Method of and apparatus for decoding movie images

The subject of the invention is a method and a device for decoding a film image, in particular for a coding-decoding system for converting a film image into a video signal, recorded on a medium.

High-performance video decoding is very important when recording a digital video signal. This especially applies to long films, recorded as video signals on small carriers with small information recording capacities. The Moving Picture Expert Group (MPEG) standard is an example of a high-performance coding standard for compressing a video signal by finding a data relationship between individual portions of video data, called frames. According to the MPEG standard, you find data dependencies between frames over time by searching for differences between consecutive frames contained in a video signal. This compresses the video signal in time dimension. The MPEG standard also relies on the data in the spatial dimension by processing the video signal within each video frame. Orthogonal transformations such as DCT discrete cosine transform and the like are used, which compress the video signal in a spatial dimension. There are three types of compressed MPEG frames: an I image frame that is compressed without being bound to other frames, a P image frame that is compressed from the preceding frame, and a B image frame that is compressed using both the preceding frame. and the next frame.

A collection of video frames is called a group of pictures (GOP), which is found in a sequence of movie images. Each group of pictures begins with a GOP start code, which defines the beginning of each group of pictures. The decoder recognizes the picture group by detecting the GOP start code, and then starts decoding at the appropriate point in the picture group. Figures 6A and 6B show an example of a picture group that consists of a video signal having nine frames F0 to F8. Type 1 picture frame is encoded

182 073 using an independent image information pertinent to it, and is therefore called a freeze frame. The P-picture frames are encoded using a preceding I-picture frame or a preceding P-picture frame. Such encoding is called forward-forecast encoding. B-type image frames are encoded with both the previous and the following frames. Such encoding is called bi-directional encoding. Since the P-picture frames and the B-picture frames different frames are used for encoding, these picture frames are called dependent frames. The first two FO, FI frames in Figures 6A and 6B are B-type image frames and have no relation to the previous frames. In this case, the B FO, FI image frames are coded with only the next reference frames. This coding is called forward-looking coding.

Figs. 7A to 7C show the timing diagrams for the encoding and decoding process of the group of pictures shown in Fig. 6A. The group of pictures is input to the encoder in the order shown in Fig. 7A. As will be further explained, the B B picture frame first fed into the decoder has the problem that the B picture frames cannot be coded without the B picture frame as the reference frame. Type I picture frame has to be prepared in advance, and here such a frame does not exist. To fix this, the encoder changes the order of the image frames. As shown in Fig. 7B, a picture frame of type 112 is coded first, although it actually appears in time after B-type picture frames. BO, BI picture frames are coded based on a picture frame of type 112. Then there are the remaining frames. The P picture frame P5 is coded by forward prediction using the picture frame 112. Then, the B picture frames B3, B4 are coded by bidirectional prediction using the previous reference picture 12 and the next time reference picture P5. In this way, the rest of the frames in the group of pictures are encoded and the resulting coded frames are fed to the encoder output and further fed to the decoder input as shown in Fig. 7B.

At this stage, the coded group of pictures may be transmitted over the air by recording and retrieving from a data carrier, or by some other known transmission method. The decoder decodes the group of pictures shown in Fig. 7B and outputs to its output in the order shown in Fig. 7C, thus reproducing the order of pictures shown in Fig. 7A. This allows the video signal to be reproduced correctly.

The MPEG-2 standard is a modified MPEG standard, especially suited for coding frames with interleaved images. As shown in Figures 8A and 8B, each frame MF1 through MF4 has top regions interleaved with bottom regions. These areas are treated by the MPEG-2 standard as separate frames that should be encoded sequentially. For example, the top area of ​​the first frame MF1 is encoded first, then the bottom area of ​​the first frame, the top area of ​​the second frame MF2, the bottom area of ​​the second frame, then redundantly repeated the top area of ​​the second frame, and so on. -2 uses a top region priority marker that determines the order of the regions in each frame. Thus, the top region priority marker of Fig. 8D is 1 when the top region appears first and 0 when the bottom region appears first.

The top region priority mark is especially important when the image group is converted from a movie to an interleaving video using a movie / TV signal converter. Following the 3: 2 split technique, each frame is converted into a two-area frame. For example, the input movie frames (30 frames per second, 60 areas per second in a moving picture) shown in Fig. 8B are made up of frames with two areas and frames with three areas. These frames are then converted to an interleaving video signal consisting of only frames with two areas. This effect can be achieved by removing unnecessarily repeated areas from each frame with three areas MF2, MF4, resulting in frames with two areas, as shown in Fig. 8C. The 3: 2 split transformation transforms frames with three areas into frames with two areas and hence its name.

182 073

The 3: 2 split converter needs to identify unnecessary duplicate areas in order to remove them. To identify redundant repetition areas in frames with three areas, MPEG-2 uses the First Region Repeat Marker to highlight redundant areas. Based on the first area repeat marker, the 3: 2 split converter identifies areas that are redundantly repeated and will be removed from split frames. For example, in Fig. 8E, the first area repeat marker for the first frame MF1 is set to 0, which means there is no unnecessarily repeated area in the first frame. In contrast, the first area repeat marker for the second MF2 frame is set to 1, which indicates that there is an unnecessarily repeated area in this frame.

From Fig. 8B it can be concluded (comparing frames MF2 and MF4) that both the top and bottom areas may be unnecessarily repeated areas. Based only on the repetition marker of the first region, a 3: 2 split converter cannot determine whether the top or bottom region is redundant. To determine whether the top or bottom region has been redundantly repeated, a 3: 2 split converter must analyze the state of the top region priority mark. When, for example, the top region priority marker (Fig. 8E), corresponding to frame MF2 in Fig. 8C, is equal to 1, the first area of ​​the second frame is the top area and it is considered redundant. When the top region priority flag is 0 and the first region repeat flag is 1, the bottom region is considered redundantly repeated as is the case of the fourth MF4 frame in FIG. 8C.

When the encoded interleaved video signal of Fig. 8C is to be decoded, redundant repeating areas removed during encoding must be reconstructed. This situation is shown in Fig. 8F, where unnecessarily repeated areas removed during encoding are recreated by repeating the first area of ​​a frame. To determine the deleted regions, the decoder analyzes the state of the top region priority mark and the first region repeat mark. For example, the top region priority flag for the second frame MF2 is 1, which means the top region occurs first, so the decoder should repeat the top region for the second frame. And the top priority flag for the fourth frame MF4 is 0, so the decoder repeats the bottom area. In this way, the decoder reconstructs the regions by converting the interleaved two-area video into a three-area movie image adapted to be displayed.

The upper and lower regions of the interleaving video are problematic during decoding, which will become clear when the encoding process is explained. Figs. 9A through 91 are timing diagrams of the interleaving video signal in an encoding process. The region synchronization signal of FIG. 9A, consisting of the high region signal and the low region signal alternately, serves to synchronize the encoding process for the top and bottom regions shown in FIG. 9B. Each of the series of areas in Fig. 9B fed to the encoder input is indicated by a letter designating the video frame type I, Β, P and a number indicating the display order.

The designations of further regions of Fig. 9B will now be described. The first area fed to the encoder is the bottom area of ​​the B BO picture frame, and the next area the top area of ​​the same frame of the same B BO picture. The next area fed to the encoder input is the lower area b1 of the next B image frame, then its upper area B1. The next area xl represents a redundant area being a repetition of the previous lower area b1. Sequentially, to the input of the encoder, the top area of ​​the I-type image frame 12 and the bottom area of ​​the I-type i2 image are transmitted. Further, the input of the encoder is given to the upper area of ​​the picture frame type Β B3 and the lower area of ​​the same frame b3. The picture frames are thus sent to the encoder input, up to the last lower area x5, which is a redundant area, being a repetition of the lower area p5 of the P-type picture frame.

The start frame marker (Fig. 9C) is set to 1 at the time when the corresponding frame is to be encoded. Before the encoding process, the encoder removes redundant areas by using the top area priority flag and the first area repeat flag (Figs. 9D and 9E) as already explained. As a result of conversion by division 3 2, blank areas are inserted in place of the repeated areas x1, x3, and x5 (Fig. 9F). Then

182 073 areas are coded as shown in Fig. 9F. In the encoding process, the order of the areas is changed so that the I picture frame is encoded before the B picture frames, which allows successive B picture frames BO, BI to be encoded bidirectionally using at least the I picture frame. For the same reason, the P picture frame P5 is coded before the B picture frames B3, B4, which again allows the B picture frames to be coded bidirectionally using at least the P picture frame P5. The encoder sets a frame start marker (Fig. 9G), which indicates the start of each group of pictures. The order of the top region priority flags and the first region repeat (Figs. 9H and 91, respectively) are fixed so as to transmit them together with their corresponding regions.

The coded signal at the output of the encoder of Fig. 9F is transmitted, e.g. by television broadcasting etc. to a receiver, where the received signal in the form of a series of areas shown in Fig. 10A is routed to a decoder. In order for the original moving image to be correctly displayed, the decoder input signal of Fig. 10A should be the same as the encoder output of Fig. 9F. The top region priority marks and the first region repeats (Fig. 10B and 10C) are, together with the encoded data, transmitted and used by the decoder to determine which areas have been removed during the already described 3: 2 division transformation. The decoder then decodes the coded areas to recreate the original movie image. Unnecessary areas removed during encoding are recreated. Assuming that the decoding process is performed without error, the signals at the decoder output and the encoder input (Fig. 9B) should be identical to ensure that the original video signal of the motion picture is accurately reproduced.

The timing of the decoding process is determined by the area synchronization clock signal shown in Fig. 10E. This signal is composed of the top region sync signal and the bottom region sync signal alternately. For simplicity, it is assumed that the decoder decodes the encoded frames without time delays, so that no area sync burst is lost during the decoding process.

Unfortunately, the first area given to the decoder input is the bottom area b0 as shown in Fig. 10D, and at this time the area sync signal points to the top area. That is, the lower region b0 must be delayed as indicated by xx at the location of the first region in Fig. 10D. This delay continues until the synchronization signal of the lower region arrives. So in order for the lower area of ​​bO to be decoded, it has to be delayed one cycle before being displayed.

It would be better to decode the bottom area first, as this area would be in sync with the bottom area sync cycle. However, when decoding the group of pictures, it is not possible to know in advance that the top area will be transmitted to the decoder during the low area sync cycle. This is due to the MPEG standard assumption that a group of pictures must be decoded and displayed in real time. So it is not possible to receive a group of images in advance in order to designate an area type.

One can compare the above possibilities, which are the source of potential problems, with Figs. HA to HE, which illustrate the desired situation when the bottom region is in sync with the bottom region sync cycle in the first region of each frame. In this case, the lower area b0 need not be delayed by one area synchronization cycle as it is already synchronized and can be displayed immediately after decoding.

In addition to delaying the display by one cycle, the decoder must stop retrieving the transmitted video signals during this cycle. This situation is shown in Fig. 10A where the decoder is stopped at the time denoted by ==. If a buffer to receive the transmitted encoder output is used, stopping the decoder will cause the buffer to overflow.

Moreover, when two groups of GOP1, GOP2 pictures are combined as shown in Fig. 12, a gap xx appears. This situation occurs when the last area in the first group of GOP1 pictures is of the same type (upper or lower area) as the first area of ​​the second group.

182 073 GOP images2. As shown in Fig. 12, the last region of the GOP1 picture group is the top region and it is synchronized with the top region synchronization cycle Tf. The second group of GOP2 pictures also starts with the top region, so the decoder must wait until it receives the next top region sync cycle, exiting the bottom region sync cycle. As a result, a gap xx is created. Since the setting of the first area in the second group of GOP2 pictures cannot be known in advance, it is necessary to decode the entire second group of pictures and check the positioning of the last area in the second group of pictures. When the second group of pictures contains a large amount of data, it takes a long time to decode. This degrades the decoder efficiency.

Moreover, from EP-0576289 A2, a method is known for encoding one of a pair of odd and even interleaved video signal fields in one of a plurality of frames of this signal, the field of the second frame is coded. The second of the pair of interleaved fields is coded such that at least a portion thereof is predictive coded with respect to a portion of the second field of the pair and with respect to the field of the second frame. The interleaved field pair is time-coded over the respective intervals of a coded field within a predetermined sequence. In-field and in-field decoding is implemented using prediction on single fields that are not encoded using bidirectional encoding.

EP-0588669 discloses a method of encoding an input video at a field frequency of 60 Hz resulting from moving motion pictures, which entails detecting duplicate fields in the fields of the input video signal. Each duplicate field is eliminated from the input video to form a progressive video signal of several frames having a frame rate of 24 Hz. The progressive signal is then encoded. Detecting the duplicate field entails generating a control signal at rt of each detected duplicate field, with the control signal used to encode the progressive signal.

A method of decoding a movie image, which decodes a movie image converted into a video signal encoded in the form of frames encoded as groups of video images containing an independent frame, the frames consisting of a plurality of areas with various settings, including fixed settings, with time control being controlled. start the process of decoding the appropriate groups of images, which is determined by the control means after detecting the presence of the first area in the individual picture groups, the area is a predetermined area, the individual picture groups are decoded in response to the signals generated in the control step, and the area synchronization signal is generated consisting of alternating cycles of area settings, for synchronizing the decoding process of said areas of different orientation, according to the invention, it is characterized in that during the control step, a start code is generated at the moment when the region synchronization signal is in a cycle for a predetermined region, and the decoding process of the first predetermined region is started by the decoding means.

Preferably, the decoded moving picture is displayed, scheduling the processes of the control step and generating a start code for two area synchronization signal cycles prior to displaying said first area.

Preferably, the decoding step stops when it is detected that the last picture of a given decoded group of pictures is a picture with a setting other than the preset setting.

It is preferred that the different area settings are an overhead and a low setting, and the first area is a coded area with an overhead setting.

It is preferred that the different area settings are the top and the bottom settings and that the last area in each group of pictures is the coded area with the top setting.

In a different solution, a film image decoding method, in which a film image is decoded into a video signal encoded in the form of frames encoded as groups of video images containing an independent frame, which frames are composed of

182 073 is provided from a plurality of regions with a different orientation, including a fixed orientation, and the encoded movie picture includes data denoting a first area of ​​a given group of images having a predetermined orientation, detecting a first area defined by said data as the fixed-orientation region prior to the decoding process. encoded film image, controlling the start point of the encoded moving picture decoding process at the time of detecting the first area, and decoding the encoded moving picture starting at the predetermined start time, according to the invention characterized in that in the control step a first area decoding start code is generated when the area sync signal is present in a cycle for an area with a fixed setting.

A movie image decoding apparatus for decoding an encoded movie image in the form of frames encoded as groups of video images containing a freelance frame, the frames consisting of a plurality of areas of varying orientation, including fixed orientation. the device is provided with a start time controller for the decoding process of the respective picture groups after detecting the presence of the first predetermined area in the individual picture groups, and connected thereto with means for decoding the individual picture groups, connected to a display having means for generating the area synchronization signal consisting of alternate cycles of area settings, to synchronize the decoding process of areas with different settings, according to the invention, it is characterized in that the controller is provided with means for generating a start code, and the decoding means is adapted to start a decoding process of the first fixed region when the region synchronization signal is on a cycle for a fixed region.

It is preferable that the controller is further provided with scheduling means coupled to the start code generating means which are adapted to generate the start code on two cycles of the area synchronization signal prior to displaying the first area.

Alternatively, a movie image decoding apparatus for decoding an encoded movie image in the form of frames encoded as groups of independent-frame video images, the frames consisting of a plurality of regions of various orientation, including fixed-setting, and the encoded movie image includes data. denoting the first area of ​​a given group of images having a predetermined orientation, the device is provided with detection means for detecting the first area denoted by said data as a predetermined area before starting a decoding process of the encoded moving picture, and a controller connected thereto for controlling the start of the decoding of the encoded moving picture when the first area is detected by the means detection and decoding of a film image based on signals from the controller, according to the invention, it is characterized in that the controller is provided with means for generating an area synchronization signal consisting of alternating area settings for synchronizing the decoding process of areas with different orientation.

It is preferable that the controller is arranged to generate a start code at the moment when the area synchronization signal is in a cycle for a fixed area, and the decoding means is adapted to start a decoding process of the first fixed area.

Using the solution according to the invention, the type of the first area of ​​each group of pictures is known in advance of the decoding process, so it can be matched to the area sync signal. As a result, a one-cycle delay in reading groups of pictures due to mismatch is not allowed. Moreover, decoder buffer overflow is not allowed because it is not possible for the decoding system to stop reading from the decoder. The invention furthermore avoids gaps between the plurality of picture groups because the first areas in each picture group are arranged so that gaps do not arise between the groups. In this way, multiple groups of pictures can be combined with no gaps between them.

182 073

Fig. 1 shows a block diagram of an encoder, Figs. 2A to 2G show timing diagrams showing the timing of the signals in the encoder of Fig. 1, Fig. 3 - block diagram of the decoder, Figs. Figs. 4A to 4E are timing diagrams showing the time distribution of the signals in the decoder of Fig. 3, Fig. 5 - syntax of the bit sequence, Fig. 6A, 6B - group of pictures (GOP) shown to explain the prediction coding mechanism. Figs. 7A to 7C - timing diagrams showing the order of frames to be processed into picture groups of Figs. 6A or 6B, Figs. 8A to 8F - 9A to 91 timing diagrams explaining the coding process of a group of pictures to which a 3: 2 division transform was applied, Figs. 9A to 91. 10A to 10E - timing diagrams explaining the coding process of the coded picture groups according to Figs. 9A to 91 and explaining the problem of out of sync of the area sync signal with the first area of ​​the group of pictures, Figs. HA to HE - timing diagrams explaining the group decoding process of images when the region synchronization signal is synchronized with the first region of a group of pictures, Fig. 12 is a diagram explaining combining multiple groups of pictures, Fig. 13 - a flowchart used in the procedure for determining the moment of generating the code of a group of pictures, and Fig. 14 is a flowchart used in the procedure of generating first and last region set marks

First, the encoder 1 used in the movie picture coding-decoding system for encoding and compressing television signals produced by processing the movie image with a movie / television converter in accordance with the MPEG-2 standard will be discussed. The signal processed in this way is the input movie signal of Fig. 8B.

A redundant region detector 3 detects the presence of redundant regions and sends the top region priority mark and the first region repeat mark to the redundant region remover 4 which removes redundant duplicate regions. The select converter 5 receives the remaining frames and converts them into blocks of coded data, which are then coded into I, B, P picture frames. These frames are recorded on a recording medium 2, such as an optical disk, a video tape or the like. Simultaneously, at the input of the encoder 7, the controller 8 sends the start code of the group of pictures GOP. The encoder 7 then starts the encoding process according to the state of the top region priority flag and the first picture repeat flag. The tags are also sent directly to the encoder to be written to the recording medium 2. The encoder 7 encodes the blocks of coded data according to the relationship between the series of frames according to the MPEG-2 standard, the state of the top region priority mark and the first picture repeat marker, the type of the picture data, and the marks indicating the start of the frames.

The encoding process will now be discussed in detail, starting with the conversion process through the 3: 2 division. A 3: 2 division transformation is performed on the video signal before inputting it to the encoder of Fig. 1. After the transformation, the interleaved video signal shown in Fig. 8B is obtained. When performing a 3: 2 split transformation, unnecessarily repeated top regions are produced, which form, for example, the MF2 and MF6 frames, and unnecessarily repeated bottom regions, which are, for example, MF4 and MF8 frames.

With the redundant area detector 3, redundant repeated areas are detected by examining whether a given area is an image formed by repeating two successive areas with the same orientation. The setting is top or bottom type. The test consists in assessing whether the sum of the absolute values ​​of the differences from the values ​​of the image elements of the two regions is less than a certain predetermined threshold value. A threshold value can be set according to the degree of similarity of the two regions. The redundant area detector generates a detection result (being the repeat flag of the first area) of 1 when the area is redundant, or 0 when the area is not superfluous. In this example, the detection result and the area itself are sent to the removal system

182 073 redundant regions 4, which removes all redundant repetition areas when the detection result is 1. Thus, redundant region remover 4 removes redundant repetition regions, resulting in a two-region interleaved video signal shown in Fig. 8C The above process also has the advantage of reducing the amount of transmitted data by skipping the repeat areas in the encoding process. The repeat mark of the first area is also transmitted to the controller 8 and to the encoder 7 to be recorded on the recording medium 2.

During the decoding process, the top region priority mark is needed to determine which of the regions (upper or lower) must be repeated in the next frame when the first region repeat marker indicates a redundant repeat region. This is because the top area repeat marker only indicates that the first area in a given frame needs to be repeated, and the first area may be either the top area or the bottom area. Therefore, it is not possible to determine from the first area repeat marker whether to repeat the top or bottom area of ​​the area. To solve this problem, the redundant region detector generates a top region priority marker to indicate which of the regions (top or bottom) is first within a given frame. Thus, it is distinguished which area should be repeated in the decoding process when the repeat mark of the first area is set.

The select converter 5 converts the frames remaining after processing by the redundant region 4 to a series of frames fD, F0, fl, FI, etc. as shown in Fig. 2B and passes them to the encoder 7. In this process, each area is treated. as a separate frame, and these frames are sequentially encoded with an encoder. The select converter 5 generates timers that synchronize each transformed frame and then sends them to the driver 8. The encoding of the frames is thus precisely synchronized. '

In the first example shown, the controller generates the region group GOP start code using the top region priority flag and the top region repeat flag. In this way, the displayed area at the beginning of a given group of pictures is the top area, and the last area of ​​this group is the bottom area. Each area processed in the selection converter into frames is encoded with an encoder. The encoding is performed using the encoding type data specifying how to encode the I, B, or P image frames generated by the controller based on the generated image group GOP start code. For example, the encoding data can describe which processed frames are I, B, and P image frames, and which frames are used for prediction coding of B and P image frames.

Figures 2A to 2G are timing diagrams explaining the encoding process with the encoder shown in Figure 1. The region synchronization signal (Figure 2A) provided from an external device such as a monitor synchronizes the coding of the given frame. The frames (Fig. 2B) along with the redundant areas x1, x3, and x5 are supplied to the input terminal 9. The frame start defining signal (Fig. 2C) generated by select converter 5 determines the start of each frame. The top region priority marks and the first region repeats (Figs. 2D and 2E), which are discussed in connection with Figs. 4B and 4C, indicate the order of the areas in each frame.

As already explained, the controller 8 generates a picture group start code GOP determining the start of each picture group when the top region appears, and determines the end of the picture group when the bottom region appears. The controller 8 determines which of the remaining frames are to be inserted into a given group of images between the first and last frames by adaptively selecting consecutive frames. According to the MPEG standard, a P frame must be at least every third frame, with two frames in between, and an I frame every sixth. However, according to the invention, the number of frames is adaptively modified to meet the condition that the first area in the group of pictures must be the top area and the last area must be the bottom area. problems because the I-picture frames are independent frames, that is, they are not predictively coded as predicted. Otherwise, if the spacing

182 073 between B and P image frames are adaptively selected, this changes the compression structure of B and P image frames.

According to the invention, the GOP start code of the picture group starts with the top region synchronization cycle. However, the GOP start code of a picture group may also be present with the occurrence of a lower region. In the case where the area sync signal starts with a low area sync cycle Bf, the first area is required to be the bottom area and the last area to be the top area. Then the areas will be synchronized according to the area synchronization signal.

In the preferred example, the controller 8 terminates the current group of pictures, e.g. GOP1, with the area in the lower position. It is preferred that the lower area x1 of Fig. 2B is a redundant area, and if it appeared at the end of the group of pictures, it could be skipped. Thus, according to the invention, unnecessary lower regions are selected as the last regions in the picture group. Note that the GOP start code of the picture group Fig. 2F for the next group of pictures begins exactly with the occurrence of the top regions F2 and F6. As shown in Fig. 2G, the data specifying the type of coding, i.e., I, B and P picture frames, produced by the controller 8 shows that the starting frame of a picture group is a B picture frame and the ending frame of a picture group is a picture frame. P. Also shown is data specifying the type of encoding for an I picture frame when the P picture frame is coded and compressed.

Upon receipt of the GOP start code of the picture group, the encoder compresses and codes each received frame into an I-type, P-type, or B-type picture frame using the coding type data shown in Fig. 2G. The first pair of areas F2, f2 of the group of GOP2 pictures are encoded as a B-type picture frame and the second pair of areas F3, f3 as an I-picture frame. These frames are encoded and transmitted as a bit sequence in accordance with the MPEG standard. The encoder appends the header at the beginning of the given bit string representing the group of pictures containing the encoded frames. The header may contain information such as the length of the bit string for a given group of pictures, as well as other data helpful in extracting and decoding the encoded frames. Then, by means of an encoder, a top region priority mark and a first region repeat marker are attached to each frame. Thereafter, the encoded bit string is transmitted and stored on the recording medium 2, preferably in the form of an optical disc, according to the syntax defined by the MPEG-2 standard (ISO / IEC 13818-2).

By means of the controller 8, the start code of the picture group GOP is generated at the moment when the first area of ​​the picture group is aligned with the synchronization signal of the areas on the decoding side. For example, as shown in Figs. 2A through 2G, the controller causes the first areas in each of the three groups of pictures GOP1, GOP2, and GOP3 to be areas in the up position, which allows synchronization with the top area sync cycle (Fig. 2A). while decoding. The controller also directs the setting of the last area by deleting the start code of the group of pictures, and as can be seen from the example of Figs. 2A to 2G, causes the last area to be in the down position. As a result, bit strings representing groups of pictures are recorded on the optical disk constituting the recording medium 2, the first area of ​​which is in the upper position and the last area in the lower position.

The solution of the movie image coding-decoding system is not limited to storing the encoded information on the recording medium. The video signal may be transmitted to the receiver by other means, for example by radio, cable, Internet, videotape, etc.

After transmission, the image data recorded on the optical disk constituting the recording medium 2 is reproduced by the decoder 10 shown in Fig. 3 The read bit sequences are sent to a decoder 14 which decodes them according to the MPEG standard, and the decoded signals are displayed on the display 15. The display 15 produces an area synchronization signal for the decoder and sends it to both the display start determining circuit 11 and the decoding start controller 12. This signal, together with a control signal from the host computer (e.g.

182 073 (not shown in the drawing of the microprocessor) connected to the input terminal 16, controls the switch 13, thus starting the decoding process.

Now, a decoding process performed according to the method of the invention will be discussed. An encoded image is reproduced from recording medium 2 in the form of a bit string representing the encoded image groups and sent to switch 13. A display defining circuit 11 receives encoding type data from a host computer via input terminal 16 to determine when to start decoding image groups. For example, data specifying the type of encoding means that a given picture should be decoded at the beginning or at the center of a given group of pictures, with respect to the region sync signal (Fig. 4E).

As shown in FIG. 4E, arrows indicating specific cycles of the region synchronization signal generated by the display 15 show the order in which the images have started to be displayed. Looking towards the timeline, the second arrow indicates where to start decoding and displaying the group of pictures. However, the point at which the decoding process begins is advanced two cycles from the first arrow. This is because two cycles are needed to process the group of pictures prior to the decoding process

The display start time is determined by three parameters: the synchronization signal of the areas forming the areas defined by the arrows, data specifying the type of encoding indicating the point at which decoding of each group of pictures begins, and the bit string representing the picture groups read from the optical disk constituting the recording medium 2. The data determining the start time of the display is sent to the decoding process start controller 12, which calculates the start time of the decoding process based on this data. When decoding starts, the decoding start controller 12 turns on the switch 13, thereby allowing the bit string representing the group of pictures to be sent to the decoder 14. Similarly, the display start controller 11 turns off the switch 13 at the end of the decoding process, cutting off the flow of the bit string representing the group of pictures to the decoder 14.

The group of pictures is encoded such that the first picture is set to a predetermined position and the last region is set to a position opposite to the first region. The first area of ​​the picture group is preferably coded as the top area and the last area of ​​the picture group as the bottom area.

The positioning of the first area is known before the picture group decoding process, therefore, during decoding, the start of the decoding process is determined at the start of the top area as shown by the second arrow in Fig. 4E. For simplicity, it is assumed that the decoder decodes the coded area without any delay. However, as already mentioned, two cycles of the area sync signal are used by the display start determining circuit 11 to create display start data and by the decoding start controller 12 to determine the switch-on time of the switch 13. Accordingly, the decoding start controller 12 turns on switch 13 two cycles before the designated start time as shown by the first arrow in Fig. 4E. This allows the decoder 14 to decode the groups of pictures two cycles later, at the time the first frame is to be displayed.

Now, the process of decoding a group of pictures will be explained with reference to the timing diagrams in Figs. 4A to 4E. The coded frames consisting of the top area and the bottom area are coded and transmitted to the decoder input in the sequence shown in Fig. 4A. I, B, and P picture frames are marked with appropriate uppercase or lowercase letters denoting the upper and lower positions of the areas, respectively. Then there is a number that indicates the order in which the frames are displayed. The top region priority flag (Fig. 4B) indicates which area (top or bottom) is first for a given frame, and the first area repeat flag indicates whether the first area of ​​a given frame is to be repeated in the next frame. According to the invention, the encoder-generated stop signal denoted by is inserted at the position where

182 073, the first area should be repeated. The occurrence of a stop signal causes the decoder 14 to stop for the duration of one area and redisplay the area indicated by the first area repeat marker. Fig. 4D shows the output of the decoder 14 in which the order of the frames in the group of pictures is restored as it was before the encoding process.

It is advantageous that the display cycle of the region synchronization signal does not have to be matched to the position of the regions at the decoder input. This prevents the development of a one area delay caused by the region synchronization signal mismatching with the first region alignment, and prevents the decoder 14 buffer from overflowing during the delay.

Each group of images is complexed so that the first area is set to a fixed position and preferably it is the top area. The last area, on the other hand, should be the bottom area. This ensures that there is no gap between the image groups, which allows you to combine multiple image groups.

Also, it is not necessary to decode the whole group of pictures any further to determine the last area setting, since this setting is known before decoding is complete. So it is possible to efficiently combine and change image groups, even when they are long and contain large amounts of data.

The decoder is compatible with encoders of other types than the type described with reference to Fig. 1. Using other encoders, it may not be possible to determine the orientation of the first display area in the group of pictures before decoding is completed. Therefore, it may turn out that the first decoded and displayed area is out of sync with the area synchronization signal produced by the display 15. This causes the decoder 14 to freeze for one cycle to match the first area to the area sync signal. This in turn halts the decoding process for one cycle and the decoder 14 stops reading data from its buffer (not shown) for one cycle. As a result, the picture data accumulates in the decoder buffer, so when the bit rate is high, the buffer may overflow.

To solve this problem, an additional memory buffer is provided in the decoder to store the next area to the area currently being processed. An additional buffer size B (not shown) can be calculated based on the size of the receive buffer VBV and bit rate R stored in the header of the group of pictures. The additional buffer capacity B can be calculated using the following formula.

B = VBV + R x (duration of one area)

Due to the additional B buffer, the main receive buffer will not overflow when the first area to be displayed is out of sync with the area synchronization signal.

Now, referring to the flowchart of Fig. 13, the operation of the controller 8 of Fig. 1 will be explained. The flowchart shows a procedure for determining when to generate a GOP start code of a picture group performed by the controller 8. The controller 8 starts operation at step 101 with initialization of the variable and value. -1. The variable i means the sequential number of the frame prepared by the selection converter 5 and sent to the encoder 7. Then the controller 8 checks whether there are further frames ready for decoding. If these frames are available, the program proceeds to step 102. If there are no more frames waiting to be decoded, controller 8 exits.

In step 102, the controller 8 initializes the variable with the value -1. The variable n here denotes the sequence of frames within a given group of pictures. In step 103, the driver 8 increases the value of the variables i and no one, i.e., before the start of the nested loop, both variables i and n have the value 0. In step 104, the driver 8 checks if the variable n is equal to 0, which means, that the current frame is the first frame in the image group. If the variable n is 0, the program proceeds to step 105, where the GOP start code of the picture group is set to 1. Otherwise

182 073 The current frame is not the first frame in the picture group and the controller continues with step 106 where the GOP start code of the picture group is set to 0.

In steps 107 to 111, a controller 8 determines the type of frame coding. Frames are identified as I, B, or P picture frames (steps 109, 111, and 110). For example, when checking in step 107, controller 8 judges whether the variable n is odd and if so, program execution continues at step 107. step 108. Otherwise, step 111 is performed, where the current frame is defined as a frame of the B type image. In step 108, the controller 8 determines whether the value of the variable n, which turned out to be odd, is 1, and if so, in step 109, the current frame is determined by appears as an I frame. Otherwise, in step 110, the current frame is marked as a P frame.

In steps 122 through 114, the controller 8 checks when the given group of pictures is to end. The controller 8 appropriately determines the end of the group of pictures so that the last displayed area is the lower area. As a result, groups of images can have a different number of N frames. In the checking step 112, the controller 8 determines whether the current frame is a frame which no longer belongs to the given group by checking whether the current frame number n is greater than or equal to the last frame number N minus one. If the controller 8 determines that the frames of the next group of pictures have not yet been reached, program execution continues from step 103, where the variable values ​​are incremented by one. Otherwise, controller 8 determines that a frame from the next group has been reached and continues with step 113.

In step 113, the steering wheel 8 judges whether the area to be displayed at the end of the current frame is the lower area. Judgment is made on the basis of the priority flag of the top region Tff and the repeat flag of the first region Rff. For example, if the top region priority marker is 1 and the first region repeat marker is 0, the first region is the top region of the current frame and should not be repeated in the next frame. If the top region priority marker is 0 and the first region repeat marker is 1, the first region is the bottom region and repeats on the next frame. In such a case, the controller determines the termination of the picture group GOP and, in step 114, it generates a picture group decoding complete signal. This means that the last area in the image group is the bottom area.

It is preferred that the next area being the first area of ​​the next group of pictures is the top area because the last area of ​​the previous group was the bottom area. As the next group of pictures is processed, the controller continues to step 103 where it increments the value of the variable by one. The above-described operation procedure of the controller 8 is repeated for successive groups of images until the cinematographic image is completed.

The encoder 1 shown in Fig. 1 may operate according to another procedure performed by the controller 8. In this case, the controller 8 changes the state of the first picture alignment marker that indicates the setting of the first area of ​​the first frame in a given group of pictures and the alignment of the last region alignment marker that is indicates the setting of the last area. The state of these marks is recorded on the recording medium. In this example, there are no restrictions on determining the start and end points of a group of pictures, since the setting of the first and last areas need not be upper and lower, respectively.

Preferably, the settings of the first and last areas are transmitted in a user data area designated according to the MPEG standard for use by the user. Fig. 5 shows the meaning of the bit string for the group of pictures with regard to user data. A group of pictures begins with the 32-bit word group_start_code. The user data area begins with the 32-bit user_data_start_code followed by the first area set bit (first_field_polarity_GOP) and the last area set bit (last_field_polarity_GOP). In the preferred example, a value of 1 as the first area set marker or the last area set marker is the top setting, and 0 is the bottom setting. User data is

182 073 grouped into units of one byte size, and therefore a six-bit padding (denoted as reserved) is necessary to make the two bits of area setting up to the size of a full byte.

By means of the encoder 7 successive frames are encoded according to a predetermined type of coding into I, P or B picture frames. The coding device in the example shown encodes I picture frames at intervals of four frames. The interval for P-type image frames is set to two frames in the image group. It is preferred that the number of frames in each of the image groups does not need to be appropriately selected because the region alignment is indicated by the region alignment markers, and the first and last regions do not necessarily have to be set to a given position (top and bottom positions, respectively). Top area priority and first area repeats are attached to each frame as in the previous example. In addition, each picture group header containing the user data is appended to each picture group when the picture group start code is detected. The meaning and setting of the bits in Fig. 5 is defined by the MPEG-2 standard (ISO / IEC 13 818-2). These bits are recorded on the optical disk constituting the recording medium 2. As shown in Fig. 5, a first area set mark and a last area set mark are recorded on the optical disk. Both tags are included in the user data of the image group header.

The decoding device for the coding device in the modified system is similar to the device shown in Fig. 3. This device differs from the previously described in the way the decoding start controller 12 controls the switch 13. In this example, the display origin designator 11 reads from user data stored in the header of the GOP start code of the picture group, area setting information. It then determines the area display start time corresponding to the area sync signal produced by the display 15. In the example shown in Fig. 4D, the setting of the areas in the group of pictures to be displayed is such that the first area is the top area. Therefore, from the region synchronization signal, the controller selects the top region cycle as the display start point. This is shown in Fig. 4E.

The operating procedure of the controller 8 of Fig. 1 for generating the first and last region alignment marks will be described in detail based on the flowchart shown in Fig. 14. The steps of this procedure are similar to the steps of the procedure described in the flowchart of Fig. 13. The controller 8 starts with variable initialization and a value of -1 at step 202. If there are no more frames waiting to be decoded, the driver exits.

In step 202, the controller 8 initializes the variable n with the value -1. In step 203, the controller 8 increases the value of variables i and no one, i.e. before the start of the nested loop, both variables i and n have the value 0. In step 204, the controller checks whether the variable n is equal to 0, which means that the current frame is the first frame in an image group. If the variable n is equal to 0, the program proceeds to step 205, where the picture group start code is set to 1. Otherwise, the current frame is not the first frame in the image group and the controller 8 continues with step 206 where it sets the start code of the image group to 0.

Contrary to the previously described example, here the controller 8 generates a first area alignment marker by setting its value to 1 As already explained, this marker is transmitted along with the current frame pointing to the first region of the corresponding group of pictures.

In steps 208 to 216, the same operations are performed as in corresponding steps 108 to 116 in the previous example. In steps 208 to 212, a controller 8 determines the type of frame coding. The frames may be identified as I, B, or P picture frames (steps 210, 211, and 212). For example, when checking in step 208 with controller 8, it judges whether the variable n is odd, and if so, program execution continues with step 209. Otherwise, step 212 is performed where the current frame is referred to as a B frame. In step 209, using the control room.

182 073 ka 8, it is determined whether the value of the variable n that is determined to be odd is 1 and if so, in step 210 the current frame is defined as an I frame. P.

In steps 213 to 215, the controller 8 checks when the given group of pictures is to end. As in the previous example, the controller 8 appropriately determines the end of the group of pictures so that the last displayed area is the lower area. As a result, groups of images can have a different number of N frames. In checking step 213, the controller determines whether the current frame is a frame that no longer belongs to the group by checking whether the current frame number n is greater than or equal to the last frame number N minus one. If the controller determines that no more frames have been reached from the next group of images, program execution continues from step 203, where the variable values ​​are incremented by one. Otherwise, the controller determines that a frame from the next group has been reached and continues to check in step 214 where the start code of the group of pictures is set to 1.

The controller then continues to step 215 where it judges whether the area to be displayed at the end of the current frame is the bottom area. Judgment is made on the basis of the priority flag of the top region Tff and the repeat flag of the first region Rff. For example, if the top region priority marker is 1 and the first region repeat marker is 0, the first region is the top region of the current frame and should not be repeated in the next frame. If the top region priority mark is 0 and the first region repeat marker is "1, the first region is the bottom region and it is repeated in the next frame. In this case, the controller determines the end of the group of pictures and sets the last area set mark to 1. As with the first area set mark, the last area set marker is transmitted with the current frame indicating that the current area is the last in the group of pictures. Otherwise, the controller sets the value of the last area set marker to 0.

As in the previous example, the next area, which is the first area of ​​the next group of pictures, is the top area because the last area of ​​the previous group of pictures was the bottom area. As the next group of pictures is processed, the controller continues with step 203 where it increments the value of the variable by one. The described controller operation procedure is repeated for subsequent groups of images, until the end of the movie image.

Thus, according to the examples above, it is possible to know the setting of the first area of ​​a group of pictures before starting the decoding of the group. As in the case of the first example, for simplicity it is assumed that the decoder 14 decodes the frames without delay. However, the decoding process start controller 12 causes the decoding process to begin two cycles of the area sync signal ahead of the designated start time as shown in FIG. 4E. This delay is introduced to allow the time needed to process the frames prior to the decoding process and to accurately synchronize the display of the decoded images.

It is also possible to know the positioning of the areas at the beginning and end of a group of pictures, which avoids gaps in the data between consecutive picture groups. Therefore, the examples described allow for the joining of multiple groups of pictures and do not create gaps between the groups.

182 073

<td>Tf Bf Tf Bf Tf Bf Tf Bf Tf Bf Tf Bf Tf Bf Tf Bf Tf Bf Tf Bf Tf</td><td>FO FI F2 F3 x3 F4 F5 F6 F7 x7 fO f1 x1 f2 f3 f4 f5 x5 f6 f7</td><td>T. T— V T- T. Ύ— V</td><td>V- o About τ- Ο O T Τ- Ι-- O ABOUT OO</td><td>Γ Ί—</td><td>CD ABOUT. CD CD 0. CD</td><td>ca C (! c 0 C c cce AND</td><td>0 L. 5 &lt;4 L. 5 5 u) 5 AND</td>

<img file="PL182073B1_D0001.tif" />

<img file="PL182073B1_D0002.tif" />

LL LL

U-CL

UO

CC0

CM Ó mm

<img file="PL182073B1_D0003.tif" />

<img file="PL182073B1_D0004.tif" />

<img file="PL182073B1_D0005.tif" />

182 073

<img file="PL182073B1_D0006.tif" />

182 073

<img file="PL182073B1_D0007.tif" />

&lt;mo olu 'ϊvr

0 0 00 ttssss tMn BCBsn mm · ημμ

182 073

MEANING NUMBER OF BITS group_of_pictures () {group_start_code32 time_code25 closed_gop1 broken_link1 if (nextbits () == extension_start_code) {extension_start_code32 while (nextbits () l = '0000 0000 00000000 0000 0000Γ) {group_extension_data8} (next_extension_data8) == user_data_start_code) {user_data_start_code32 first_field_parity_GOP1 last_field_parity_GOP1 reserved6 while (nextbits () l = '0000 0000 00000000 0000 00001') {user_data8} next_start_code ()}

do {picture ()} while (nextbits () == picture_start_code)

FIG.5

182 073

IMAGE GROUP (GOP)

<img file="PL182073B1_D0008.tif" />

FORWARD FORWARD (P PICTURE)

CGOP IMAGE GROUP)

<img file="PL182073B1_D0009.tif" />

DWL / KEEPER FORECASTING (PICTURE T / PU B)

FIG.6B

182 073

CN

CN co O.

mo CD

<img file="PL182073B1_D0010.tif" />

every eight wounds LŁ

182 073

FIG.8A

MOVIE (24 Ι «ΑΤΚΙ / 5ΕΚή

MF1

MF2

MF3

MF4

FIG.8B

HEJŚILOkD PICTURE RLMONY / PLANE SY6NAŁ NIIYM (30 FRAMES / SEC, 60 OBSMÓN / SEC)

<img file="PL182073B1_D0011.tif" />

EXCESSIVELY f ^ EÓRZDKLATKI FOR VET. CODER

FIG.8D

WE MEAN PLEAWSENΜΑΜΑ &amp; 012N £ 6fl AREA 7F

<img file="PL182073B1_D0012.tif" />

BF

EXCESSIVE AREA

FIG.8E

ZUAURtk POHTÓIZENIA PtERN $ ZE60 AREA

F1G.8F

KLATkl MA A / YL. DECODER

REPEATED TF AREA

REPEATED Bp area

<img file="PL182073B1_D0013.tif" />

182 073

<img file="PL182073B1_D0014.tif" />

182 073

<img file="PL182073B1_D0015.tif" />

ο ο ο οο

0 0 00

MM MBW WIMU ΜΜ! Μβ

L LL LŁ LŁLL

182 073

<img file="PL182073B1_D0016.tif" />

182 073

<img file="PL182073B1_D0017.tif" />

<img file="PL182073B1_D0018.tif" />

CM Υ “· 0 LL

182 073

<img file="PL182073B1_D0019.tif" />

NOPE

<img file="PL182073B1_D0020.tif" />

SUCH-

i) —102-103

Ϊ ++ Π ++

104

<img file="PL182073B1_D0021.tif" />

NOPE

YES

105 --- GOP start [i] = 1 n == 0

106

GOP_start [i] = 0

107

108

<img file="PL182073B1_D0022.tif" />

NO n% 2 == 1

YES n == 1

YES

110

111 ______ Ϊ ____ pc [i] = P pc [i] = B

112

NOPE

NOPE

<img file="PL182073B1_D0023.tif" />

YES

114-— GOP end [i] = 1

YES nc (Nl) (tff i == 1 &amp;&amp; rff [11 == 0) II (tftn == 0 &amp;&amp; rff [il == 1) tff: top_field_first rff: repeatjirstjield pc: picture_coding_type

FIG. 13

182 073

<img file="PL182073B1_D0024.tif" />

FIG.14 tff: top_field_first rff: repeat_first_field pc: picture_coding_type

182 073

<img file="PL182073B1_D0025.tif" />

Publishing Department of the Polish Patent Office. Circulation of 60 copies

Price PLN 4.00.