Nova Patents
EP0933947A2

Adaptive quantization compatible with the jpeg baseline sequential mode

Abstract

Pixel blocks (102) of an input image are type classified based on an analysis of pixel values for each respective pixel block (102). A discrete cosine transform (DCT) is performed on the pixel values of each pixel block (102), and a quantization modification process thresholds and/or quantizes the resulting DCT coefficients (104) based on the type classification of the respective pixel block (102). Once the coefficients (104) are modified in this way and encoded, the resulting data can be decoded and dequantized in compliance with the standard JPEG sequential mode data syntax in order to construct a perceptually faithful representation of the image, without passing any additional information to the decoder concerning the quantization modification.

EP0933947A2, drawing sheet 1
Sheet 1 of 28

Term

Term ended

Projected expiry passed 13 January 2019, 7.7 years ago.

  1. Priority
  2. Filed
  3. Published
  4. Projected expiry
  5. Today

10 claims: 7 independent, 3 dependent

  1. 1
    A method for encoding data representing an image to produce output data from which the image can be effectively restored by a decoder, comprising:inputting image signal data representing at least one plane of the image, the image signal data including a first pixel block and a second pixel block, the first and second pixel blocks comprising data for a plurality of pixel rows and pixel columns in respective regions of the image;obtaining first coefficients by performing a transform on the first pixel block;determining a first image type based on an analysis of the first pixel block;determining a first modification table optimized for quantizing coefficients representing an image of the first image type;determining a fine quantization table optimized for quantizing base image coefficients that could be produced by performing the transform on a base image type different from the first image type, the fine quantization table differing from the first modification table;obtaining first modified coefficients by modifying the first coefficients based on both the first modification table and the fine quantization table;and producing output data from which the image, including said respective regions of said image, can be effectively restored by the decoder;said output data including data representing the first modified coefficients and information concerning the fine quantization table, but being devoid of information describing said first modification table.
  2. 5
    The method of any one of claims 1 to 4, wherein determining the first or second image type based on the analysis of the first or second pixel block comprises obtaining texture values along the pixel rows and pixel columns by a spatial domain analysis of the data for the plurality of pixel rows and pixel columns.
  3. 6
    The method of any one of claims 1 to 4, wherein determining the first or second image type based on the analysis of the first or second pixel block comprises:obtaining texture values along the pixel rows and pixel columns by a spatial domain analysis of the data for the plurality of pixel rows and pixel columns;calculating histogram values to determine a numerical distribution of the texture values obtained for the pixel rows and pixel columns;and determining the image type based on the distribution of the texture values.
  4. 7
    The method of any one of claims 1 to 6, wherein the transform is a discrete cosine transform.
  5. 8
    The method of any one of claims 1 to 7, wherein the fine quantization table is a text-optimized quantization table and the decoder is JPEG compliant.
  6. 9
    An apparatus for encoding an input data signal employing the method of any one of claims 1 to 8.
  7. 10
    An apparatus for encoding an input data signal representing an image, comprising:a transform module (1110, 1310) that receives data representing a block-like region of one plane of said image and performs a transform thereon to obtain transform coefficients (104) corresponding to the data representing said region;a segmenter (1112, 1312) that characterizes said block-like region as being of a first image type based on one of (i) the transform coefficients (104) obtained by the transform module (1110, 1310) and (ii) a spatial domain analysis of the data representing said region;a modified quantizer (1116, 1316) that receives the transform coefficients (104) from the transform module (1110, 1310) and obtains modified coefficients by modifying the transform coefficients (104) as a function of both (i) a first quantization table optimized for quantizing the first image type and (ii) a fine quantization table optimized for quantizing a base image type different from the first image type;and an output module (1120, 1320) that outputs the quantized coefficients along with information that identifies the fine quantization table, but does not identify the first quantization table.
Independent claims7