Data coding/decoding method and apparatus and coded data recording medium

The invention relates to a method and a data recording apparatus for recording data on a recording medium, a method and a data reproduction apparatus for reproducing data from a recording medium to a data recording and reproduction apparatus. , and also refers to a recording medium in which data is recorded. The data is image data. The data can include 5 audio data.

Digital or similar image signals to be recorded on a disc in a digital video disc system (referred to hereinafter as DVD), are compressed and encoded using the MPEG method (Group of Experts in Coding of Motion Images).

Figure 14A is a schematic representation of the interframe prediction structure used in the MPEG system. In this example, a GOP (Group of Images) is composed, for example, of fifteen frames, which include a frame of an image I (image encoded by intraframe), four frames of images P (images encoded by forward interframe prediction) , and the remaining ten frames consisting of B images (images encoded by bidirectional prediction, forward and backward).

As used herein, image I is an image encoded by intraimage, in which either a frame or a field is compressed into data using intraframe or intrafield coding in accordance with the MPEG system; image P is an image encoded by forward interimage prediction, in which either a frame or a field is encoded with compression using interframe or interfield coding in reference to the temporarily preceding frame or field (image I or image P) and encoded; and image B is an image encoded by bidirectional prediction, in which either a frame or a field is encoded with compression using interframe or interfield coding in reference to the temporarily preceding and successive frames or fields.

More specifically, as indicated by arrows in the diagram, an image I, I0, is encoded alone by intraframe processing and without references to any other frame; an image P, P0, is encoded by interframe prediction in reference to image I, I0; and an image P, P1, is encoded by frame prediction 25 in reference to the image P, P0. In addition, images B B0 and B1 are encoded by interframe prediction in reference to both image I, I0, and image P, P0; and images B B2 and B3 are encoded by interframe prediction in reference to both image P, P0, and image P, P1. Similarly, subsequent images are encoded by said prediction according to the manner indicated by the arrows.

When decoding the mentioned images encoded by prediction, the image I is decoded alone, since it is not encoded in reference to any other frame. However, a preceding image I or a previous image P is required to decode a particular image P, since an image P is encoded by prediction in reference to the image I or the temporarily preceding image P. Similarly, preceding and successive P images I or P images are required to decode a given B image since a B image is encoded in reference to the I images or the temporarily preceding and successive P 35 images.

For this reason, to provide proper decoding, the positions of the images on a recording medium are changed from those depicted in Figure 14A to the positions illustrated in Figure 14B, so that the images necessary for decoding are decoded in advance.

As illustrated in the diagram, said positional changes are made so that image 40 I, I0, precedes images B B-1 and B-2 since image I, I0, is necessary to decode images B B-1 and B-2, and the position of the image P, P0, is also changed to precede images B B0 and B1 since decoding of images B B0 and B1 requires image I, I0, and the image P, P0. Similarly, other images are repositioned so that image P, P1, precedes images B B2 and B3 since images P P0 and P1 are necessary to decode images B B2 and B3, and also the image P, P2, 45 precedes images B B4 and B5 since decoding of images B B4 and B5 requires images P P1 and P2. In the same way, positional changes are made so that image P, P3, precedes images B B6 and B7.

The video data composed of the image I, images P and images B, arranged in the order of Figure 14B, and other data including audio data and subtitle data (captions), are structured in 50 packages (multiplexed) and are recorded on a recording medium such as a disc or transmitted on a transmission channel. The amount of code in each frame in the image data is not fixed between images and depends on the complexity or flatness of the individual image. Typically, an image I is represented by

more data than a P image, which is represented with more data than a B image.

Figures 15A to 15C represent an example of how data is structured in packages. In these diagrams, Figure 15A represents a continuous flow of the MPEG2 system that is multiplexed after packet structuring; Figure 15B represents the content of a video packet in the multiplexed stream; and Figure 15C depicts a continuous stream of MPEG2 video of a video layer. 5

In each of the image data V, V + 1, V + 2, ..., and so on, which constitute the video layer of Figure 15C, in the front position header information of the images and extension information encoding the images. In the example shown, a continuous stream of video ranging from the position identified as D1 to the position identified as D3 of the video layer forms a video packet with a packet header fixed in its forward position, and a continuous stream of video 10 that goes from position D3 to the position identified as D5 of the video layer forms another video packet with a packet header fixed in its forward position.

The video packages thus structured are multiplexed with audio packages and subtitle packages, to thereby form the continuous stream of the MPEG2 system depicted in Figure 15A.

Figure 16 represents the content of an image header, and Figure 17 represents the content of an image coding extension.

In the image header, there are information elements such as a unique image_start_code, a temporary_reference (TR) (which is a serial number in temporary succession provided by each image), and an image_coding_type (image I, P or B).

In the image coding extension, there are information elements such as an exclusive extension_start_code, an exclusive extension_sign_code_identifier, an image_structure, first_most_field, progressive_frame, and so on.

As for the image data, two data structures can coexist: a frame structure, in which an image is composed of a frame, and a field structure, in which an image is composed of two fields. It can be identified if the image data has a frame structure corresponding to one frame per image or a field structure corresponding to two fields per image, from the following three information elements, namely (1) presence of GOP header, (2) time_referral (TR) in the image header, and (3) image_ structure in the image coding extension.

Figure 18 is a block diagram illustrating an example of a data decoder apparatus 30 adapted to perform a special reproduction of data, such as slow image reproduction, rapid reproduction, backward reproduction, and the like. An optical disk 1 can be rotated by a rotation motor (not shown) with a predetermined rotation speed, and a laser beam is projected from a sensor 2 to a track of the optical disk 1, so that from it they read the compressed digital data in MPEG, recorded on the track. The digital data is processed by a demodulator circuit 35 that demodulates an eight to fourteen modulation (EFM) and is supplied to a sector detection circuit 4. The output of the sensor 2 is also supplied to a phase 9 loop loop circuit (PLL), where a clock signal is reproduced and it is supplied to the demodulator circuit 3 and the sector detection circuit 4.

The digital data recorded on disk 1 includes multiplexed continuous streams recorded in 40 units of a fixed length sector, with a sector synchronism and a sector header set at the beginning of each sector. The sector detection circuit 4 detects each of the sectors from the sector synchronism and the sector address from the sector header; and this information is supplied to a control circuit 6.

The demodulated digital data is supplied through the sector detection circuit 4 to an ECC circuit (error correction) 33 that executes the error detection and correction. The ECC circuit 33 supplies data with corrected errors to a circular buffer 5 for writing to it under the control of the control circuit 6.

The output of the ECC circuit 33 is also supplied to a continuous flow detector 50, which determines the type of image from the image header of the continuous flow of data read from disk 1 50 in a special playback mode and, at Then, it supplies the image type information to the control circuit 6. In response to this information, the control circuit 6 executes its control operation of

such that, in the special playback mode, the data of the image I and the data of the two successive images P are written to the circular buffer 5.

A focus control circuit (not shown) and a tracking servo circuit 8 control the focus and tracking of the sensor 2, respectively, under the control of a system controller (not shown) in response to a focus error signal and a tracking error signal obtained at 5 from the information read by the sensor 2.

According to the sector address of each sector detected by the sector detection circuit 4, the control circuit 6 designates, by means of a write pointer WP, a write address to write the corresponding sector in the circular buffer 5. Furthermore, according to a code request signal obtained from a video code buffer 10 (Figure 18B), the control circuit 6 also designates, by means of a read pointer RP, an address for reading the written data in the circular buffer 5. The control circuit 6 is adapted to read the data from the position of the reading pointer RP and supplies the read data to a demultiplexer 32.

Since the encoded data recorded on the disk 1 comprises multiplexed video, audio and subtitle data, the demultiplexer 32 separates the read data in the direction thereof, from the circular buffer 15, in the video data, the audio data and the subtitle data and then supplies the respective data to a video decoder 20 (Figure 18B), an audio decoder (not shown), and a subtitle decoder (not shown). The video decoder 20 stores the video data in the video code buffer 10.

After this, the data stored in the video code buffer 10 is supplied to an image header detector 34, which detects the image header. The information of the detected image header is also used to identify the type of image (image I, P or B) of the video data and the temporal reference (TR) that means the order of the frames in the GOP. An image data selection circuit 35 selects only image I and image P as identified by the image type information supplied from the image detector 34 in the special playback mode, and supplies the image data selected to a reverse VLC (variable length coding) circuit 11. In a normal playback mode, the image data selection circuit 35 is controlled to deliver all image data to the reverse VLC circuit 11 without any preselection.

The data supplied to the reverse VLC circuit 11 is processed using a reverse VLC; and then supplied to a quantifier 12. The code request signals are returned to the video code buffer 10 from the reverse VLC circuit in order to allow the transfer of new data from the buffer 10 video codes.

In addition, the reverse VLC circuit 11 outputs a size of the quantization step towards the quantifier 12 and outputs information of motion vectors to a motion compensator 15. The size of the quantization step and the motion vector information are included with the video data. The quantifier 12 decrypts the input data according to the size of the designated quantization step and outputs the quantified data to a reverse circuit 13 of DCT (discrete cosine transform). The reverse DCT circuit 13 processes the unquantified data using a reverse DCT to retrieve video information, and supplies the recovered video information to an adder 14. 40

Adder 14 adds the output of the reverse DCT circuit 13 and the output of the motion compensator 15 according to the type of image (I, P or B) and supplies the result, that is, compensated video data in motion, to a bank 16 of memories of paintings.

After this, the data read from the frame memory bank 16 is redisposed in the original order of the frames (as shown in Figure 14A, by the switch 16E). The redisposed data 45 is supplied to a digital-to-analog (D / A) converter 17 that converts the data into an analog video signal to be displayed on a display device 18.

Referring again to Figure 18A, the output of the ECC circuit 33 is supplied to a continuous flow detector 50, which detects the type of image from the continuous flow data read from disk 1 and supplies the type information. Image to control circuit 6. In response to this information, the control circuit 6, in the special playback mode, writes in the circular buffer 5, the data of the image I and the two successive images P.

Therefore, at the beginning of each GOP, they are written at high speed in the buffer

5 three frames corresponding to the image I and the two images P circulate, and these data can be acquired and decoded by the decoder 20 with any desired timing, thus allowing efficient decoding of the data in the special playback mode.

For example, reverse reproduction is supposedly initiated with the image P, P3, of the original frame order depicted in Figure 14A. It is necessary to display the decoded images in the following order:

P3 → B7 → B6 → P2 → B5 → B4 → P1 → B3 → B2 → P0 → B1 → B0 → I0 → ...

However, since each image P is encoded by interimage prediction as described, it is necessary to decode the images I0, P0, P1 and P2 before decoding the image P, P3. Similarly, it is necessary to decode images P P2 and P3 before decoding image B, B7. Therefore, if reverse reproduction is to be performed by decoding each image simply once, as in normal reproduction, it is necessary to use a bank 16 of large capacity frame memories, which can store as many frames as the images that constitute a GOP.

The storage capacity of the frame memory bank 16 must be increased beyond what is required in a normal reproduction mode to satisfy said requirement. In addition, the decoded data 15 must be stored sequentially in the frame memory bank to deliver the images in the proper reverse reproduction order.

Although other reverse reproduction techniques can be adopted to simply work with images I and P, thus omitting images B, there is still a need to store more frames than those required for normal reproduction. twenty

For this reason, the data decoder apparatus of Figure 18 functions to perform reverse playback using the same frame memory bank as used in normal playback, that is, using three memory elements in the example of Figure 18. to store one image I and two temporarily successive P images. The continuous flow detector 50 provided for this purpose writes the image I and the two successive images P in the circular buffer 5. However, this causes the construction and operation of the continuous flow detector 50 to detect the image I and the two successive P images are more complicated.

Figure 15A represents the continuous flow of the MPEG2 system structured in packets (multiplexed). When a packet of the MPEG2 video stream is defined in a D3 position, as shown in Figure 15C, during the packet structuring process, the image header and the coding extension of the data image Image (V + 2) extend over two packages, as shown in Figure 15B.

If the image header and the image coding extension are spread over two video packages, it is necessary to detect two video packages to obtain the required elements of image information. In addition, and as shown in Figure 15A, there may be another package 35 (for example, an audio package) between the two video packages to complicate the detection process, thereby complicating the construction and operation of the detector 50 of continuous flows.

According to the MPEG2 technique, the video data of a frame structure, in which an image is composed of a frame, and the video data of a field structure, in which an image is composed of two fields, they can intermingle. As each image header 40 is set to each field, the image headers and image coding extensions corresponding to two consecutive images must be read to determine the data structure of the video data.

Therefore, a determination is made to find out if the image data has been constituted in the frame structure or in the field structure based on the three information elements mentioned above, that is, (1) header presence of GOP; (2) temporal reference (TR) in the image header; and (3) image structure information in the image encoding extension.

A detailed explanation is given below in relation to a method to differentiate between a frame structure and a field structure.

Figures 19A and 19B represent video data in the field structure and frame structure formats, respectively. In the field structure format, a video data frame 50 is composed of two image data fields, each of which is set to a header of

image and an extension of image coding. In the frame structure format, a video data frame is composed of an image data frame to which an image header and an image coding extension are attached.

In the format of the field structure, the numerical values ​​of the TR information in the respective image headers of the image data pair are set equal to each other. The image structure information in the image coding extension is "01" and "10" for the Upper Field and the Lower Field, respectively, as shown in Figure 20. In addition, the image structure information in the image coding extension of the frame structure is "11" as shown in Figure 20.

The format (field or frame structure) of the image data can be ascertained by first reading the GOP header at the starting position of the GOP and then reading the structure information_image of the image coding extension, in the beginning of that image data.

Although the image data in the frame structure can be loaded into the circular buffer 5 (Figure 18A) when detecting a single frame, it is difficult to similarly load the video data in the field structure, in which a pair Image data is a video data frame, because the paired image data must be detected before they can be properly loaded. Consequently, the TR information is read in each image header to find two units of image data with numerically equal TR values. When such a pair is found, they are identified as paired image data and then a loading of them is performed. twenty

Image headers with paired field structure are arranged in one of two different orders: upper / lower and lower / upper. These provisions will be described below with reference to Figure 21. A GOP header (GOP H), a frame structure image I, a field structure image B, another field structure image B, a GOP header separated from them, and a structure image I are sequentially recorded of fields, a GOP header, another 25 field structure image I, ... and so on.

For example, when a total of three frames (one image I and two successive P images) are loaded into circular buffer 5 (figure 18A), the image I of the frame structure that is next to the upper GOP header ( first) the image_coding_type in the image header is detected and identified from the GOP header, and the image structure information ("11" in the case of the frame structure) in the image coding extension at the beginning of the image data.

When an access is made for reproduction in the position identified as random access 1 in the continuous bit stream, the image header and the image coding extension of the first image B with field structure are read. At this time, the TR expressed as "0" is also read. Then, as the TR expressed as 0, the image header and the image coding extension of the second image B with field structure are read. Since the TR value of the two B images with field structure is the same, these images are detected as paired data.

When an access is made at the position identified as random access 2 in the continuous bit stream, the image header and the image coding extension of the first image are read together with the TR expressed as "0". Next, the image header and the image coding extension of the following image are read together with the TR expressed as "1". As the respective numerical values ​​of the TR do not coincide with each other, the data of the two field-structured images are not detected as paired data.

If an access is made in the position identified as random access 3 in the continuous bit stream, the respective numerical values ​​of the TR in the two image headers coincide with each other (TR = 1), 45 as in the previous case of the access random 1, so the images are detected as a pair. If an image structure is detected in the image coding extension of "01" or "10", it is considered as a field structure, and paired data is detected.

If an access is made at the position identified as random access 4 in the continuous bit stream, the image header and the image coding extension of the first image are read together with the TR 50 expressed as "0". This image data is considered as image I with field structure according to the image coding type information in the image header and the image_ structure information in the image coding extension.

After subsequent detection of the GOP header, the image header and the image coding extension corresponding to the following image are read, together with the TR expressed as "0". In this case, the respective numerical values ​​of the TR of the two consecutive images coincide with each other, but these two images are not considered as paired since there is a GOP header between the two images. It should be noted that the TR is reset to "0" if there is a GOP header and there is no GOP header interposed between a paired image.

As described, the continuous flow detector 50 executes a process of detecting various information elements in relation to the images, in accordance with the GOP header, the image headers and a plurality of flags of the coding extensions image, to load the image data into circular buffer 5. However, this processing routine is extremely complex, which makes it difficult to construct the continuous flow detector 50.

The manner in which the continuous flow detector 50 detects that the charging process has been completed is explained in conjunction with the flow chart depicted in Figure 22. It is assumed, in this flow chart, that random access is made to an input sector written immediately prior to an image I, so that an appropriate image can be obtained instantly, in response to random access.

In step S10, the continuous flow detector searches for the image_start_code in the image header to detect the image header corresponding to the image I and in step S12 a query is made to determine if the image_start_code has been detected. If the query of step S12 is answered affirmatively, that is, if the image_start_code is detected, operation continues towards step S14. However, if the response to the query of step S12 is negative, that is, if no image_start_code is detected, the process of step S12 is repeated until the image_start_code is detected.

In step S14, the temporal reference of the detected image header is read and its numerical value is stored in a register as TR0. 25

In step S16, another search (SRCH) of the image_start_code is performed in the image header, in order to detect the following image and in step S18 a query is made to determine if the image_start_code has been detected. If the answer to the query of step S18 is affirmative, that is, if the image_start_code is detected, the operation continues towards step S20. However, if the response to the query of step S18 is negative, the process of step S18 is repeated until the image_start_code is detected.

In step S20, a query is made to determine if a GOP header has been detected in the image start code, in order to determine in this way whether the detected image data is part of a pair. If the response to the query of step S20 is negative, that is, if no GOP header has been detected, operation continues towards step S22. However, if the answer to this query is affirmative, that is, if the GOP header has been detected, the operation proceeds towards step S26 since the existence of a GOP header between the image data eliminates the possibility of that these imaging units are paired.

When, from the detected image header, the time reference is read, its numerical value is stored in a register as TR1, as represented by step S22, and operation proceeds 40 towards the query of step S24 to determine if numerical values ​​of the TR stored, respectively, in records TR0 and TR1 are equal. If the query is answered affirmatively in step S24, that is, if there is a match between the two numerical values, the operation returns to step S16 and the processes described above are repeated in combination with steps S16 to S24. It will be appreciated that the coincidence of two numerical values ​​means that a pair of 45 image data has been detected.

However, if the query of step S24 is answered negatively, that is, if the numerical values ​​of the TR are not equal, the operation continues towards step S26. In this case, the image header corresponding to the following image is detected; and the type of image coding read from the image header is stored in the register. The operation proceeds, then, until the consultation of the step S28 to determine if the type of stored image coding represents an image B. If the answer to the query of step S28 is answered affirmatively, that is, if the detected image is a image B, operation returns to step S16 since image B is not being searched; and the processes described above are repeated in combination with steps S16 to S28 to detect the next image.

However, if the query in step S28 is answered negatively, that is, if the detected image is not an image B, the time reference of the detected image header is read and its numerical value is stored in the register as TR2 , as represented by step S30. It will be appreciated that this detected image is the first image P that appears after image I.

In step S32 another search (SRCH) of the image_start_code in the image header is performed, to detect the following image and, in step S34, a query is made to determine if the image_start_code has been detected. If the answer to step S34 is answered in the affirmative, the operation continues to step S36. However, if the query in step S34 is answered negatively, that is, if no image_start_code has been detected, the process is repeated in step S34 until the image_start_code is detected. 10

In step S36, a query is made to determine if a GOP header was detected during the search of the image_start_code, to thereby determine whether the detected image data is part of a pair. If the query of step S36 is answered negatively, that is, if no GOP header has been detected, operation continues towards step S38. However, if the query of step S36 is answered in the affirmative, that is, if a GOP header has been detected, operation proceeds towards step S42 because the existence of a GOP header between image units eliminates the possibility that these imaging units are paired.

When the time reference of the image header detected in step S38 is read, its numerical value is stored in the register as TR3 and the operation proceeds until the consultation of step S40, to determine if a match between the numerical values ​​is obtained of the stored TR, 20 respectively, in the register as TR2 and TR3. If the answer to step S40 is answered in the affirmative, that is to say if the two numerical values ​​are equal, the operation returns to step S32 and the processes described above are repeated in combination with steps S32 to S40. It will be appreciated that the coincidence of two numerical values ​​means that a pair of image data has been detected.

However, if the query of step S40 is answered negatively, that is, if the numerical values ​​of the TR do not match, the operation continues towards step S42 to read the type of image. Operation proceeds, then, to the consultation of step S44 to determine if the type of stored image coding represents an image B. If the query of step S44 is answered affirmatively, that is, if the detected image is an image B, the operation returns to step S32 since the image B is not being searched; and the processes described above are repeated in combination with steps 30 S32 to S44 to detect the following image.

However, if the query in step S44 is answered negatively, that is, if the detected image is not an image B, the time reference detected in the image header is read and its numerical value is stored in the register as TR4 , as represented by step S46. It will be appreciated that this detected image is the second image P that appears after image I. 35

Continuing to step S48, another search (SRCH) of the image_start_code is performed in the image header to detect the next image and a query is made in S50 to determine if the image_start_code has been detected. If the answer to step S50 is answered in the affirmative, that is, if the image_start_code has been detected, operation continues until step S52. However, if the query of step S50 is answered negatively, the process of step S50 is repeated until the image_start_code is detected.

In step S52, a query is made to determine if a GOP header has been detected during a search of the image_start_code, in order to determine in this way whether the detected image data is part of a pair. If the query in step S52 has a negative answer, that is, if no GOP header has been detected, the operation continues to step S54. However, if the answer to step S52 is answered in the affirmative, that is, if a GOP header has been detected, the loading of the image data in the circular buffer is completed and the process is terminated.

When the time reference is read from the detected image header and its numerical value is stored in the register as TR5, as represented by step S54, the operation proceeds until the consultation of step S56, to determine if a match is obtained between the numerical values ​​50 of the TR stored, respectively, as TR4 and TR5. If the answer to step S56 is answered in the affirmative, that is, if there is a coincidence between the two numerical values, the operation returns to step S48 and the processes described above are repeated in combination with steps S48 to S56. However, if the query in step S56 is answered negatively, that is, if the two numerical values ​​of the TR do not

they are the same, the loading of the image data is completed and the process is finished.

Thus, the continuous flow detector 50 can load a continuous stream of bits by executing the above processing routine to load an image I and two successive images P. However, as its extensive description suggests, the execution of this complex processing routine is quite expensive.

WO-A-94/07332 describes an apparatus and a method of processing digital video signals, in which video data and audio data are recorded in units of a batch on a disk. A batch is constituted from a plurality of packages. When a video data package is found, if the image at the top of the video package is an image I, then, immediately before the video package, an input package is arranged. In the input package, the positions of three previous and subsequent packages are recorded. The position of the image I is detected from the data 10 of the input package. This makes it possible to quickly locate the access point to achieve a quick search.

An embodiment of the present invention is to provide a method and an apparatus for encoding data in order to perform a special reproduction, for example, a fast forward and fast reverse reproduction, which overcomes the drawbacks of the aforementioned technique , that is, 15 the intricate operations mentioned above.

Another embodiment of the present invention is intended to provide a method and an apparatus for decoding data in order to perform a special reproduction, which overcomes the drawbacks of the technique described above.

Another embodiment of the present invention is to provide a recording medium 20 to be used in conjunction with a processor-controlled apparatus, in order to perform special reproduction without the need for the processor-controlled apparatus to perform the intricate operations mentioned above. .

In accordance with one aspect of the present invention, an apparatus and method for recording encoded image data on a recording medium are provided, in which: the image data is encoded using an intraframe encoding and / or a predictive encoding , to provide an image I and two successive P images; positional information is generated representing the positions of the two images P in relation to the image I, and the image I, the two images P and the positional information are recorded on the recording medium.

In an embodiment of one aspect of the present invention, the positional information 30 represents the length of data in bytes from the image I to the end of the first image P and / or the end of the second image P.

In accordance with another aspect of the present invention, an apparatus and a method for reproducing encoded image data from a recording medium are provided, in which: positional information representing the positions of two P images relative to one image is detected I, and a continuous flow of data is generated that includes image I, the two images P and positional information. The continuous flow of data can be decoded and visualized.

In accordance with yet another aspect of the present invention, a recording medium is provided for use in conjunction with a processor-controlled apparatus, in which at least one image I, at least two images P, and positional information representing the positions of the at least two images P in relation to the at least one image I, are recorded on the support. The processor-controlled device uses the information to perform a special reproduction in a relatively simple and straightforward way.

The following detailed description, provided by way of non-limiting example of the present invention, will be made more clear by reference to the accompanying drawings, in which:

Figure 1 is a block diagram showing an embodiment of the data coding apparatus of the present invention;

Figure 2 represents an example of a packet structured continuous stream, encoded by the data coding apparatus of Figure 1;

Figure 3 represents an arrangement of the entry point information; fifty

Figure 4 represents a syntax of a map of continuous program streams (PSM);

Figure 5 represents a syntax of elementary continuous flow descriptors;

Figure 6 represents a syntax of a descriptor_ip_ipp;

Figure 7 represents a syntax of global descriptors; 5

Figures 8A and 8B are a block diagram representing an embodiment of the data decoding apparatus of the present invention;

Figures 9A to 9C are diagrams representing an example of the order of video data to which reference will be made in describing how video data is read in a fast reverse playback mode of the data decoding apparatus of Figures 8A and 8B ; 10

Figure 10 is a read / write timing diagram that will be referred to when describing the fast reverse playback mode of the data decoding apparatus of Figures 8A and 8B;

Figures 11A to 11C are diagrams representing an example of the order of video data to which reference will be made in describing how video data is read in a fast forward playback mode of the data decoding apparatus of Figures 8A and 8B ; fifteen

Figures 12A to 12C are diagrams representing another example of the order of video data to which reference will be made in describing how video data is read in a fast reverse playback mode of the data decoding apparatus of Figures 8A and 8B ;

Figure 13 is a read / write timing diagram that will be referred to when describing how video data is read using two frame memories in the fast reverse playback mode 20 of the data decoding apparatus of Figures 8A and 8B;

Figures 14A and 14B are schematic representations of original interframe prediction image structures and recorded pictures of images in the MPEG system;

Figures 15A to 15C are schematic representations of a continuous stream of MPEG video;

Figure 16 represents the structure of an image header in the MPEG system; 25

Figure 17 represents the structure of an image coding extension in the MPEG system;

Figures 18A and 18B are a block diagram representing a data decoding apparatus;

Figures 19A and 19B are schematic representations of the structure of video data in a frame format and in a field format; 30

Figure 20 is a table representing the content of an image structure;

Figure 21 is a diagram of a continuous stream of video depicting how two video formats differ (field and frame); and

Figure 22 is a flow chart of a routine executed in a continuous flow detector to load three frames of the image data (one image I and two images P) when the 35-field and frame video formats are intermingled.

The illustrative embodiments of the present invention are described in detail below, with reference to the accompanying drawings.

Figure 1 is a block diagram showing an embodiment of the data coding apparatus of the present invention. An audio encoder 102 encodes with compression an input audio signal 40 supplied to its audio input terminal and a video encoder 101 compressively encodes an input video signal supplied to its video input terminal. The encoded audio and video signals are supplied to multiplexer 113. It is assumed that a continuous stream that has been output from the audio encoder 102 is a continuous stream of MPEG2 audio (audio layer) and a stream

The continuous output from the video encoder 101 is a continuous stream of MPEG2 video (video layer), the latter being represented in Figure 15C.

The multiplexer 113 bundles the continuous stream of MPEG2 video and the continuous stream of incoming MPEG2 audio by time division multiplexing, to form the continuous stream of the system, shown in Figure 15A. 5

Although not shown, a continuous stream of subtitles can also be introduced into multiplexer 113 and it can be multiplexed with continuous video stream and continuous audio stream. In such a case, the continuous flow of the MPEG2 system that has been output from multiplexer 113 is as shown in Figure 15A.

An input terminal of an input point data memory circuit 133A is connected 10 to the video encoder 101; and an input point detector 131 allows the input memory data circuit 133A to store an entry point (data relative to an image generation point I) received from the video encoder 101.

A table of contents data generator (TOC) 156 generates TOC data based on the content of the data memory circuit 133A of input points. TOC data includes a disc name 15 on which video and audio data is recorded, the name of each chapter recorded on the disc, the starting address of each chapter on the disc, the playing time of the disc , the playing time of each chapter, the starting address of each input sector and the like.

Continuous packet stream exiting multiplexer 113 is temporarily stored in a DSM (Digital Storage Media) 110 and then supplied to a 20 TOC suffrage circuit 150. The TOC suffixing circuit 150 adds the TOC data to the packet-structured continuous flow and supplies it to an image header detector / data overwriting device of the program continuous flow map (PSM) map, 155.

The image header detector / generator-data overwriting device of the PSM, 155, detects an image header and generates PSM data that includes information representing the length of data in bytes from the beginning of an input sector to the end of the first image P that appears and / or from the beginning of the input sector to the end of the second image P that appears. Preferably, the PSM data includes information representing the length of data in bytes from the beginning of an image I to the end of the first image P that appears and / or from the beginning of an image I to the end of the second image. P to appear. The generated PSM data is written in an area of ​​the input sector previously reserved for this in the continuous flow structured in packets by means of multiplexer 113. A detailed description of the PSM data is provided herein below.

The output of the image header detector / generator-data overwriting device of the PSM, 155, is supplied to a circuit 151 for sector header suffixing, where the continuous flow 35 structured in packets is divided into sectors, adding a header from sector to each of the sectors.

The output of the sector header suffixing circuit 151, that is, the video and audio data together with all other data added thereto, as described above, are encoded for error correction by means of a ECC 152 encoder. 40

Next, a modulator 153 modulates the encoded data from the ECC encoder 152 using an eight to fourteen (EFM) modulation and the modulated data is supplied to a cutting machine 154. The cutting machine 154 forms depressions on a master disk 160, in accordance with the data supplied from the modulator 153, whereby the data of the continuous stream structured in packets is written on a master DVD disk 160. A DVD disc copy is produced, for example, by pressure molding of this master disc 160.

In this way, the data coding apparatus of Figure 1 encodes and structures in packets, using a time division multiplexing, an audio signal and a video signal introduced therein, to produce a continuous structured flow in packets. In addition, the image header detector / generator-data overwriting device of the PSM, 155, generates and writes data from the PSM in the 50 continuous stream structured in packets. The continuous stream structured in packets is recorded on the DVD master disc 160.

Figure 2 represents an example of the continuous flow structured in packets, for example, a continuous flow of the MPEG2 system, which has been output from the image header detector / generator-data overwriting device of the PSM, 155. For simplify, only video data and audio data structured in packets are represented. Audio data is inserted into certain parts of the MPEG2 system's continuous stream to ensure that sound is not interrupted during playback, and video data of images I, P and B is inserted between audio data 5.

An entry point means the upper (or starting) position of an image I, and a sector that includes that entry point is called the input sector. In Figure 2, the positions of said entry points are indicated as entry point n, entry point n + 1, ... and so on. The position at which input point information is written is predetermined to be immediately prior to an image I, so that a complete image can be instantly displayed when a sensor has read the data from the input sector.

There may be audio data between the entry point information and an image I, but among them there are no P or B images.

Figure 3 represents an arrangement of the entry point information. The 15 entry point information includes a packet header with an optional system header, a PSD (Program Continuous Flow Directory), a PSM (Program Continuous Flow Correspondence Map) and other packages.

Figure 4 represents the syntax of the PSM. The PSM includes a 24-bit package_start_code prefix that forms an exclusive code, an 8-bit Cont_corresp stream_id, a program_flowCont_program composed of 20 an arbitrary number of global descriptors, a type_Cont_flow, and an element_flow_content_flow that includes an arbitrary number of continuous stream descriptors.

Figure 5 represents the syntax of the elementary continuous stream descriptors that are composed of a dvd_vd_ descriptor and an ip_ipp descriptor if the continuous stream is video data, or composed of a dvd_audio_ descriptor and an ISO_6__language descriptor if the continuous stream is audio data, or composed of a descriptor_subtitles_dvd and a descriptor_language_ISO_639 if the continuous stream is subtitle data. In addition, other information elements are represented in Figure 5.

As depicted in Figure 6, the ip_ipp descriptor consists of an 8-bit descriptor_tag that means an ip_ipp descriptor, an 8-bit description_length that means the length of the 32-bit descriptor, bytes_a_primer_img_P which means the number of bytes from the first byte of the current input sector 30 to the last byte of an image P that appears first, and 32-bit bytes_a_second_img_P which means the number of bytes from the first byte of the current input sector to the last byte of the P image that appears second.

Bytes_a_primera_img_P and bytes_a_segunda_img_P represent the data lengths, as shown in Figure 2. It can be seen that the number of offset bytes indicated by the information 35 of bytes_a_primera_img_P and bytes_a_segunda_img_P includes not only I and P images but also intermediate B images and packets audio, as depicted in figure 2.

Figure 7 represents the syntax of the global descriptors of Figure 4. The PSD included in each input sector represents the distances from the current input sector to the preceding input sector and the next input sector, and the distances to input sectors after a span of 40 of a second, three seconds, and so on. These distances are referred to as directions of travel.

Figures 8A and 8B are a block diagram of a preferred embodiment of the data decoder apparatus of the present invention. For simplicity, the elements represented in Figures 8A and 8B corresponding to those represented in Figures 18A and 18B are indicated by the same numerical references.

An optical disk 1 is rotated at a predetermined rotation speed by means of a rotation motor (not shown), and from a sensor 2 a laser beam is projected onto a track on the optical disk 1, so that they are read from the same digital data compressed by MPEG, recorded on the track. The digital data is demodulated in EFM by means of a demodulator 3 and entered into a detector 4 of 50 sectors. The output of the sensor 2 is also supplied to a phase 9 loop loop circuit (PLL), which recovers a clock signal that is supplied to both demodulator 3 and sector detector 4.

As described above, digital data is recorded on disk 1 in fixed-length sector units, in which at the beginning of each sector a sector synchronism and a sector header are recorded. The division of sectors is determined from the detection of the sector synchronism and a sector address from the sector header, which are supplied to a control circuit 6. Preferably, the control circuit is implemented with microprocessors and performs a processor control 5 on the illustrated apparatus.

The demodulated output is supplied through the sector detector 4 to an ECC circuit 33 (error correction), which performs error detection and correction. The ECC circuit 33 supplies the data with the corrected errors to a circular buffer 5 for writing to it under the control of the control circuit 6. 10

The output of ECC circuit 33 is also supplied to a PSM detector 40. In the special playback mode, the PSM detector 40 detects the PSM information in the input sector from the continuous flow data read from the disk 1, and supplies the detected PSM information to the control circuit 6. The control circuit 6 uses this PSM information to control the writing (or loading) of images I and P in the circular buffer 5 in a special reproduction mode in accordance with the information regarding the number of offset bytes in the descriptor_ip_ipp , to ensure that the length information in the continuous flow data, from an image I after the input sector to the second image P, is recorded in the circular buffer 5.

A focus control circuit (not shown) and a tracking servo circuit 8 control the focus and tracking of the sensor 2, respectively, under the control of a system controller (not represented 20), in accordance with an error signal of focus and a tracking error signal obtained from the information read by the sensor 2.

On the basis of the sector address of each sector detected by the sector detector 4, the control circuit 6 designates, by means of a write pointer WP, a write address to write the appropriate sector in the circular buffer 5. Furthermore, on the basis of a code request signal obtained from a video code buffer 10 (Figure 8B), the control circuit 6 designates, by means of a read pointer RP, a read address from the which the data is read from the circular buffer 5. The data read from the position designated by the reading pointer RP is supplied to a demultiplexer 32.

Since the encoded data recorded on the disc 1 comprises video data, audio data and 30 data of structured subtitles in packets, multiplexed, the demultiplexer 32 separates the data supplied to it, into the video data, audio data and data of subtitles and then supplies the respective data to a video decoder 20 (Figure 8B), an audio decoder (not shown), and a subtitle decoder (not shown).

As a consequence, the video data read from the circular buffer 5 is stored in the video code buffer 10 of the video decoder. Continuous flow data, from an image I to the second successive image P, contains packages that are not the video packages, as shown in Figure 2. In the special playback mode, any unnecessary data, that is, packets other than video data, are excluded by the demultiplexer 32.

The data stored in the video code buffer 10 is supplied to an image header detector 40 34 where the image header thereof, the image type information meaning the image type I, P, is detected. or B, and the temporal reference (TR) that means the order of the tables in the GOP. The image type information detected is supplied to an image data selector 35, in which only images I and P are selected in the special playback mode; and the selected images are supplied to a reverse VLC circuit 11. In normal playback mode, the image data selector 35 is controlled to output all the image data without making any previous selection.

The data supplied to the reverse VLC circuit 11 is processed using a reverse VLC therein; and then supplied to a quantifier 12. The code request signals are returned to the video code buffer 10 from the reverse VLC circuit to allow the transfer of new data from the code buffer 10 Of video.

In addition, the reverse VLC circuit 11 outputs a quantization step size towards the quantifier 12 and outputs motion vector information to a compensator 15 of

movement. The quantifier 12 decrypts the input data according to the size of the specified quantization step and outputs the quantified data to a reverse DCT circuit 13. The reverse DCT circuit 13 processes the quantized data using a reverse DCT and supplies the processed data to an adder 14.

Adder 14 adds the output of the reverse DCT circuit 13 and the output of the motion compensator 15 of 5 according to the type of image (I, P or B) and supplies the resulting video data, compensated in motion, to a bank 16 of memories of paintings.

After this, the data read from the frame memory bank 16 is redisposed in the original frame order shown in Figure 14A by switch 16e. The redisposed data is supplied to a digital-to-analog (D / A) converter 17, to be converted into an analog video signal 10, which is then displayed on a display device 18.

In response to a code request signal from the video code buffer 10, the control circuit 6 supplies the data stored in the circular buffer 5 to the video code buffer 10. When the amount of data transferred from the video code buffer 10 to the reverse VLC circuit 11 is reduced, for example, as a consequence of the continuous processing of simple image data presenting a small amount of data, it is also reduced the amount of data transferred from the circular buffer 5 to the video code buffer 10. Consequently, the amount of data stored in the circular buffer 5 can be increased, and the write pointer WP can potentially exceed the read pointer RP to cause an overflow of the circular buffer 5. 20

To avoid this problem, the control circuit 6 calculates the current amount of data stored in the circular buffer 5 based on the address positions of the write pointer WP and the read pointer RP. When the calculated amount of data exceeds a predetermined magnitude, a track skip decision circuit 7 determines that, potentially, the circular buffer 5 can overflow, and sends a track skip command to the tracking servo circuit 8. In response to the track jump command, the tracking servo circuit 8 causes the sensor 2 to skip tracks, depending on the storage capacity of the circular buffer 5, in order to avoid any overflow or underflow of the buffer memory circular 5. This advantageously allows continuous video playback with uniform image quality, regardless of the complexity (or flatness) of the images recorded on disc 1. 30

The data transfer rate from the circular buffer 5 to the video code buffer 10 is previously set to be equal to or less than the data transfer rate from the ECC circuit 33 to the circular buffer 5, in order to allow the video code buffer 10 to transmit to the circular buffer a request for data transfer regardless of a track skip. 35

In a normal playback mode, it is assumed, for example, that the image data I, P and B, I0, B-2, B-1, P0, B0, B1, ... are recorded on disc 1 in the order represented in figure 14B. In this example, a GOP is composed of fifteen frames of images, including one frame of image I, four frames of images P and ten frames of images B. Normal reproduction of the images is done by reading and decoding sequentially the encoded data in the order of recording represented in Figure 14B and 40 by displaying the decoded data in the order illustrated in Figure 14A.

More specifically, at the time of decoding the image I, I0, the decoded output obtained from the reverse DCT circuit 13 is supplied directly to the frame memory bank 16. However, at the time of decoding the image B, B-2, the previously decoded image P (not shown) and the image I, I0, both used as references for the predictive coding of the image B, B-2, They are supplied from the frame memory bank 16 to the motion compensator 15, and a predicted motion image is generated according to the motion vector information supplied from the reverse VLC circuit 11. The motion image with prediction, generated, is then supplied to adder 14 which adds the motion image with prediction to the output of the reverse DCT circuit 13, whereby image B, B-2, is decoded and 16 of 50 frame memories are stored in bank 16.

Image B, B-1, is decoded in the same manner as image B, B-2, and is overwritten on top of image B, B-2, which is stored in one of memories 16a to 16c of frames of the frame memory bank. When decoding the image P, P0, the image I, I0, is supplied from the bank 16 of

frame memories to the motion compensator 15 together with the motion vector information supplied from the reverse VLC circuit. The motion compensator supplies a moving image with prediction to adder 14, which adds this moving image with prediction to the image P, P0, supplied from the reverse DCT circuit, to thereby decode the image P, P0. The image P, P0, thus decoded is written on the oldest data (which can be either image I or image P) stored in the bank 16 of frame memories.

Thus, the images are decoded sequentially, as mentioned above, but they are read from the frame memory bank 16 in their original order and displayed on the display device 18 in the order represented in Figure 14A.

In a fast playback mode, it is necessary that the data recorded on disc 1 be decoded and displayed in reverse order. For example, if an image B, B07, of disk 1 is decoded, in which video data is recorded in the order represented in Figures 9A and 9B (to simplify, only video data is represented), it is necessary to decode images P P08 and P05 before decoding image B, B07, since these images P have to be used as a reference to decode the desired image B, B07. However, it is appreciated that the image P, P05, decoded is necessary to decode the image P, P08, and that the image I, I02, decoded is necessary to decode the image P, P05. Consequently, decoding should start from image I located at the beginning of the GOP. At the end of decoding a GOP, it is necessary that the operation goes back to the previous GOP to continue with the decoding process.

However, if a reverse reproduction is performed with said decoding technique, an excessive time delay will be introduced in the display of the images, which results in the display of unnatural images. The illustrative embodiments of the present invention solve this problem by decoding only images I and P, as in the normal playback mode, but perform a reverse playback using only the same three memories 16a, 16b and 16c of frames necessary for normal reproduction Thus, the present embodiments 25 advantageously decode a total of three images, an image I appearing after an input sector and two successive images P, instead of decoding a complete sequence of images I and P, making it possible thus a fast reproduction in backward movement with a minimum complexity of the circuits.

The present embodiment can perform rapid reverse reproduction by means of a PSM detector 40 with a simplified circuit configuration, without using a continuous flow detector 30 having a complicated configuration as described above. As the PSM detector 40 detects information representing the number of offset bytes provided in the PS_ip_ descriptor in a PSM, it is possible to write in the circular buffer 5 only the required interval of the continuous flow of data from an image I that appears immediately after from an input sector to the two successive P images that appear after image I. 35

It will be described with reference to Figures 9A to 9C the operation performed by the data decoder apparatus of Figures 8A and 8B in a fast reverse playback (FR) mode.

Figures 9A and 9B represent the order of the video data recorded on disc 1. The video data represented in these diagrams correspond to four GOPs and, in the FR playback mode, the control circuit 6 executes its action of control such that the sensor 2 reads the video data of the disk 1 in the order indicated by the arrows below the video data. More specifically, the sensor 2 reads successively the video data of the image I, I32, the image B, B30, the image B, B31, the image P, P35, the image B, B33, the image B, B34, and image P, P38, in this order, then jumps to an immediately preceding GOP, and reads the video data of image I, I22, to the second image P, P28. Next, the sensor 2 jumps to another previous GOP and reads the video data of the image I, I12, until the second image P, P18. Next, the sensor 2 jumps to an additional preceding GOP and then reads the video data of the image I, I02, to the second image P, P08. Then, similar to that described above, the sensor 2 reads the video data of an image I positioned at the beginning of a successive preceding GOP until the second image P positioned after that image I.

Said read operation is possible since the aforementioned entry point information 50 is written in the input sector formed at the top of each GOP and the PSM detector 40 detects the descriptor_ip_ipp in the input sector and then , supplies the detected_ip_ipp descriptor to the control circuit 6. Consequently, the control circuit 6 controls the sensor 2 to read, from the top (or beginning) of the input sector, the data corresponding to the number of bytes represented by the byte_a_second_img_P information provided in the descriptor_ip_ipp, with the

which enables the sensor 2 to read the video data in the order indicated by the arrows in figures 9A to 9C.

To access the upper part of the immediately preceding GOP, the distance information (data length) represented as a direction of travel in the PSD in the input sector is used, and the same means the distance to the preceding input sector. 5

The read video data comprised from the image I positioned at the beginning of the GOP to the second image P positioned after the image I, is separated from the audio data and other data by the demultiplexer 32 and then written to the buffer 10 of video codes. Images B are removed using the detection information obtained from the image header detector 34, and decoding only the image data I and P that are written in the bank 16 of frame memories 10. The video data thus written is read from the frame memory 16 in the reverse order of image display depicted in Figure 9C and displayed on the display device 18.

It is described with reference to FIG. 10 the data read / write timing for the frame memory bank 16 in the FR playback mode. Preferably, the frame memory bank 16 is provided with three frame memories 16a, 16b and 16c, as shown in Figure 8. The writing of the decoded image I, I32, at the top of the start GOP (see Figure 9B) in the frame memory 16a begins at the time instant t0, and ends at the time instant t1 after a period of time. of a painting Next, the writing of the image P, P35, decoded with reference to the image I, I32, in the frame memory 16b begins at the instant of time t1, and ends at the instant of time t2 after a period of one picture. twenty

In addition, the writing of the image P, P38, decoded in reference to the image P, P35, in frame memory 16, begins at the time of time t2, and ends at the time of time t3 after a period of one picture. The reading of the image P, P38, from the frame memory 16c can begin at an intermediate point between the two instants of time t2 and t3, provided that a field of the image P, P38 has already been written in the memory 16c of frame before this instant of start reading time. Therefore, it is possible to read from and write in the same frame memory simultaneously by delaying in a field the reading timing with respect to the writing timing.

The reading of the image P, P38, of the frame memory 16c ends at an intermediate time between the two instants of time t3 and t4, and the writing of the decoded image I, I22, of the previous GOP in the memory 16c of frame begins at the instant of time t3. This writing ends in 30 seconds of time t4 after a period of one frame.

In the frame memory 16c different image data can be written while simultaneously reading from it the previously written image data, as mentioned above, since the write timing presents a delay of a field with respect to Read timing. 35

After this, the decoded image data as depicted in Figures 9A to 9B are written in memories 16a, 16b and 16c of frames in the order of I32, P35, P38, I22, P25, P28, I12, P15 , P18, I02, P05, ... and so on. Meanwhile, the image data is read from memories 16a, 16b and 16c of frames in the order ranging from the oldest (largest) image number to the most recent (minor) image number, such as P38, P35, I32, P28, P25, I22, P18, P15, I12, ... and so on. 40

Consequently, fast reverse playback is performed and the images are displayed in the order depicted in Figure 9C. For example, three images can be played back by GOP when three frame memories are used.

In the reverse playback mode, identification numbers assigned to the images are detected, and these are read from the frame memory bank 16 in the order ranging from the oldest to the most recent number (from highest to lowest). The temporary references (TR), which mean the numbers that indicate the order of visualization of the images, are reset in the respective upper parts of the GOPs, and the values ​​of said temporary references are in a range of 0 to 1,023.

Referring now to Figures 11A to 11C, an explanation of the fast forward reproduction (FF) performed in the data decoder apparatus of Figures 8A and 8B is provided. The 50 figures 11A and 11B represent the video data of four GOPs in the order in which they are recorded on disc 1, in which the arrows below the video data indicate the reading order thereof in the mode of FF playback.

In the FF playback mode, the PSM detector 40 detects a descriptor_ip_ipp in the input sector written at the top of each GOP and supplies the detected_ip_ipp descriptor to the control circuit 6, as in the aforementioned playback mode in FR previously. The control circuit 6 controls the sensor 2 to read, from the top of the input sector, the data corresponding to the number of bytes represented by the information of bytes_a_second_img_P in the descriptor_ip_ipp, so that the 5 video data is read in the order indicated by the arrows in figure 11C.

Images B, whose identities are present in the respective image headers, are deleted from the read video data, so that only images I and P are decoded. Images I and P decoded are read from the memory memory bank 16 frames in the order of decoding and are displayed on the display device 18 in the order of I02 P05 P08 I12 P15 P18 I22 P25 P28 I32 P35 P38, as shown in Figure 11C.

Although, preferably, three frame memories are incorporated into the frame memory bank 16, the number of frame memories is not limited to three, and any desired number can be selected. Fast forward (FF) playback will be performed with images I and P in a number equal to frame memories. fifteen

The reproduction in FR made with only two memories 16a and 16b of frames in the bank 16 of frame memories will be described below, with reference to Figures 12A to 12C and Figure 13.

Figures 12A and 12B represent the order in which video data is recorded on disc 1. The video data in Figures 12A and 12B correspond to four GOPs and, in the 20 FR playback mode, circuit 6 of control controls the sensor 2 to read the video data from disk 1 in the order indicated by the arrows below the video data. More specifically, the sensor 2 reads successively the video data of the image I, I32, the image B, B30, the image B, B31, and the image P, P35, in this order, then jumps to an immediately preceding GOP and reads the video data of image I, I22, to the first image P, P25. Next, the sensor 2 jumps to another previous GOP, and reads the video data of the image I, I12, to the first image P, P15. Next, the sensor 2 jumps to an additional preceding GOP and then reads the video data of the image I, I02, to the first image P, P05. After this, similar to that described above, the sensor 2 reads the video data of an image I positioned at the beginning of the next preceding GOP until the first image P positioned after that image I. 30

Said operation is carried out thanks to the fact that the PSM detector 40 detects the information of bytes_a_primera_img_P in a descriptor_ip_ipp in the input sector written at the top of each GOP and then supplies the detected information to the control circuit 6. More specifically, the control circuit 6 controls the sensor 2 to read, from the top of the input sector, the data corresponding to the number of bytes represented by the information of bytes_a_primera_img_P 35 provided in the descriptor_ip_ipp, thereby enabling the sensor 2 to read the video data in the order indicated by the arrows in Figures 9A to 9B.

The read video data comprised from the image I positioned at the beginning of the GOP to the first image P positioned after that image I is separated from the audio data and other data by means of the demultiplexer 32 and written to the buffer memory 10 video codes Then, the images B are deleted, whose identities are present in the respective image headers, and only the data of the images I and P are decoded and written to the memory bank 16, the video data thus written they are read from the frame memory 16 in the image display order shown in Figure 12C and are displayed on the display device 18.

Figure 13 is a data read / write timing diagram for bank 16 of 45 frame memories with a capacity of two frames. The writing of the decoded image I, I32, at the beginning of the last GOP in Figure 12B in frame memory 16a starts at time instant t0 and ends at time instant t1 after a period of one frame. Next, the writing of the image P, P35, decoded in reference to the image I, I32, in frame memory 16b begins at time t1 and ends at time instant t2 after a period of a frame. fifty

The reading of the image P, P35, of the frame memory 16b begins at an intermediate point between the two instants of time t1 and t2, provided that a field of the image P, P35 has already been written in the memory 16b of box at this time of reading start time. Therefore, it is possible to read from and record in the same frame memory 16b, simultaneously, delaying in one field the timing of the reading

with respect to write timing. Thus, in the frame memory 16b different image data can be written while, simultaneously, previously written image data is read from it.

The reading of the image P, P35, of the frame memory 16b ends at an intermediate point between the two instants of time t2 and t3, and the writing of the decoded image I, I22, of the preceding GOP in the memory 16b of box starts at the instant of time t2. This writing ends at the time of time t3 after a period of one frame.

Image I, I32, is read from frame memory 16a at an intermediate point between the two instants of time t2 and t3, and after approximately a delay of a field with respect to this instant of reading start, writing begins. of the decoded image P, P25, in frame memory 16a. A picture of image I, I32, is read completely 10 at an intermediate point between the instants of time t3 and t4 and then a picture of image P, P25, of frame memory 16a is completely read. In addition, a frame of image I, I12, of the preceding GOP is written in frame memory 16a, between time t4 and t5.

Decoded image data as depicted in Figures 12A to 12B are written in memories 16a and 16b of frames in the order I32, P35, I22, P25, I12, P15, I02, P05, ..., and they are read in the order that goes from the oldest (largest) image number to the most recent (minor) image number such as P35, I32, P25, I22, P15, I12, P05, I02, ...

As described hereinbefore, in the special playback mode one image I and two or one P images per GOP are displayed. It will be appreciated that the present invention can be modified to decode and display only one image I per GOP and eliminate both the P images and the B images. In this case, the information to detect the number of bytes until the end of the image I It is recorded on a PSD (Program Continuous Flow Directory). More specifically, in the directory of program streams defined in accordance with the MPEG system (ISO 13818-1), information relating to image I immediately after the PSD is recorded as a reference access unit, and 25 are added. with each other three values ​​of shift_position_head_head, offset_placement and bytes_to_report, in order to determine the data length (total number of bytes) from the first byte of the PSD to the end of image I.

When the storage capacity of frame memories exceeds three frames, more than three frames can be decoded and played back by GOP in the special playback mode. In such a case, the information representing the data length is written in the PSM, so that three or more P images appearing after the image can be accessed.

Although the examples above represent a jump to an adjacent GOP in the special playback mode, the jump can be performed to a distant GOP when performing special playback.

In the present embodiment of the invention in which the sensor 2 jumps in the special playback mode, the video data of the continuous video stream has different speeds, since the degrees of compression thereof are different depending on the types of image (I, P or B) or of the attributes of said images (flat or complex). Consequently, the search time is not fixed and some difficulties may arise when performing playback in FF / FR with equimultiple speeds. To avoid such difficulties, the system controller measures the search time or display interval, and the distance is changed to the next search according to the measured time, whereby the speed control is achieved by means of a speed control. feedback. For example, if any search has taken a longer time, the sensor 2 jumps to a slightly distant position in a GOP unit to gain the necessary distance.

Although this invention has been described in conjunction with an optical recording medium, it should be appreciated that the method and the recording / playback apparatus of the invention can be used with other recording media such as a magnetic disk or the like or to transmit data from compressed video

As with the present invention a special reproduction can be performed, such as a backward reproduction, with cheaper and less complex circuits, the dimensions of some components of the apparatus, including substrates and special circuits, can be reduced to consequently decrease the consumption of energy, which minimizes heat generation and minimizes the structure necessary for heat radiation.

In the special playback mode, an individual image I can be recovered, an image I and an image P can be recovered, or an image I and two images can be recovered. These settings can be selectively switched so that the speed of special playback by changing the number of images that should be loaded into frame memory and that will be displayed. By using one or two images P in addition to an image I, scenes can be displayed in a uniform manner to provide a satisfactory visual presentation.

Although preferred embodiments have been represented and described in the present invention, it will be readily appreciated that several changes can be made without departing from the scope of the invention.

10