MXPA98010155A - Image processing method, image processing apparatus, and da storage media - Google Patents

Abstract

The invention relates to an image coding apparatus comprising a DCT unit for transforming an image signal corresponding to a reference block that is processed into frequency components, a quantizer for quantifying the output of the DCT unit for generate coefficients of difference which are differences between the transformation coefficients of the reference block, and an identification cycle to identify whether or not the values of the difference coefficients are within the range of numerical values which can be expressed by length codes Fixed N-bits. When the values of the difference coefficients are within the range, these difference coefficients are coded. When at least one of the values of the coefficients is out of range, the transformation coefficients of the reference block are coded. Therefore, when the transformation coefficients, which have been obtained in a comprehensive manner by transforming an image signal, are encoded using predetermined code words, the predicted intra-frame coding of these transformation coefficients can be reversibly carried out without reduce coding efficiency

Description

METHOD OF PROCESSING OF IMAGES, APPARATUS OF PROCESSING OF IMAGES, AND MEANS OF STORAGE OF DATA FIELD OF THE INVENTION The present invention relates to an image processing method, an image processing apparatus, and a data storage medium and, more particularly, to a method and apparatus for predicted intraframe encoding or intraframe predicted decoding of a image signal, and a data storage medium containing a program that carries out the coding or decoding of an image signal by software or set of programs.

BACKGROUND OF THE INVENTION To store or transmit digital image information with high efficiency, it is necessary to compressively encode the digital image information. As a typical method for compressive coding of digital image information, there is DCT (Discrete Cosine Transformation) represented by image processing technologies based on JPEG Ref. 029027 (Board of Photographic Experts Group) and MPEG (Group). of Experts in Mobile Images). In addition, there are methods of waveform coding, such as subband coding, small wave coding, and fractional coding.

In addition, to eliminate redundant image information between adjacent display images, such as frames, interframe prediction is carried out using motion compensation. That is, a point value of a point in a reference frame that is presently processed is expressed using a difference between this point value and a point value of a point in the previous frame, and this difference signal is subjected to waveform coding.

A description of an image coding method and an image decoding method based on the MPEG is now given, including the ~ DCT with motion compensation.

In the image coding method, initially, an input image signal is divided into a plurality of image signals corresponding to a plurality of blocks (macroblocks) that make up an image (a frame), and the image signals corresponding to the particular macroblocks are coded block by block. A macroblock is a visual presentation area of an image that comprises 16x16 points, in a frame. When an image signal corresponding to an arbitrary object image is input, the image signal is divided into a plurality of image signals corresponding to a plurality of blocks (macroblocks) that make up a visual presentation area corresponding to the subject image in a box (object area).

The image signal corresponding to each macroblock is further divided into image signals corresponding to a plurality of sub-blocks, each sub-block corresponding to an image presentation area comprising 8x8 points, and the image signal corresponding to each sub-block is subjected to DCT to generate DCT coefficients of each sub-block. Subsequently, the DCT coefficients of each sub-block are quantized to generate quantization coefficients (hereinafter, also referred to as transformation coefficients). The method of encoding the image signal corresponding to each sub-block by DCT and quantization as described above is called "intra-frame coding".

At the end of reception, the quantization coefficients corresponding to the particular subblocks are successively subjected to inverse quantization and inverse DCT to reproduce the corresponding image signal to each macroblock.

Meanwhile, there is a method of encoding an image signal called "interframe coding". In this coding method, initially, an area comprising 16x16 points is detected and having a smaller error in the point value of a macroblock that is presently coded (hereinafter referred to as a reference macroblock) as a prediction macroblock of an image signal corresponding to a frame, which has already been coded and temporarily is adjacent to a frame that is presently coded (hereinafter referred to as a frame of reference), by a method of detecting the movement of the frame. picture in the picture, such as the matching of blocks.

Subsequently, the image signal of the prediction macroblock is subtracted from the image signal of the reference macroblock to generate a difference signal corresponding to the reference macroblock, and this difference signal is divided into a plurality of signals corresponding to the subblocks each which comprises 8x8 points. Subsequently, the difference signal corresponding to each sub-block is subjected to DCT to generate DCT coefficients, and the DCT coefficients are subjected to quantization to generate the quantization coefficients (transformation coefficients).

Even when the input image signal corresponds to an object image, the interframe coding for the image signal is carried out in the same manner as described above.

At the end of the reception, the quantization coefficients (quantified DCT coefficients) of the particular subblocks are successively subjected to inverse quantization and inverse DCT to restore the corresponding difference signal to each macroblock. Subsequently, a prediction signal for an image signal corresponding to a reference macroblock in a reference frame is generated from an image signal of a frame which has already been decoded, by motion compensation, and then the signal of prediction and the difference signal restored are added to reproduce the image signal of the macroblock.

By the way, in an ordinary image, there are many areas that have similar visual representation data such as shape and luminance in the same frame, and it is possible to approximate the image signals to these areas in the frame using this image property.

Therefore, in recent years, intraframe predicted coding has been used as a method to approximate image signals as described above, in image processing technologies based on ITU-T (International Telecommunication Union - Standardization Sector). of Telecommunications) and MPEG4.

It is specific that, in the intraframe predicted coding, an image signal corresponding to a sub-block comprising 8x8 points is subjected to DCT (frequency transformation) to generate DCT coefficients (frequency components), and these DCT coefficients are quantified for generate quantification coefficients (transformation coefficients). The values of the quantization coefficients (hereinafter also referred to as quantization values) are in a range from "-127" to "127".

Then, using the quantization coefficients of an adjacent block as prediction coefficients, the differences between the prediction coefficients and the quantization coefficients of the reference sub-block that are processed are obtained, and the difference coefficients are coded. The prediction coefficients are called "intraframe prediction signal" or "intraframe prediction coefficients".

Next, the predicted intraframe encoding will be described in greater detail using figure 3. In the description, "block" means "sub-block".

In Figure 3, Bx is a reference block which is presently coded, and Ba, Bb and Be are adjacent blocks which are placed above, to the left of, to the upper left of the reference block Bx, respectively. Each of the blocks Ba, Bb and Be have a plurality of quantization coefficients obtained by transforming the corresponding image signal. In Figure 3, these quantization coefficients are shown by black dots corresponding to the positions of the points in the blocks Ba, Bb and Be.

More specifically, a quantization coefficient Qxl placed in the upper left hand corner of the reference block Bx is a quantized DC coefficient obtained by quantifying a DC component among the plural frequency components (DCT coefficients) of this block., and other quantization coefficients in the reference block Bx that the quantized DC coefficient Qxl are quantized AC coefficients obtained by quantifying the AC components between the frequency components. A quantization coefficient Qal placed in the upper left hand corner of the adjacent block Ba is a quantized DC coefficient obtained by quantifying a DC component between the plural frequency components of this adjacent block, and other quantization coefficients in the adjacent block Ba that the quantized DC coefficient Qal are quantified AC coefficients obtained by quantifying the AC components between the frequency components. A quantization coefficient Qbl placed in the upper left hand corner of the adjacent block Bb is a quantized DC coefficient obtained by quantifying a DC component between the plural frequency components of this adjacent block, and other quantization coefficients in the reference block Bb that the quantized DC coefficient Qbl are quantized AC coefficients obtained by quantifying the AC components between the frequency components. A quantization coefficient Qcl placed in the left hand corner above the adjacent block Be is a quantized DC coefficient obtained by quantifying a DC component between the plural frequency components of this adjacent block.

In conventional intraframe predicted coding, however, the quantization coefficients in the reference block different from those that are contiguous with the upper and left edges of this block are not subject to predicted coding. For this reason, in Figure 3, only the quantized DC coefficient Qxl, a group Qxa of AC coefficients comprising the quantized AC coefficients that are contiguous with the upper edge, and a Qxb group of AC coefficients comprising the coefficients Quantized ACs that are contiguous to the left edge are displayed in the reference block Bx. In the adjacent block Ba, only the quantized DC coefficient Qal and a group Qa of quantization coefficients comprising the quantized AC coefficients that are contiguous with the upper edge are shown. In the adjacent block Bb, only the quantized DC coefficient Qbl and a group Qb of quantization coefficients comprising the quantized AC coefficients which are contiguous with the left edge are shown. In the adjacent block Be, only the quantized DC coefficient Qcl is shown.

The predicted intraframe coding is carried out as follows. Initially, a first difference value between the quantized DC coefficient Qcl of the upper left adjacent block Be and the quantized DC coefficient Qal of the upper adjacent block Ba is compared with a second value of difference between the quantized DC coefficient Qcl of the adjacent block upper left Be and the coefficient of quantized Qbl of the adjacent left block B.

When the first difference value is smaller than the second difference value, the quantization coefficients of the Qb group of quantization coefficients of the adjacent left-left block Bb are selected as prediction coefficients for the quantized DC coefficient Qxl and the quantization coefficients of the quantization coefficient. group Qxb of AC coefficients in the reference block Bx. On the other hand, when the second difference value is smaller than the first difference value, the quantization coefficients of the group Qa of quantization coefficients in the upper adjacent block Ba are selected as prediction coefficients for the quantized DC coefficient Qxl and the quantization coefficients of the group Qxa of coefficients of AC in the reference block Bx.

Usually, the prediction of the quantization coefficients in the reference block Bx is carried out by either the quantized DC coefficient Qxl and the quantization coefficients of the Qxa group of AC coefficients, which are arranged in the horizontal direction or the quantized DC coefficient Qxl and the quantization coefficients of the Qxb group of AC coefficients, which are arranged in the vertical direction.

Therefore, the quantization coefficients of the group Qa (Qb) of quantization coefficients in the adjacent block Ba (Bb), which have been determined as prediction coefficients, are extracted from the quantized DC coefficient Qxl and the quantification coefficients of the group Qxa (Qxb) of coefficients of AC, therefore the quantized difference coefficients of the reference block are obtained.

Subsequently, the quantized difference coefficient corresponding to the quantized DC coefficient Qxl, the quantized difference coefficients corresponding to the quantization coefficients of the group Qxa (Qxb) of AC coefficients, and the other quantization coefficients which have not been subjected to predictions are decoded using code words comprising a predetermined number of variable length codes and a predetermined number of fixed length codes having the same bit length (e.g., 8 bits).

In the coding process, if a part of the quantized difference coefficients described above do not have variable length codes corresponding to their values, these difference coefficients are decoded using fixed length codes each having a _ 8-bit code length , which in a specific way are shown in figure 13.

Similar to the quantization coefficients, the quantized difference coefficients are transformed into fixed-length codes corresponding to the values that range from "-127" to "127".

Incidentally, in the intra-frame predicted coding described above, it is necessary to reversibly encode the quantized difference coefficients. That is, the quantized difference coefficients which have been decoded must have the same values as those previously encoded. To satisfy this, a sufficient number of bits of the code length of the fixed length codes expressing the quantized difference coefficients is required.

As described above, in the predicted intraframe encoding, the values of the quantized difference coefficients are in the range from "-127" to "127", and the quantized difference coefficients are expressed using fixed length codes each having a 8-bit code length. In this case, however, since the quantization coefficients have their values in the range from "-127" to "127", the quantized difference coefficients have their values in the range from "-254" to "254" and , therefore, they originate a case where a part of the quantized difference coefficients can not be encoded by the 8-bit fixed-length codes shown in Figure 13.

This case can be avoided by using fixed length codes each having a code length of 9 bits. In this case, however, the values of the quantized difference coefficients ranging from "-127" to "127" must be encoded by the fixed-length codes of 9 bits, resulting in reduced coding efficiency by the quantified difference coefficients.

To avoid the case described above, there is another method in which, when the values of a part of the quantized difference coefficients exceeds the range from "-127" to "127", the values of these coefficients are set at the lower limit value "-127" or the upper limit value "127" in the range. In this case, however, the quantized difference coefficients can not be reversibly encoded.

BRIEF DESCRIPTION OF THE INVENTION An object of the present invention is to provide an image processing method and apparatus capable of carrying out the predicted intraframe coding to reversibly transform transformation coefficients, without reducing the coding efficiency, when the transformation coefficients (quantization coefficients) ) obtained compressively by transforming an image signal are encoded using fixed length codes having the same bit length.

Another objective of the present invention is to provide an image processing method and apparatus capable of carrying out the predicted intraframe decoding of transformation coefficients which have been encoded by the reversed intraframe predicted coding described above with no reduction in coding efficiency. .

Yet another object of the present invention is to provide a data storage medium containing a program that performs, by a computer, the predicted intraframe reversible coding described above with no reduction in coding efficiency, or the predicted intraframe predicted decoding. to the predicted intraframe coding.

Other objects and advantages of the invention will become apparent from the detailed description that follows. The detailed description and specific embodiments described are provided for illustration only since various additions and modifications within the scope of the invention will be a. to those of specialty in the technique of detailed description.

According to the first aspect of the present invention, an image coding method for transforming an image signal into transformation coefficients is provided by compressing data of the image signal for each processing unit, and encoding the transformation coefficients in a coding process using code words that they comprise a predetermined number of codes, and the coding process comprises the steps of: identifying whether or not the values of the reference difference coefficients, which are differences between the reference transformation coefficients that are processed and the transformation coefficients as different prediction coefficients of the reference transformation coefficients, they correspond to any of the codes that make up the code words; coding the reference difference coefficients when the values of the reference difference coefficients correspond to any of the codes in the code words; and coding the reference transformation coefficients when at least one of the values of the reference difference coefficients correspond to none of the codes in the code words. Therefore, when the transformation coefficients obtained in a compressive manner by transforming an image signal are encoded using predetermined code words, the predicted intraframe encoding of the transformation coefficients can be carried out reversibly without reducing the coding efficiency.

According to a second aspect of the present invention, in the image processing method of the first aspect, the code words are composed of a plurality of variable length codes, and a plurality of fixed length codes having the same length of bits; In the process of coding each of the reference transformation coefficients, when there is a variable length code corresponding to the value of the reference transformation coefficient, the reference transformation coefficient is coded using the variable length code, and when not there is a variable length code corresponding to the value of the reference transformation coefficient, the reference transformation coefficient is coded using a fixed length code; and in the coding process each of the reference difference coefficients, when there is a variable length code corresponding to the value of the reference difference coefficient, the reference difference coefficient is coded using the variable length code, and when there is no variable length code corresponding to the value of the reference difference coefficient, the reference difference coefficient is coded using a fixed length code. Therefore, efficient coding is performed.

According to a third aspect of the present invention, there is provided an image coding method for dividing an image signal into image signals respectively corresponding to a plurality of unit areas into which an image space for visual presentation is divided. , transforming an image signal corresponding to a reference unit area that is coded into data compression transformation coefficients, and subjecting the transformation coefficients in the reference unit area to a coding process using code words comprising a predetermined number of codes, and the coding process comprises the steps of: identifying whether or not the values of the reference difference coefficients of the reference unit area, which are differences between the conversion coefficients in the area of reference unit and the transformation coefficients as coefficients d prediction in another unit area adjacent to the reference unit area, correspond to any of the codes that make up the code words; coding the difference coefficients in the reference unit area when the values of the difference coefficients in the reference unit area correspond to any of the codes in the code words; and encoding the transformation coefficients in the reference unit area when at least one of the values of the difference coefficients in the reference unit area corresponds to none of the codes in the code words. Therefore, when the transformation coefficients obtained in a compressive manner by transforming an image signal are encoded using predetermined code words, the predicted intraframe encoding of the transformation coefficients can be carried out reversibly without reducing the coding efficiency.

According to a fourth aspect of the present invention, there is provided an image decoding method for decoding a coded image signal based on an identification signal from the outside, and? Decoding comprises the steps of: deciding whether an image signal encoding that is decoded corresponds to reference transformation coefficients obtained by compressing data from an image signal or reference difference coefficients which are differences between the reference transformation coefficients and the transformation coefficients as different prediction coefficients of the reference transformation coefficients, based on the identification signal; decoding the encoded image signal to generate a decoded image signal that is processed as reproduced coefficients when the encoded image signal that is decoded corresponds to the reference transform coefficients; generating reproduced coefficients by adding prediction coefficients obtained from a decoded image signal different from the decoded image signal that is processed, to the decoded image signal that is processed, when the decoded image signal that is decoded corresponds to the coefficients of reference difference; and subjecting the reproduced coefficients to data expansion to generate a reproduced image signal. Therefore, it is possible to perform a decoding method adapted to a coding method in which either the coding of the reference transformation coefficients or the coding of the difference coefficients between the reference transformation coefficients and the coefficients of Prediction is selected according to whether the coefficients that are encoded have corresponding codes or not.

According to a fifth aspect of the present invention, there is provided an image coding method for transforming an image signal into transformation coefficients by compressing the image signal data for each processing unit, and subjecting the transformed coefficients to a coding process using code words comprising a plurality of fixed length codes, and the coding process comprises the steps of: identifying whether or not the values of the reference difference coefficients, which are differences between the coefficients of reference transformation that are processed and the transformation coefficients as different prediction coefficients from the reference transformation coefficients, correspond to any of the codes that make up the code words; encoding the reference difference coefficients that produce an encoded image signal when the values of the coefficients of different reference correspond to any of the codes in the code words; and transforming the reference difference coefficients to binary codes corresponding to their values and having a code length greater than that of the fixed length codes and then producing code sequences, each having a bit width equivalent to the length of code of the fixed-length codes and corresponding to the lower part of each binary code, as an encoded image signal, when at least one of the values of the reference difference coefficients corresponds to none of the codes in the words of code. Therefore, it is not necessary to limit the values of the difference coefficients within a range from - (2IN "1) -1) to (2 (N" 1) -1), therefore the difference coefficients can be encode in a reversible manner without reducing the coding efficiency.

According to a sixth aspect of the present invention, there is provided an image decoding method for decoding a coded difference signal reproducing transformation coefficients, the coded difference signal is obtained through the following steps: submitting a signal of image to compression of data for each processing unit to generate transformation coefficients; and encoding difference coefficients, which are differences between the reference transformation coefficients that are encoded and other transformation coefficients which have been previously generated to the reference transformation coefficients, using first code words comprising a plurality of codes fixed length or secondly code words comprising binary codes each having a length of code longer than that of fixed length codes. The image decoding method comprises the steps of: desodifying a coded difference signal that is processed to generate a decoded difference signal that is processed; adding transformation coefficients already reproduced to the decoded difference signal that is processed, as prediction coefficients, to generate reproduced coefficients that are processed; compare the value of each of the reproduced coefficients that are processed with a positive maximum value and a negative minimum value between the binary reference numbers each having the number of digits one bit less than the code length of the length codes fixed; add a maximum absolute value of an arithmetic binary number that has the number of digits equal to the code length of the fixed length codes, to the value of the reproduced coefficient, and then produce the sum, when the value of the reproduced coefficient is less than the minimum negative value of the reference binary number; subtract the maximum absolute value of the arithmetic binary number of the reproduced coefficient and produce the difference when the value of the reproduced coefficient is greater than the maximum positive value of the reference binary number; and producing the reproduced coefficient like this is when the value of the reproduced coefficient is greater than the minimum negative value and less than the positive maximum value. Therefore, it is possible to perform an intraframe predicted decoding process adapted to an intraframe predicted coding process using binary codes and not having reduction in coding efficiency.

According to a seventh aspect of the present invention, there is provided an image coding apparatus comprising a data compression unit for transforming an image signal into transformation coefficients by data compression for each processing unit, and a unit encoding to encode the transformation coefficients using code words comprising a predetermined number of codes. The data compression unit comprises means for generating difference coefficients to generate reference difference coefficients which are differences between the reference transformation coefficients that are processed and the transformation coefficients as different prediction coefficients of the transformation coefficients reference. The coding unit comprises identification means for identifying whether the values of the reference difference coefficients correspond to any of the codes in the code words, and coding means for coding the reference difference coefficients when the values of the reference difference coefficients correspond to any of the codes in the code words, while coding the reference transformation values when at least one of the values of the reference difference coefficients correspond to any of the codes in the words of code, based on the result of the means of identification. Therefore, when the transformation coefficients obtained in a compressive manner by transforming an image signal are encoded using predetermined code words, the predicted intraframe encoding of the transformation coefficients can be carried out reversibly without reducing the coding efficiency.

According to an eighth aspect of the present invention, an image decoding apparatus is provided for decoding a coded image signal based on an identification signal from the outside, and the decoding apparatus comprises: decision means for deciding whether a signal encoded image that is decoded corresponds to reference transformation coefficients obtained by compression of data from an image signal, or reference difference coefficients which are differences between the reference transformation coefficients and the transformation coefficients as prediction coefficients different from the reference transformation coefficients, based on the identification signal; first, means for generating coefficients to generate a decoded image signal that is processed as reproduced coefficients first by decoding the coded image signal that is processed; second, means for generating coefficients to generate secondly reproduced coefficients by adding prediction coefficients obtained from a decoded image signal different from the decoded image signal being processed, to the decoded image signal that is processed; switching means for selecting the output of the first coefficient generation means when the coded image signal that is processed corresponds to the reference transformation coefficients, while selecting the output of the second coefficient generation means when the signal of The coded image that is processed corresponds to the reference difference coefficients, based on the result of the decision means; and data expansion means for the data expansion of the output of the reproduced coefficients of the switching means to generate a reproduced image signal. Therefore, it is possible to perform a decoding method adapted to a coding method in which either the coding of the reference transformation coefficients or the coding of the difference coefficients between the reference transformation coefficients and the coefficients of Predictions are selected according to whether the coefficients that are encoded have corresponding codes or not.

According to a ninth aspect of the present invention, there is provided an image coding apparatus comprising a data compression unit for transforming an image signal into transformation coefficients by data compression for each processing unit, and a unit encoding to encode the transformation coefficients using code words comprising a predetermined number of codes. The data compression unit comprises means for generating difference coefficients to generate reference difference coefficients which are differences between the reference transformation coefficients that are processed and the transformation coefficients as different prediction coefficients of the transformation coefficients reference. The coding unit comprises decision means for deciding whether the values of the reference difference coefficients correspond to any of the codes in the code words; encoding means for encoding the reference difference coefficients using the code words and producing a coded image signal; binary code converter for converting the reference difference coefficients to binary codes corresponding to their values and having a longer code length than the fixed length codes; code output means for producing codes, each having an amplitude of bits equivalent to the code length of the fixed-length codes and corresponding to the lower part of each binary code, as an encoded picture signal; and switching means for connecting the output of the difference coefficient generation means to one of the input of the coding means and the input of the binary code converter, based on the result of the decision means, such that the coefficients reference difference are supplied to the coding means when the values of the reference difference coefficients correspond to any of the codes in the code words, while they are supplied to the binary code converter when at least one of the values of The reference difference coefficients correspond to none of the codes in the code words. Therefore, it is not necessary to limit the values of the difference coefficients within a range from (N-1) 1) up to (2"-1), therefore the difference coefficients can be encoded reversibly without reducing the coding efficiency.

According to a tenth aspect of the present invention, an image decoding apparatus is provided for decoding a coded difference signal reproducing transformation coefficients, the coded difference signal is obtained through the following steps: subjecting a signal of image to compression of data for each processing unit to generate transformation coefficients; and encoding the difference coefficients, which are differences between reference transformation coefficients that are encoded and other transformation coefficients which have been previously generated to the reference transformation coefficients, using first code words comprising a plurality of fixed length codes or secondly code words comprising binary codes each having a longer code length than that of fixed length codes. The image decoding apparatus comprises: a data analyzer for generating a decoded difference signal that is processed by data analysis of a coded difference signal that is processed; a reproduction unit of coefficients to generate reproduced coefficients that are processed by adding transformation coefficients already reproduced as prediction coefficients to the decoded difference signal that is processed; decision means to decide whether or not the value of each of the reproduced coefficients that are processed is greater or less than a positive maximum value and a negative minimum value between the reference binary numbers that have the number of digits a bit smaller than the code length of the fixed-length codes; and reproduced coefficient processing means including an adder to add a maximum absolute value of an arithmetic binary number having the number of digits equal to the code length of the fixed length codes to the value of the reproduced coefficient being processed, and a subtractor to subtract the maximum absolute value of the arithmetic binary number from the value of the reproduced coefficient that is processed. The reproduced coefficient processing means are carried out as follows based on the result of the decision in the decision means: when the value of the reproduced coefficient is less than the minimum negative value of the reference binary number, it produces the sum obtained in the adder; when the value of the reproduced coefficient is greater than the positive maximum value of the reference binary number, it produces the difference obtained in the subtractor; and when the value of the reproduced coefficient is greater than the minimum negative value and less than the positive maximum value, it produces the reproduced coefficient as it is. Therefore, it is possible to perform an intraframe predicted decoding process adapted to an intraframe predicted coding process using binary codes and that has no reduction in coding efficiency.

According to an eleventh aspect of the present invention, a data storage medium is provided that contains a program that processes an image signal, and the program is a program which causes a computer to process an image signal by a method of image processing according to any of the first to sixth aspects.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a block diagram illustrating an image coding apparatus according to a first embodiment of the present invention.

Figure 2 is a circulation diagram for explaining an intraframe predicted coding process by the image coding apparatus according to the first embodiment.

Figure 3 is a schematic diagram for explaining the prediction of transformation coefficients in the intraframe predicted coding process by the image coding apparatus according to the first embodiment.

Figure 4 is a block diagram illustrating an image decoding apparatus according to a second embodiment of the present invention.

Figure 5 is a circulation diagram for explaining an intraframe predicted decoding process by the image decoding apparatus according to the second embodiment.

Figure 6 is a block diagram illustrating an image coding apparatus according to the third embodiment of the present invention.

Figure 7 is a circulation diagram for explaining an intraframe predicted coding process by the image coding apparatus according to the third embodiment.

Figures 8 (a) and 8 (b) are schematic diagrams to explain the specific arithmetic processes in the intraframe predicted coding process by the image coding apparatus according to the third embodiment.

Figure 9 is a block diagram illustrating an image decoding apparatus according to the fourth embodiment of the present invention.

Figure 10 is a circulation diagram for explaining an intraframe predicted decoding process by the image decoding apparatus according to the fourth embodiment.

Figures 11 (a) and 11 (b) are schematic diagrams to explain the specific arithmetic processes in the intraframe predicted decoding process by the image decoding apparatus according to the fourth embodiment.

Figures 12 (a) -12 (c) are diagrams for explaining a data storage medium which contains a program that carries out the processing of the image signal according to any of the first to fourth modes using a computer system.

Figure 13 is a diagram illustrating the fixed-length codes used to encode and decode according to the prior art and the aforementioned embodiments of the invention, and the quantization coefficients (transformation coefficients) corresponding to the particular codes.

DESCRIPTION DETAT. ADA OF THE PREFERRED MODALITIES Modality 1.

Figure 1 is a block diagram for explaining an image coding apparatus 100 according to a first embodiment of the present invention.

The image coding apparatus 100 receives an input image signal Sg through an input terminal la, encodes the image signal Sg by a coding method that includes the predicted intracuadr coding, and produces an encoded image signal. through the exit terminal Ib.

In this first embodiment, the input image signal Sg is a digital image signal comprising a luminance signal and a chrominance signal. Also, the? image encoding device 100 processes, as the input image signal Sg, an image signal corresponding to an image having an arbitrary shape in an object area, as well as an image signal corresponding to an image in an image space that It has a rectangular shape (a square).

The loo image coding apparatus includes a blocking unit 101, a DCT unit 102, and a quantizer 103. The blocking unit 101 divides the input image signal Sg into a plurality of respectively image signals corresponding to a plurality of image signals. of blocks (processing units) into which an image (frame) or an object area is divided, and produces Bg image signals (hereinafter, also referred to as blocked image signals). The DCT unit 102 transforms the image signal Bg corresponding to a reference block which is presently processed, in frequency components Fg by discrete cosine transformation (DCT). The quantizer 103 quantizes the frequency components Fg and produces quantization coefficients (transformation coefficients) Qgl.

The image coding apparatus 100 further includes a difference coefficient generator 104, a switch 105, and an encoder 106. The difference coefficient generator 104 employs the quantization coefficients of a block which is placed in the vicinity of the block and has already been coded as prediction coefficients for the reference block, generates difference values between the Qgl quantization coefficients of the reference block and the prediction coefficients, and produces the difference values as difference coefficients (coefficients of quantized difference) Qg2 of the reference block. The switch 105 selects either the Qgl output of the quantizer 103 or the Qg2 output of the difference coefficient generator 104 according to a control signal of the switch Csl, and produces this as a selected output Qg. The encoder 106 subjects the output Qg of the switch 105 to an encoding process using code words comprising a predetermined number of variable length codes and a predetermined number of fixed length codes having the same bit length (8 bits) , according to a coding control signal Cco, and produces an encoded image signal Eg.

The image coding apparatus 100 further includes a control unit 110 which generates the control signals Csl and Cco. The coding control unit 110 comprises an identification circuit 111 and an encoding control circuit 112. The identification circuit 111 identifies whether or not the value of the output Qg2 of the difference coefficient generator 104 corresponds to any of the codes plurals that compose the code words in the encoder 106, and produces the control signal of the switch Csl to the switch 105. The coding control circuit 112 controls the encoder 106, based on the outputs Qgl and Qg2 to carry out either variable length coding or fixed length coding for particular outputs.

The switch 105 selects a signal that is input to the encoder 106 in the following manner, based on the control signal Csl. When there is a code corresponding to the value of the output Qg2, the switch 105 supplies the output Qg2 of the difference coefficient generator 104 to the encoder 106. When there is no code corresponding to the value of the output Qg2, the switch 105 supplies the output Qgl. from quantizer 105 to encoder 106.

A description of the operation of the image coding apparatus 100 is given.

Figure 2 is a circulation diagram for explaining the intraframe predicted coding by the image coding apparatus 100.

When the input image signal Sg which is processed is input to the image coding apparatus 100 (step S10), the image signal Sg is divided into a plurality of image signals corresponding to a plurality of unit areas (blocks) in which an image (a frame) or an object area is divided into the blocking unit 101 (step Sil). In this first modality each block is a visual presentation area of image that comprises 8X8 points.

Then, the image signals Bg corresponding to the particular blocks are compressed block by block and transformed into quantization coefficients (transformation coefficients). More specifically, the image signal Bg corresponding to the reference block that is compressed is transformed to frequency components Fg by DCT in the DCT unit 102, and the frequency components Fg are transformed to quantization coefficients Qgl by quantization in the quantizer 103 (step S12). The values of the quantization coefficients are integers within a range from "-127" to "127".

In the difference coefficient generator 104, the prediction coefficients for the reference block are decided by referring to the quantization coefficients of a block which is placed in the vicinity of the reference block and has already been compressed, and the coefficients of prediction are subtracted from the quantization coefficients of the reference block to obtain the difference coefficients Qg2 of the reference block (step S13). It is specific that, 8 coefficients of difference Qg2 are produced for the reference block and, in addition to the difference coefficients there are fifty-six quantization coefficients in the reference block. The method of deciding the prediction coefficients for the reference block by referring to the quantization coefficients of a block in the vicinity of the reference block is identical to the method already described with respect to the conventional intraframe predicted coding using Figure 3.

Returning to step S14, in the identification unit 111 it is identified whether or not any of the plural difference coefficients in the reference block exceeds the range from "-127" to "127".

When the result in step S14 is that at least one of the difference coefficients is less than "-127" or greater than "127", "0" is set as an identifier of this block in the identification circuit 111, and the switch 105 is controlled by the control signal of the switch Csl of the identification circuit 111 to select the output Qgl of the quantizer 103. Receiving the output Qgl of the quantizer 103 through the switch 105, the encoder 106 encodes the output (transformation coefficients) ) Qgl (step S16).

When the result in step S14 is that none of the difference coefficients in the reference block exceeds the range from "-127" to "127", "i" is set as an identifier of the reference block in the identification circuit 11, and the switch 105 is controlled by the control signal of the switch Csl of the identification circuit III to select the output Qg2 of the difference coefficient generator 104. Receiving the output Qg2 of the difference coefficient generator 104 through the switch 105 , the encoder 106 codes the output (difference coefficients) Qg2 (step S15).

In this coding process, similar to the coding process based on MPEG1, the quantization values (quantization coefficients or difference coefficients) are fundamentally encoded by the variable length coding and, when there are no variable length codes corresponding to the quantization values, the quantization values are encoded using the fixed length codes shown in Figure 13. Therefore, in the coding control unit 112, it is decided whether there are variable length codes corresponding to the quantization values, based on the output Qgl of the quantizer 103 and the output Qg2 of the difference coefficient generator 104. Furthermore, in the encoder 106, according to the coding control signal Cco based on the result of the decision, it is carried out either variable length coding or fixed length coding.

Subsequently, it is decided whether or not the reference block is the last block between the blocks that make up a frame or an object area (step S17). When the reference block is the last block, the predicted interframe encoding is completed. When the reference block is not the last block, the processes in raisins S12-S17 are repeated.

The quantization coefficients or the difference coefficients thus encoded are produced from the encoder 106 in conjunction with the identifier and are then recorded in a recording medium or transmitted to an image decoding apparatus.

As described above, the first embodiment of the present invention relates to a method for transforming an image signal of input-transformation coefficients by data compression and then encoding the transformation coefficients using N-bit codes (N = integer ), that is, fixed length codes each having a code length of N bits. In this method, the transformation coefficients of a block placed in the vicinity of a reference block that is processed are used as prediction coefficients for the reference block, and the difference coefficients which are values of difference between the coefficients of transformation of the reference block and the prediction coefficients are encoded using the N-bit codes. In the coding process, when the values of the difference coefficients are within the range of the codes that can be expressed by N bits, the difference coefficients of the reference block are coded. On the other hand, when the "minus one of the difference coefficients exceeds the range, the transformation coefficients of the reference block are coded." Therefore, the predicted intra-frame coding adapted to the transformation coefficients obtained by data compression of the Image signal can be carried out reversibly without reducing coding efficiency.

It is specific that, when at least one of the values of the difference coefficients exceeds the range that can be expressed by 8-bit codes, the transformation coefficients are coded in place of the difference coefficients. In this case, the difference coefficients can always be reversibly coded because it is not necessary to limit the values of the difference coefficients in a range from - (2 (N_1) -1) to (2 (N "1) -1) (in this mode, from -127 to 127).

Although it is stated as a premise that the image coding apparatus 100 quantizes the corresponding DCT coefficients between the adjacent blocks with the same quantization amplitude, the apparatus 100 can quantify the corresponding DCT coefficients between the adjacent blocks with different quantization amplitudes. In this case, however, it is necessary to generate difference coefficients based on the coefficients obtained in an inverse manner by quantifying the transformation coefficients obtained by quantifying the DCT coefficients, and the difference coefficient generator 104 must be provided with a circuit structure for This process.

Furthermore, although in this first DCT mode it is used for data compression of an image signal corresponding to each block, the small wave transformation can be used.

Mode 2 Figure 4 is a block dia for explaining an image decoding apparatus 200 according to a second embodiment of the present invention.

The image decoding apparatus 200 of this second embodiment is adapted to the image coding apparatus 100 of the first embodiment. That is, an encoded picture signal Eg produced from the image coding apparatus 100 is input to the decoding apparatus 200 through the input terminal 200a, and the decoding apparatus 200 decodes the image signal encoded Eg by a method of decoding that includes the predicted intra-frame decoding, and produces a reproduced image signal Rsg through the output terminal 2b.

The image decoding apparatus 200 includes a data analyzer 201 and a quantization coefficient restoration unit 202. The data analyzer 201 decodes the co-encoded image signal by analyzing the code sequence, therefore it restores the quantization coefficients RQgl for each block that is a decoding unit (hereinafter, also mentioned as restored coefficients RQgl). When the restored coefficients correspond to the difference coefficients of the first embodiment, the restoration unit of quantization coefficients 202 uses the restored coefficients of a block which is placed in the vicinity of a reference block that is decoded and has already been decoded, as prediction coefficients for the reference block, and obtains the sum of the restored coefficients RQgl of the reference block and the prediction coefficients, as quantification coefficients (restored coefficients) RQg2 of the reference block.

In addition, the image decoding apparatus 200 includes a switch 203, an inverse quantizer (IQ) 205, a reverse DCT unit 206, and a reverse blocking unit 207. The switch 203 selects either the RQgl output of the data analyzer. 201 or the output RQg2 of the restoration unit of quantization coefficients 202 according to a control signal of the switch Cs2, and output as restored coefficients RQg. The inverse quantizer 205 subjects the output RQg from the switch 203 to inverse quantization to generate restored DCT coefficients. The reverse DCT unit 206 subjects the restored DCT coefficients RFg to inverse DCT and produces a restored image signal RBg. The reverse blocking unit 207 integrates the restored image signals RBg of the particular blocks to generate a reproduced image signal RSg having a scanning line structure.

A description of the operation of the image decoding apparatus 200 is given.

Figure 5 is a circulation diagram for explaining an intraframe predicted decoding process by the image decoding apparatus 200.

In the image decoding apparatus 200, when the decoding is started and an encoded picture signal Eg corresponding to each of the plural blocks in which a frame or an object area is divided, the apparatus is input together with a control signal (an identifier in the first mode) (step S21), the data analyzer 210 transforms the code sequence of the coded picture signal Eg to restored coefficients RQgl by data analysis (step S22). The data analyzer 201 performs variable length decoding to the code sequence when the code sequence comprises variable length codes, and performs fixed length decoding to the code sequence when the code sequence comprises codes of fixed length.

In step S23, the decision circuit 204 decides the identifier of the reference block. When the identifier is not "1", since the restored RQgl coefficients are restored quantization coefficients, the switch 203 is controlled by the control signal Cs2 of the decision circuit 204 such that the restored coefficients RQgl of the data analyzer 201 produce as an output of the switch RQg to the inverse quantizer 205 (step S24).

On the other hand, when the decision in step S23 is that the identifier of the reference block is "1", the restored RQgl coefficients of the reference block are restored difference coefficients. Therefore, in the restoration unit of quantization coefficients 202, the restored quantization coefficients of a block which is placed in the vicinity of the reference block and which has already been decoded are used as prediction coefficients for the block reference, and the sum of the restored coefficients (restored difference coefficients) RQgl of the reference block and the prediction coefficients are obtained as restored coefficients (restored quantization coefficients) RQg2 of the reference block (step S25). In this case, the switch 203 is controlled by the control signal Cs2 of the decision circuit 204 such that the restored coefficients RQg2 of the restoration unit of quantization coefficients are sent to the inverse quantizer 205.

The restored coefficients (restored quantization coefficients) thus obtained are transformed to restored DCT coefficients RFg by inverse quantization in the inverse quantizer 205, and the restored DCT coefficients RFg are transformed to a restored image signal RBg corresponding to each block by DCT reverse in the reverse DCT unit 206.

Then, in the image decoding apparatus 200, it is decided whether or not the reference block is the last block between the blocks that make up a frame (step S27). When the reference block is the last block, the intraframe predicted decoding is completed, and the reverse blocking unit 207 integrates the restored image signals RBg corresponding to the particular blocks and produces a reproduced image signal RSg having a scanning line structure and corresponding to a frame or an object area (step S28). When the reference block is not the last block, the processes of steps S21-S27 are repeated.

As described above, according to the second embodiment of the invention, either the restored difference coefficients or the restored quantization coefficients, which have been restored by the data analyzer 201, are selected according to the identifier, which it has been set for each block _at the end of the coding, and the selected coefficients are subjected to inverse transformation. Therefore, it is possible to perform an intraframe predicted decoding method adapted to an intraframe predicted coding method in which only the difference coefficients having corresponding 8-bit codes are encoded and, instead of the difference coefficients that do not have corresponding 8-bit codes, the transformation coefficients are coded.

Modality 3.

Figure 6 is a block diagram for explaining an image coding apparatus 300 according to the third embodiment of the present invention.

The image coding apparatus 300 encodes an input image signal Sg applied to an input terminal 3a by a coding method that includes intraframe prediction, and produces an encoded image signal Eg through an output terminal 3b .

In this third embodiment, the input image signal Sg is a digital image signal comprising a luminance signal and a chrominance signal. In addition, the image coding apparatus 300 processes, as an input image signal Sg, an image signal corresponding to an image having an arbitrary shape in an object area, as well as an image signal corresponding to an image in an image. Image space that has a rectangular shape (a box).

The image coding apparatus 300 includes a blocking unit 301, a DCT unit 302, and a quantizer 303. The blocking unit 301 divides the input image signal Sg into a plurality of image signals respectively corresponding to a plurality of blocks (processing units) in which an image (frame) or an object area is divided, and produces Bg image signals (hereinafter, also referred to as blocked image signals). The DCT unit 302 transforms the image signal Bg of a reference block which is presently processed into frequency components Fg by discrete cosine transformation (DCT). The quantizer 303 quantizes the frequency components FG and produces quantization coefficients (transformation coefficients) Qg.

The image coding apparatus 300 further includes a difference coefficient generator 304, a coding unit 310, a binary code converter 307, and an output unit for the code sequence 308. The difference coefficient generator 304 employs the quantization coefficients of a block which is placed in the vicinity of the reference block and which has already been coded, as prediction coefficients for the reference block, generates different values between the quantization coefficients Qg of the reference block and the prediction coefficients, and produces the different values as coefficients of difference Qgd of the reference block. The coding unit 310 codes the difference coefficients Qgd using code words comprising a predetermined number of variable length codes and a predetermined number of fixed length N-bit codes, and produces a coded signal Cg. The binary code converter 307 binary encodes the difference coefficients Qgd until they are converted into coded coefficients BCg. The output unit for the code sequence 308 produces a sequence of. N-bit code Blg comprising lower N-bits of each of the coded coefficients BCg. The coding unit 310 comprises a variable length encoder 311 and a fixed length encoder 312. The variable length encoder 311 performs variable length coding for the difference coefficients having variable length codes corresponding to their values, and produce them as a Cg coded signal. The encoder 311 does not encode the difference coefficients that do not have variable length codes corresponding to their values, and pass these coefficients completely. The fixed-length encoder 312 performs the fixed-length encoding of the difference coefficients that pass through the variable-length encoder 311 using fixed-length N-bit codes, and produce the coefficients as a coded signal Cg.

The image coding apparatus 300 further includes a switch 305 and a decision unit 306. The switch 305 supplies the output Qgd of the difference coefficient generator 304 to either the coding unit 310 or the binary code converter 307, based in the control signal Cs2. The decision unit 306 decides whether or not the coding of the difference coefficients using the code words in the coding unit 310 is possible or not, based on the output of the difference coefficient generator 304, and the control signal is output Cs2 corresponding to the result of the decision towards the switch 305.

A description of the operation of an image coding apparatus 300 is given. FIG. 7 is a circulation diagram for explaining an intraframe predicted coding process by the image coding apparatus 300.

In this third embodiment, the blocking of an image signal in step S31, the generation of the transformation coefficients in step S32, and the generation of the difference coefficients in step S33 are identical to those already described with respect to steps Sil, S12, and S13 in the intraframe predicted coding process of the first mode, respectively.

In step S34, it is decided by the decision unit 306 whether or not all the difference coefficients of the reference block have corresponding codes in the code words of the coding unit 310.

When it is decided in step S34 that all the difference coefficients of the reference block have corresponding codes, the switch-305 is controlled by the control signal Cs2 of the decision unit 306 such that the output Qgd of the coefficient generator of difference 304 is supplied to the coding unit 310. Receiving the output Qgd through the switch 305, the coding unit 310 codes the output Qgd (step S35).

It is specific that, when there are variable length codes corresponding to the difference coefficients, the difference coefficients are subjected to variable length coding in the variable length encoder 311 and are produced as a coded signal Cg of the reference block.

When there are no variable length codes corresponding to the difference coefficients, the difference coefficients pass through the variable length encoder 311 and are input to the fixed length encoder 312, where the difference coefficients are-length encoded fixed and then they come out as a coded signal Cg of the reference block.

On the other hand, when it is decided in step S34 that at least one of the difference coefficients of the reference block does not have a corresponding code, the switch 305 is controlled by the control signal Cs2 of the decision unit 306 such that the output Qgd of the coefficient generator-e difference 304 is supplied to a binary code converter 307.

In the binary code converter 307, the difference coefficients of the reference block are converted to coding coefficients BCg by binary coding. In addition, in the output unit for the code sequence 308, a N-bit code sequence BLg comprising N lower bits of each of the coded coefficients BCg is extracted from the coded code BCg (step S37).

Subsequently, it is decided whether or not the reference block is the last block between the blocks that make up a frame or an object area (step S38). When the reference block is the last block, the predicted intraframe encoding is completed. When the reference block is not the last frame, the processes are repeated in steps S32-S38.

Next, the binary coding in the intraframe predicted coding process of this third mode will be described in more detail. Figures 8 (a) and 8 (b) are schematic diagrams to explain the binary coding.

In these figures, the numerical values Al and Bl are values of the transformation coefficients, A2 and B2 are values of the prediction coefficients, and A3 and B3 are values of the difference coefficients. Figure 8 (a) shows an example where A3 is 254 and this is transformed to a code sequence A4 as a binary number "1111 1110". Figure 8 (b) shows an example where B3 is -129 and this is transformed to a code sequence B4 'as a binary number "1 0111 1111".

From the code sequence A4 (B4 '), the A4 data (B4) comprising the lower 8 bits of the code sequence is extracted and output as a BLg code sequence. In other words, the code sequence A4 corresponding to the difference coefficient A3 shown in Figure 8 (a) occurs as is. With respect to the code sequence B4 'corresponding to the difference coefficient B3 shown in FIG. 8 (b), the code sequence B4 comprising the lower 8 bits of the code sequence B' is produced.

As described above, according to the third embodiment of the present invention, the transformation coefficients of a block adjacent to a reference block that is processed are used as prediction coefficients for the reference block, and the values of difference between the transformation coefficients of the reference block and the prediction coefficients are obtained as difference coefficients. When the values of the difference coefficients are within a range of codes which can be expressed by N bits (8 bits), the difference coefficients of the reference block are coded using predetermined code words and then output as code sequences . On the other hand, if at least one of the values of the difference coefficients exceeds the range, the difference coefficients of the reference block are transformed to binary codes, and the lower N bits (8 bits) of each binary code are produced as a code sequence. Therefore, the predicted intra-frame coding of the transformation coefficients obtained in a compressive manner by transforming an image signal can be carried out reversibly without reducing the coding efficiency.

Modality 4.

Figure 9 is a block diagram for explaining an image decoding apparatus 400 according to a fourth embodiment of the present invention. The image decoding apparatus 400 of this fourth embodiment is adapted to an image coding apparatus 300 of the third embodiment. That is, an encoded image signal Eg (an encoded signal Cg or N-bit code sequences Blg) produced from the image coding apparatus 300 is input to the decoding apparatus 400 through an input terminal 4a. The decoding apparatus 400 decodes the coded image signal Eg and produces a reproduced image signal Rsg through an output terminal 4b.

The image decoding apparatus 400 includes a data analyzer 401 and a coefficient reproduction unit 402. The data analyzer 401 decodes the coded image signal Eg by analyzing the code sequence, thereby restores the difference coefficients REg for each block that is a decoding unit (below, also referred to as restored coefficients REg). The coefficient reproduction unit 420 uses the reproduced coefficients (restored quantization coefficients) of a block which is placed in the vicinity of a reference block that is processed and which has already been reproduced, as prediction coefficients for the block of reference, and obtains the sum of the restored coefficients REg of the reference block and the prediction coefficients as reproduced coefficients RQgl of the reference block. When the input encoded image signal Eg is a variable length encoded signal Cg, the data analyzer 401 decodes the coded image signal Eg using the code words (code table) in the coding unit 310 of the third mode to obtain the restored difference coefficients REg (decimal values). On the other hand, when the coded picture signal Eg is a fixed-length coded signal Cg or N-bit code sequences BLg, the data analyzer 401 decodes the coded picture signal Eg to transform the signal Eg to the coefficients of difference restored REg (decimal values).

In addition, the image decoding apparatus 400 includes a reproduced coefficient reprocessing unit 410 and a decision unit 403. The reproduced coefficient processing unit 410 submits the output RQgl of the coefficient reproduction unit 402 to arithmetic processing, by the same 'generates reproduced and processed coefficients RQg2. The decision unit 403 controls the reproduced coefficient processing unit 410 according to a control signal Cs3 based on the output RQgl of the coefficient reproduction unit 402.

The reproduced coefficient processing unit 410 comprises an adder 412 to add 256 to the reproduced coefficients, a subtractor 413 to subtract 256 from the reproduced coefficients, and a switch 411 controlled by the control signal Cs3. The switch 411 has an input terminal 411a connected to the output of the coefficient reproduction unit 402, an output terminal 411b connected to the input of the adder 412, an output terminal 411c connected to the input of the subtractor 413, and a output terminal 411d to pass the output RQgl of the coefficient reproduction unit 402 as it is. The switch 411 connects the input terminal 411a to one of the three output terminals 411b, 411c, 411d, according to the control signal Cs3. The decision unit 403 decides whether or not the values of the reproduced RQgl coefficients are within the range of the values corresponding to the fixed length codes each having an 8-bit code length (-127-127), and sends the control signal Cs3 according to the result of the decision towards the switch 411.

The image decoding apparatus 400 further includes an inverse quantizer (IQ) 404, an inverse DCT unit 405, and a reverse blocking unit 406. The inverse quantizer 404 submits the output RQg2 of the reproducing coefficient reproducing unit 410 to inverse quantization to generate restored DCT coefficients RFg. The inverse DCT unit 405 'subjects the restored DCT coefficients RFg to inverse DCT and produces a restored image signal RBg for each block. The reverse blocking unit 406 integrates the restored image signals RBg of the particular blocks to generate a reproduced image signal RSg having a scanning line structure.

A description of the operation of the image decoding apparatus 400 is given.

Fig. 10 is a circulation diagram for explaining an intraframe predicted decoding process by the image decoding apparatus 400.

In the image decoding apparatus 400, when the decoding is started and an encoded image signal Eg (either the encoded signal Cg or the code sequences of N-bits BLg described by the third embodiment), which has been obtained encoding an image signal Sg in the image coding apparatus 300 of the third embodiment, is input to the apparatus 400 (step S41), the data analyzer 401 transforms the encoded image signal Eg, by data analysis, into decimal values in a decoding process according to the encoded signal Cg or the code sequences of N-bits BLg (step S42).

It is specific that, in the data analyzer 401, when the image signal Eg is the variable-length coded signal Cg, the coded image signal Eg is subjected to variable length coding using code words (code table), by the same generates restored difference coefficients REg represented by decimal numbers. On the other hand, when the coded image signal Eg is the fixed-length coded image signal Cg or the N-bit code sequences BLg, the coded image signal Eg is transformed to restored difference coefficients REg as the decimal numbers corresponding.

In step S43, the coefficient reproduction unit 402 uses the reproduced coefficients (restored quantification coefficients) of a block which is placed in the vicinity of a reference block that is processed and which has already been reproduced, as prediction coefficients of the reference block, and obtains the sum of the restored difference coefficients REg of the reference block and the prediction coefficients. Unit 402 produces the sum as reproduced coefficients RQgl of the reference block.

Furthermore, in step S44, it is decided in the decision unit 403 whether or not the values of the reproduced coefficients RQgl are within the range of the values corresponding to the fixed-length codes each having an 8-bit code length. (-127-127), and the reproduced coefficients RQgl are subjected to arithmetic processing according to the result of the decision in the processing unit of reproduced coefficients 410.

It is specific that, when the values of the reproduced RQgl coefficients are within the range mentioned above, the switch 411 is controlled by the control signal Cs3 of the decision unit 403 such that the input terminal 411a is connected to the output terminal 411d, therefore the output (reproduced coefficients) RQgl of the coefficient reproduction unit 402 is produced as reproduced and processed coefficients P g2 (step S48).

If at least one of the values of the reproduced coefficients RQgl is not within the range mentioned above, it is further decided in decision unit 403 whether the value of this reproduced coefficient RQgl is less than -128 or not. (step S45). When the result of the decision in step S45 is that the value of the reproduced coefficient RQgl is less than -128, the switch 411 is controlled by the control signal Cs3 of the decision unit 403 such that the input terminal 411a is connected to the exit terminal 41_.r > , and 256 is added to the value of the reproduced coefficient RQgl in adder 412 and the sum is produced as a reproduced and processed coefficient RQg2 (step S46).

When the result of the decision in step S45 is that the value of the reproduced coefficient RQgl is greater than -128, the switch 411 is controlled by the control signal Cs3 of the decision unit 403 such that the input terminal 411a is connected to the output terminal 411c, and 256 is subtracted from the value of the reproduced coefficient RQgl and the difference is produced as a reproduced and processed coefficient RQg2 (step S47).

Figures 11 (a) and 11 (b) are schematic diagrams for explaining the decoding processes for N-bit code sequences in intraframe predicted decoding by the image decoding apparatus 400.

Figures 11 (a) and 11 (b) show the decoding processes using specific values, and correspond to Figures 8 (a) and 8 (b), respectively. In Figures 11 (a) and 11 (b), the numerical values C2, D2, C3 (= C2), and D3 (= D2) indicate the values of the restored difference coefficients (reproduced coefficients) transformed to decimal numbers, C4 and D4 indicate the values of the prediction coefficients, and C5 and D5 indicate the values produced from the adder 412.

In the process shown in Figure 11 (a), a sequence of code C1 introduced as a code sequence of N-bits BLg is transformed to a decimal number, and -2 is obtained as the value C2 of the reproduced coefficient RQgl. In addition, the value C4 of the prediction coefficient, -127, is added to -2 which is the value C3 (= C2) of the reproduced coefficient, and -129 is obtained as the value C5 of the output of the adder 412. Since this value -129 is less than -128, 256 are added to this value in step S46, and 127 is generated as the C6 value of the reproduced and processed coefficient RQg2. This value is equal to the Al value of the transformation coefficient shown in Figure 8 (a).

Returning to FIG. 11 (b), a DI code sequence introduced as an N-bit BLg code sequence is transformed to a decimal number, and 127 is obtained as the value D2 of the reproduced coefficient RQgl. Then, the value C4 of the prediction coefficient, 2, is added to 127 as the value D3 (= D2), and 129 is obtained as the value D5 of the output of the adder 412. Since 129 is greater than 128, it is subtracted 256 of this value in step S47, therefore -127 is generated as the value D6 of the reproduced and processed coefficient RQg2. This value is equal to the value Bl of the transformation coefficient shown in Figure 8 (b).

Adding or subtracting 256 a / from the value of the coefficient reproduced according to the decision results in steps S45 and S46 is equivalent to an arithmetic processing in which a reproduced coefficient expressed by a binary number is transformed to a binary number comprising 8 lower bits of the reproduced coefficient.

The reproduced and processed coefficients RQg2 thus transformed are transformed to restored DCT coefficients RFg by inverse quantization in the inverse quantizer 404, and the restored DCT coefficients are transformed to a restored image signal RBg corresponding to each block by inverse DCT in the unit DCT 405 (step S49).

Subsequently, in the image decoding apparatus 400, it is decided whether or not the reference block is the last block between the blocks that make up a frame or an object area (step S50). When the reference block is the last block, the predicted intraframe decoding is terminated, and the restored image signals RBg corresponding to the particular blocks are integrated and a reproduced image signal RSg having a corresponding scanning line structure is produced. to a frame or object area (step S51). When the reference block is not the last block, the processes of steps S42-S50 are repeated.

As described above, according to the method of the present invention, a coding signal Cg obtained by coding the difference coefficients, which have been obtained compressively by transforming an image signal, by variable length coding or fixed length coding, and the BLg code sequences each comprising 8 lower bits of a binary code corresponding to each of the difference coefficients, in particular they are transformed to restored difference coefficients REg represented by decimal numbers of agreement to the appropriate decoding processes. Subsequently, the reproduced coefficients obtained by adding prediction coefficients to the restored difference coefficients are compared with 256 (28) and -256 (-28) and, according to the result of the comparison, 256 a / of the coefficients is added or subtracted. reproduced, and the reproduced and processed coefficients RQg2 are subjected to inverse transformation to generate a reproduced image signal RSg. Therefore, it is possible to perform an intraframe predicted decoding process adapted to an intraframe predicted coding process in which the difference coefficients which can not be encoded by variable length coding or fixed length coding (8 bits) are transformed to binary codes, and a code sequence comprising 8 lower bits of each of the binary codes is produced as a coded signal.

When a coding program or a decoding program for carrying out the image coding process or the image decoding process according to the aforementioned modes is recorded in a storage medium such as a floppy disk , the coding or decoding process can be easily carried out in a separate computer system.

Figures 12 (a) -12 (c) are diagrams to explain the case where the process of »encoding images by the image coding apparatus according to the first or third embodiment, or the process of decoding images by the The image decoding apparatus according to the second or fourth mode is executed by a computer system, using a flexible disk which contains a program of the encoding or decoding process.

Figure 12 (b) shows a front view of a flexible disk FD, a cross-sectional view thereof, and a body of the flexible disk D. Figure 12 (a) shows an example of a physical format of the flexible disk body D. The body of the flexible disk D is included in a case F. On the surface of the body of the disk D, a plurality of tracks Tr are formed concentrically from the outer circumference of the disk to the inner circumference. Each track is divided into 16 sectors (Se) in the angular direction. Therefore, in the flexible disk FD containing the program mentioned above, the program data is recorded in the assigned sectors in the body of the floppy disk D.

Figure 12 (c) shows the structure for registering the program on the flexible disk FD and carrying out the image processing by software using the program stored on the flexible disk FD. When the program is registered on the FD floppy disk, the program data is written to the FD floppy disk from the Cs computer system to the FDD floppy disk drive. When the image coding apparatus or the image decoding apparatus mentioned above is constructed in a computer system Cs by the program registered in the flexible disk FD, the program is read from the flexible disk FD by the drive of the flexible disk FDD and then it is loaded into the Cs computer system.

Although the processing of images by the computer system is described using a flexible disk as an example of a data storage medium, the image processing can be carried out in a similar manner with an optical disk. The data storage medium is not limited to these disks, and any means may be employed since the program may be, for example, an IC card or a ROM cartridge.

It is noted that in relation to this date, the best method known by the applicant to carry out the aforementioned invention, is the conventional method for manufacturing the objects to which it refers.

Having described the invention as above, the content of the following is claimed as property.

Claims (11)

1. An image coding method for transforming an image signal into transformation coefficients by compressing data of the image signal for each processing unit, and encoding the transformation coefficients in a coding process using code words comprising a number predetermined codes, characterized in that the coding process comprises the steps of: identify whether or not the values of the reference difference coefficients, which are differences between the reference transformation coefficients that are processed and the transformation coefficients as prediction coefficients different from the reference transformation coefficients, correspond to any of the codes that make up the code words; Y coding the reference difference coefficient when the values of the reference difference coefficients correspond to any of the codes in the code words; and coding the reference transformation coefficients when at least one of the values of the reference difference coefficients corresponds to none of the codes in the code words.

2. The method of image processing according to claim 1, characterized in that: the "code" words are composed of a plurality of variable length codes, and a plurality of fixed length codes having the same bit length; In the process of coding each of the reference transformation coefficients, when there is a variable length code corresponding to the value of the reference transformation coefficient, the coefficient? e reference transformation is encoded using the variable length code, and when there is no variable length code corresponding to the value of the reference transformation coefficient, the reference transformation coefficient is coded using a fixed length code; Y in the coding process each of the reference difference coefficients, when there is a variable length code corresponding to the value of the reference difference coefficient, the reference difference coefficient is coded using the variable length code, and when not there is a variable length code corresponding to the value of the reference difference coefficient, the reference difference coefficient is coded using a fixed length code.

3. An image coding method for dividing an image signal into image signals respectively corresponding to a plurality of unit areas into which an image space is divided for visual presentation, transforming an image signal corresponding to a unit area reference which is coded into transformation coefficients by data compression, and subjecting the transformation coefficients in the reference unit area to a coding process using code words comprising a predetermined number of codes, characterized in that the coding process It includes the steps of: identify whether or not the values of the reference difference coefficients of the reference unit area, which are differences between the transformation coefficients in the reference unit area and the transformation coefficients as prediction coefficients in another unit area adjacent to the reference unit area, correspond to any of the codes that make up the code words; and encoding the difference coefficients in the reference unit area when the values of the difference coefficients in the reference unit area correspond to any of the codes in the code words; and encoding the transformation coefficients in the reference unit area when at least one of 1: > s values of the difference coefficients in the reference unit area correspond to none of the codes in the code words.

4. An image decoding method for decoding a coded image signal based on an external identification signal, characterized in that the decoding comprises the steps of: deciding whether an encoded image signal that is decoded corresponds to reference transformation coeients obtained by compressing data from an image signal or reference difference coeients which are differences between the reference transformation coeients and the transformation coeients as different prediction coeients of the reference transformation coeients, b ada in the identification signal; decoding the encoded image signal to generate a decoded image signal that is processed as reproduced coeients when the encoded image signal that is decoded corresponds to the reference transform coeients; and generating reproduced coeients by adding prediction coeients obtained from a decoded image signal different from the decoded image signal that is processed, to the decoded image signal that is processed, when the decoded image signal that is decoded corresponds to the decoded image signals that are decoded. reference difference coeients; Y subject the reproduced coeients to data expansion to generate a reproduced image signal.

5. An image coding method for transforming an image signal into transformation coeients by compressing data of the image signal for each processing unit, and subjecting the transformed coeients to an encoding process using code words comprising a plurality of fixed length codes, characterized in that the coding process comprises the steps of: identifying whether or not the values of the reference difference coeients, which are differences between the reference transformation coeients that are processed and the transformation coeients as different prediction coeients of the reference transformation coeients, correspond to any of the codes that make up the code words; Y encoding the reference difference coeients to produce an encoded image signal when the values of the reference difference coeients correspond to any of the codes in the code words; and transforming the reference difference coeients to binary codes corresponding to their values and having a longer code length than that of the fixed length codes and then producing code sequences, each having a bit width equivalent to the code length of the fixed-length codes and corresponding to the lowest part of each binary code, such as an encoded image signal, when at least one of the values of the reference difference coeients corresponds to any of the codes in the codes. code words.

6. An image decoding method for decoding a coded difference signal reproducing transformation coeients, the coded difference signal that is obtained through the following steps: subjecting an image signal to data compression for each processing unit to generate transformation coeients; and encoding difference coeients, which are differences between the reference transformation coeients that are encoded and other transformation coeients which have been previously generated to the reference transformation coeients, using first code words comprising a plurality of codes fixed length or secondly code words comprising binary codes each having a length of code longer than that of fixed length codes: characterized in that the decoding method comprises the steps of: decoding a coded difference signal that is processed to generate a decoded difference signal that is processed; adding transformation coefficients already reproduced to the decoded difference signal that is processed, as prediction coefficients, to generate reproduced coefficients that are processed; compare the value of each of the reproduced coefficients that are processed with a positive maximum value and a negative minimum value between the binary reference numbers each having the number of digits one bit less than the code length of the length codes fixed; Y add a maximum absolute value of an arithmetic binary number that has the number of digits equal to the code length of the fixed length codes, to the value of the reproduced coefficient, and produce the sum, when the value of the reproduced coefficient is less than the negative minimum value of the reference binary number; subtract the maximum absolute value of the arithmetic binary number of the reproduced coefficient and produce the difference when the value of the reproduced coefficient is greater than the maximum positive value of the reference binary number; and produce the reproduced coefficient as it is when the value of the reproduced coefficient is greater than the minimum negative value and less than the positive maximum value.

7. An image coding apparatus, characterized in that it comprises a data compression unit for transforming an image signal into transformation coefficients by data compression for each processing unit, and a coding unit for encoding the transformation coefficients using words of code comprising a predetermined number of codes: the data compression unit comprises; generating means of difference coefficients to generate reference difference coefficients which are differences between the reference transformation coefficients that are processed and transformation coefficients as different prediction coefficients of the reference transformation coefficients; Y the coding unit comprises; means of identification to identify whether the values of the reference difference coefficients correspond to any of the codes in the code words, and coding means for encoding the reference difference coefficients when the values of the reference difference coefficients correspond to any of the codes in the code words, while coding the reference transformation coefficients when at least one of the values of the reference difference coefficients correspond to none of the codes in the code words, based on the result of the identification means.

8. An image decoding apparatus for decoding a coded image signal based on an identification signal from the outside, characterized in that the apparatus comprises: decision means for deciding whether a decoded image signal that is decoded corresponds to reference transformation coefficients obtained by compressing data from an image signal, or reference difference coefficients which are differences between the reference and the reference transformation coefficients. the transformation coefficients as different prediction coefficients of the reference transformation coefficients, based on the identification signal; first, means for generating coefficients to generate a decoded agent signal that is processed as first reproduced coefficients by decoding the encoded image signal that is processed; second, means for generating coefficients to generate second reproduced coefficients by adding prediction coefficients obtained from a decoded image signal different from the decoded image signal that is processed, to the decoded image signal that is processed; switching means for selecting the output of the first coefficient generation means when the coded image signal that is processed corresponds to the reference transformation coefficients, while selecting the output of the second coefficient generation means when the signal of The coded image that is processed corresponds to the reference difference coefficients, based on the result of the decision means; Y data expander means for expanding data from the output of the reproduced coefficients of the switching means to generate a reproduced image signal.

9. An image coding apparatus, characterized in that it comprises a data compression unit for transforming an image signal into transformation coefficients by data compression for each processing unit, and a coding unit for encoding the transformation coefficients using words of code comprising a predetermined number of codes: the data compression unit comprises; means for generating difference coefficients to generate reference difference coefficients, which are differences between the reference transformation coefficients that are processed and the transformation coefficients as different prediction coefficients from the reference transformation coefficients; Y the coding unit comprises; decision means to decide whether the values of the reference difference coefficients correspond to any of the codes in the code words, encoding means for encoding the reference difference coefficients using the code words and producing a coded image signal, binary code converter for converting the reference difference coefficients to binary codes corresponding to their values and having a code length greater than that of the fixed-length codes, code output means for producing codes, each having a bit width equivalent to the code length of the fixed length codes and corresponding to the lowest part of each binary code, as an encoded picture signal, and switching means for connecting the output of the difference coefficient generating means to one of the input of the coding means and the input of the binary code converter, based on the result of the decision means, such that the coefficients of reference difference are supplied to the coding means when the values of the reference difference coefficients correspond to any of the codes in the code words, while they are supplied to the binary code converter when at least one of the values of the reference difference coefficients corresponds to none of the codes in the code words.

10. An image decoding apparatus for decoding a coded difference signal reproducing reproduced reproduction coefficients, the coded difference signal that is obtained through the following steps: subjecting an image signal to data compression for each processing unit for generate transformation coefficients; and encoding transformation coefficients, which are differences between the reference transformation coefficients that are encoded and other transformation coefficients which have been previously generated to the reference transformation coefficients, using first of all code words comprising a plurality of fixed length codes or secondly code words comprising binary codes each having a code length greater than that of the fixed length codes: characterized in that the decoding apparatus comprises: a data analyzer for generating a decoded difference signal that is processed by data analysis of a coded difference signal that is processed; a reproduction unit of coefficients to generate reproduced coefficients that are processed by adding transformation coefficients already reproduced as prediction coefficients to the decoded difference signal that is processed; decision means to decide whether or not the value of each of the reproduced coefficients that is processed is greater or less than a positive maximum value and a negative minimum value between the reference binary numbers that have the number of digits a bit smaller than the code length of the fixed-length codes; Y reproduced coefficient processing means including an adder to add a maximum absolute value of an arithmetic binary number having the number of digits equal to the code length of the fixed length codes to the value of the reproduced coefficient being processed, and a subtractor to subtract the maximum absolute value of the arithmetic binary number from the value of the reproduced coefficient being processed, and reproduction means of reproduced coefficients that are carried out as follows based on the result of the decision in the decision means: when the value of the reproduced coefficient is less than the minimum negative value of the binary reference number, produce the sum obtained in the adder; when the value of the reproduced coefficient is greater than the maximum positive value of the reference binary number, produce the difference obtained in the subtractor; and when the value of the reproduced coefficient is greater than the minimum negative value and less than the positive maximum value, produce the reproduced coefficient as it is.

11. A data storage medium, characterized in that it contains a program that processes an image signal, the program is a program which causes a computer to perform the processing of the image signal by an image processing method in accordance with any of Claims 1 to 6. SUMMARY OF THE INVENTION The invention relates to an image coding apparatus comprising a DCT unit for transforming an image signal corresponding to a reference block that is processed into frequency components, a quantizer for quantifying the output of the DCT unit to generate transformation coefficients, a generator of difference coefficients to generate difference coefficients which are differences between the transformation coefficients of the reference block and the prediction coefficients for the reference block, and an identification circuit to identify whether or not values of the difference coefficients are within the range of numerical values which can be expressed by fixed-length codes of N-bits. When the values of the difference coefficients are within the range, these difference coefficients are coded. When at least one of the values of the difference coefficients is outside the range, the transformation coefficients of the reference block are coded. Therefore, when the transformation coefficients, which have been obtained in a compressive manner by transforming an image signal, are encoded using predetermined code words, the predicted intra-frame coding of these transformation coefficients can be reversibly carried out without reduce coding efficiency.