JP6018012B2 - Image encoding method, image decoding method, image encoding device, image decoding device, program thereof, and recording medium - Google Patents

Description

  The present invention relates to encoding and decoding of an image signal, and more particularly to a technique for encoding or decoding an image signal using color space conversion.

The initial profile group defined as Version 1 of the next generation video coding international standard system High Efficiency Video Coding (HEVC) currently being formulated is composed of three types of profiles, which are Main Profile and Main 10 Profile (10-bit). For the input signal), and provided as the Main Still Picture Profile (for still images). The input image signal of each profile has only a YUV 4: 2: 0 color format (hereinafter referred to as 4: 2: 0). 4: 2: 0 indicates a format in which sampling of a color difference with respect to luminance is thinned in half in the horizontal direction and the vertical direction. The following formats are listed in addition to 4: 2: 0, but these formats are mainly used for professional video editing and broadcasting station material transmission.
YUV 4: 4: 4 format (hereinafter referred to as 4: 4: 4): format with the same sampling of luminance and color difference YUV 4: 2: 2 format (hereinafter referred to as 4: 2: 2): color difference with respect to luminance A format that thins sampling in half horizontally

  As a color space, there is an RGB format in addition to the YUV format described above. The RGB format is widely used in video reproduction such as output from a digital camera or various displays. In the case of the RGB format, sampling corresponding to each channel 4: 4: 4 is generally performed.

  In order to support 4: 4: 4, 4: 2: 2, RGB, and input signals of 10-bit or more, standardization of HEVC Version 2 (for example, refer to Non-Patent Document 1) proceeds in parallel with Version 1. In particular, highly efficient coding tools have been proposed for compression of color difference signals. For example, a color space conversion matrix is derived from the decoded signal, a signal obtained by applying the conversion matrix to the residual signal after prediction is coded in Coding Unit, and the color space conversion in the reverse direction is performed on the decoding side. A method (for example, see Non-Patent Document 2) that returns to the original color space has been proposed.

D. Flynn, JS Rojals, T. Suzuki, "High Efficiency Video Coding (HEVC) Range Extensions text specification: Draft 2 (for PDAM)," JCTVC-L1005.doc, Joint Collaborative Team on Video Coding (JCT-VC) of ITU-T SG16 WP3 and ISO / IEC JTC1 / SC29 / WG11, 12th Meeting: Geneva, CH, 14-23 Jan. 2013. K. Kawamura, T. Yoshino, S. Naito, "AHG7: In-loop color-space transformation of residual signals for range extensions," JCTVC-L0371.doc, Joint Collaborative Team on Video Coding (JCT-VC) of ITU- T SG16 WP3 and ISO / IEC JTC1 / SC29 / WG11, 12th Meeting: Geneva, CH, 14-23 Jan. 2013.

  By the way, in the method described in Non-Patent Document 2, a color space conversion matrix is generated without overhead using a pixel group of an encoded adjacent coding unit and a pixel group in a reference block by motion prediction. However, since the color space conversion process is located inside the encoding loop, it causes a delay in the encoding device and the decoding device, and since information between channels is used, each channel of RGB or YUV is independent. There are problems such as being unable to perform processing.

  In addition, color space conversion is performed by inputting a signal that has been pre-converted by a pre-filter with respect to a sequence to be encoded as an original signal to an encoding device, and performing reverse conversion of the color space by a post-filter during decoding. The effect can be realized. When color space conversion is performed using a pre-filter and a post-filter, the problem of independent processing of each channel is solved, and the color space conversion processing inside the encoding loop is eliminated, so that the delay of the encoding device and decoding device is eliminated. be able to. Further, since color space conversion is realized by determinant multiplication and addition / subtraction independently for each pixel, the calculation load can be greatly reduced by a processor specialized for parallel operations. However, when color space conversion is performed using a pre-filter and a post-filter, it is necessary to transmit a color conversion matrix to a decoding device, and overhead for expressing the matrix is generated.

  On the other hand, there is a method for deriving a rotation matrix using principal component analysis in color space conversion. In color space conversion by principal component analysis, a given signal group is converted to a different color space so that the first principal component to the third principal component are assigned in the order of the concentration of the signal group. Here, the obtained rotation matrix can be expressed by three orthogonal rotation angles by the xyz axis in consideration of positive / negative sign inversion, and thus a 3 × 3 rotation matrix can be expressed by only three rotation angles. It becomes like this.

For example, the rotation matrix M _ENC obtained on the encoding device side is expressed as follows by rotation in zyx order using a rotation matrix generated from rotation angles (rx _ENC , ry _ENC , rz _ENC ) of three axes. Is done.

At this time, the inverse transformation rotation matrix M _DEC to be applied on the decoding device side can be expressed as follows by rotation in xyz order.

  Here, when the color space of the original image is converted as a prefilter, a value that cannot be expressed numerically in the conventional color space (a value of 0 or less or a value of 255 or more in the case of an 8-bit sequence) due to the rotation matrix. May occur. These values can be avoided by forcing the clipping to limit from 0 to the maximum allowed by the bit depth of the sequence, or by scaling, but in the case of clipping, the reproducibility of the original image may be significantly reduced, In the case of scaling, it is necessary to specify how much the value is divided.

  Further, when a signal indicating grayscale luminance equivalent is assigned to the first principal component by color space conversion, a signal corresponding to a color difference is assigned to the second and third principal component signals. The contribution ratio of the color difference component in the image is lower than the luminance, and the original image reproducibility is not greatly impaired even if the luminance is quantized larger than the luminance signal. Here, ITU-RBT. 709 (Rec. ITU-R BT.709-3, PARAMETER VALUES FOR THE HDTV STANDARDS FOR PRODUCTION AND INTERNATIONAL PROGRAMME EXCHANGE, Feb. 1998.), the color difference component signal in the YCbCr conversion defined by RGB is 0. .0 to 1.0 are defined as follows, and these coefficient values are designed so that no significant loss occurs subjectively / objectively.

Y = 0.2126R + 0.7152G + 0.0722B (9)
Cb = 0.5389 (B−Y) = − 0.1146R−0.3854G + 0.5000B (10)
Cr = 0.6350 (R−Y) = 0.024R−0.4563G−0.0461B (11)

  In addition, when each coefficient value is obtained by a rotation matrix by principal component analysis, the square sum of each coefficient value from the first principal component to the third principal component is always designed to be 1.0. At this time, if the second and third principal components corresponding to the color difference component are not scaled, the reproducibility of the original image is improved, but the amount of generated code of the color difference component is increased, so that there is a problem that the encoding efficiency is not improved so much.

  The present invention was made in view of such circumstances, and by scaling within a range that does not significantly impair the original image reproducibility, the generated code amount that clearly specifies the scaling value for the luminance component is suppressed, An image encoding method, an image decoding method, an image encoding device, an image decoding device, and a program and recording thereof that can improve the encoding efficiency when encoding the image after pre-filtering for the color difference component The purpose is to provide a medium.

  The present invention is an image encoding method performed by an image encoding apparatus that converts an image input as an RGB signal into a signal of a different color space using a matrix and encodes the converted signal, Expressing the matrix with orthogonal three-axis rotation angles, encoding the rotation angles of each axis, encoding the scaling values of each axis in the coordinate axes after the color space conversion, and one or more of each axis Encoding the difference information between the scaling value and a predetermined value, and encoding using a value obtained by dividing the value on each axis subjected to color space conversion by the scaling value as an encoding target image. It is characterized by that.

  The present invention is characterized in that, among the predetermined values, the value of one axis is √3 or an integer value at the time of an integer calculation according to the value.

  The present invention relates to a value obtained by dividing the value on each axis subjected to the color space conversion by the scaling value, and a value that exceeds the maximum bit depth of the input image and a minimum value for the encoding target image. It is characterized by clipping values below.

  The present invention multiplies a decoded signal by a scaling value of each axis included in an input encoded stream, generates a matrix from orthogonal three-axis rotation angles included in the encoded stream, and converts the decoded signal into an RGB signal. An image decoding method performed by an image decoding device to convert, wherein the information to be decoded is a differential scaling value from a predetermined value with respect to the scaling value, and the original scaling value is subtracted from the predetermined value by subtracting the differential scaling value The method includes a step of deriving a scaling value.

  The present invention is characterized in that, among the predetermined values, the value of one axis is √3 or an integer value at the time of an integer calculation according to the value.

  The present invention is an image encoding device that converts an image input as an RGB signal into a signal of a different color space using a matrix and encodes the converted signal, and the matrix is orthogonal to three axes. Means for encoding the rotation angle of each axis, encoding the rotation angle of each axis, means for encoding the scaling value of each axis in the coordinate axis after the color space conversion, and one or more scaling values of each axis Encoding means for encoding difference information with respect to a predetermined value and using a value obtained by dividing a value on each axis subjected to color space conversion by the scaling value as an encoding target image.

  The present invention multiplies a decoded signal by a scaling value of each axis included in an input encoded stream, generates a matrix from orthogonal three-axis rotation angles included in the encoded stream, and converts the decoded signal into an RGB signal. An image decoding device for conversion, wherein the information to be decoded is a differential scaling value from a predetermined value with respect to the scaling value, and the original scaling value is derived by subtracting the differential scaling value from the predetermined value Means are provided.

  The present invention is an image encoding program for causing a computer to execute the image encoding method.

  The present invention is an image decoding program for causing a computer to execute the image decoding method.

  The present invention is a computer-readable recording medium on which the image encoding program is recorded.

  The present invention is a computer-readable recording medium on which the image decoding program is recorded.

  According to the present invention, by scaling within a range that does not significantly impair the original image reproducibility, the generated code amount that clearly specifies the scaling value is suppressed for the luminance component, and the pre-filter is performed for the color difference component. Coding efficiency when coding an image can be improved.

It is a block diagram which shows the structure of the image coding apparatus in one Embodiment of this invention. It is a flowchart which shows operation | movement of the image coding apparatus shown in FIG. FIG. 2 is a block diagram showing a detailed configuration of a prefilter color conversion matrix generation unit 101 shown in FIG. 1. 4 is a flowchart illustrating an operation of a prefilter color conversion matrix generation unit 101 illustrated in FIG. 3. FIG. 2 is a block diagram showing a detailed configuration of a prefilter processing unit 102 shown in FIG. 1. 6 is a flowchart illustrating an operation of the prefilter processing unit 102 illustrated in FIG. 5. It is a block diagram which shows the structure of the image decoding apparatus in one Embodiment of this invention. It is a flowchart which shows operation | movement of the image decoding apparatus shown in FIG. It is a block diagram which shows the detailed structure of the post filter process part 210 shown in FIG. 10 is a flowchart showing an operation of a post filter processing unit 210 shown in FIG. 9. It is a figure which shows the hardware structural example of the system which comprised the image coding apparatus by the computer and the software program. FIG. 3 is a diagram illustrating a hardware configuration example of a system in which an image decoding device is configured by a computer and a software program.

  Hereinafter, an image encoding device and an image decoding device according to an embodiment of the present invention will be described with reference to the drawings. First, the image coding apparatus will be described. FIG. 1 is a block diagram showing a configuration of an image encoding device according to the embodiment. In the present specification, an image refers to an image for one frame constituting a still image or a moving image. A video has the same meaning as a moving image. The image encoding device shown in FIG. 1 is different from a general image encoding device represented by HEVC or the like in that a pre-filter color conversion matrix generation unit 101 and a pre-filter processing unit 102 are newly provided. And the part of the function of the entropy encoding unit 112 is different. The configuration excluding the prefilter color conversion matrix generation unit 101, the prefilter processing unit 102, and the entropy encoding unit 112 is an example and is not limited to this configuration.

  Here, the configuration of the image encoding device will be described with reference to FIG. The image encoding device receives an encoding target sequence, divides each encoding target image into blocks, encodes each block, and outputs the bit stream as an encoded stream.

  The encoding target block is obtained by using a subtractor 114 to obtain a difference from the prediction block of the intra prediction unit 103 or the inter prediction unit 104, and this is used as a prediction residual signal. The transform unit 105 performs orthogonal transform (DCT) on the prediction residual signal and outputs transform coefficients. The quantization unit 106 quantizes the transform coefficient and outputs the quantized transform coefficient. The inverse quantization unit 107 performs inverse quantization of the output of the quantization unit 106, and the inverse transform unit 108 performs inverse transform of the transform unit 105. The inverse transformed value is added and synthesized by the prediction block and the adder 115 and stored in the decoded image storage unit 109.

  The image in the decoded image storage unit 109 is used in the intra prediction unit 103. When the above-described processing is completed for all the blocks of the processing target frame, the in-loop filter processing unit 110 performs image processing filtering for removing the coding distortion of the decoded image, and the decoded image after filtering is stored in the frame buffer 111. Stored in The decoded image stored in the frame buffer 111 is used by the inter prediction unit 104. Prediction information and post-quantization transform coefficients of the intra prediction unit 103 and the inter prediction unit 104, and in-loop filter information of the in-loop filter processing unit 110 are input to the entropy encoding unit 112, and an encoded stream is output. The encoded information storage unit 113 stores various values such as various encoded block sizes and prediction information that can be referred to on the decoding device side, coefficient values after quantization, in-loop filter overhead, and the like. The information 113 is quoted by various processing units in the apparatus.

  Next, functions of the prefilter color conversion matrix generation unit 101, the prefilter processing unit 102, and the entropy encoding unit 112 newly provided for the above-described general image encoding device will be described. The prefilter color conversion matrix generation unit 101 generates a color conversion matrix by applying it to the encoding target sequence, and outputs color conversion information and scaling information. The prefilter processing unit 102 converts the color space by applying the color conversion matrix to the encoding target image and outputs the result. The entropy encoding unit 112 encodes and outputs the color conversion information and scaling information output from the prefilter color conversion matrix generation unit 101.

  Next, the operation of the image encoding device shown in FIG. 1 will be described with reference to FIG. FIG. 2 is a flowchart showing the operation of the image encoding device shown in FIG. First, the prefilter color conversion matrix generation unit 101 generates a color conversion matrix, color conversion information, and scaling information from the input encoding target sequence (step S1). Subsequently, the prefilter processing unit 102 reads the encoding target image from the encoding target sequence (step S2), and performs color conversion on the encoding target image based on the information generated by the prefilter color conversion matrix generation unit 101. Is applied (step S3).

  Next, the encoding target image is read for each block (step S4), and the subtractor 114 takes the difference from the prediction block, generates a difference block, and outputs the difference block (step S5). Subsequently, the transform unit 105 applies orthogonal transform to the difference block and outputs it (step S6). In response to this, the quantization unit 106 quantizes and outputs the coefficient value after the orthogonal transformation (step S7).

  Next, the inverse quantization unit 107 inversely quantizes the quantized coefficient value and outputs it (step S8). In response to this, the inverse transform unit 108 performs inverse orthogonal transform on the coefficient value after inverse quantization and outputs the result (step S9). Subsequently, the adder 115 performs addition synthesis with the prediction block and stores it in the decoded image storage unit 109 (step S10). Then, it is determined whether all the blocks have been processed (step S11). If there is an unprocessed block, the process returns to step S4.

  Next, the in-loop filter processing unit 110 applies an in-loop filter to the decoded image stored in the decoded image storage unit 109, and stores it in the frame buffer 111 (step S12). Subsequently, the entropy encoding unit 112 performs entropy encoding on the prediction information, the quantized coefficient value, the in-loop filter information, the color conversion information, the scaling information, and the like (step S13). Then, it is determined whether or not the processing of all frames has been completed (step S14). If there is an unprocessed frame, the process returns to step S2 to repeat the processing, and when all the frames are completed, the entropy coding unit 112 outputs the encoded stream (step S15).

  Next, a detailed configuration of the prefilter color conversion matrix generation unit 101 shown in FIG. 1 will be described with reference to FIG. FIG. 3 is a block diagram showing a detailed configuration of the prefilter color conversion matrix generation unit 101 shown in FIG. The prefilter color conversion matrix generation unit 101 receives an encoding target sequence, and outputs a color conversion matrix, color conversion information, and scaling information from the input encoding target sequence. The principal component analysis unit 1011 receives the encoding target sequence and obtains a color conversion matrix. As a color conversion matrix generation method, for example, a method obtained using principal component analysis is conceivable. The transposed matrix conversion unit 1012 changes the color conversion matrix to a transposed matrix and outputs it to the triaxial rotation angle analysis unit 1013. A triaxial rotation angle analysis unit 1013 minimizes an error from a value output by multiplication of an input transposed matrix and a triaxial rotation matrix orthogonal to each other, and outputs a rotation angle around each axis. When the output rotation angle is rounded to integer precision, the output value of the transposed matrix conversion unit 1012 and the rotation matrix reconstructed from the three rotation angles may not match. In addition, a reconfigurable color conversion matrix is output.

  The scaling value analysis unit 1014 calculates a scaling value by applying a rotation matrix to the encoding target sequence, calculating the maximum absolute value on each axis after color space conversion. The scaling value obtained at this time is half of the maximum pixel value allowed by the bit depth by subtracting half the maximum pixel value allowed by the bit depth of the sequence from each RGB signal and multiplying by the color conversion matrix. The value divided by. Each row of the color conversion matrix is divided by the scaling value obtained here, and a scaled color conversion matrix is output. The scaling value analysis unit 1014 outputs the difference between the scaling value and a predetermined value as scaling information. For example, in the case of All Intra in which all frames are encoded by intra-screen prediction, the color conversion matrix is generated in units of frames, in the case of a Closed GOP (Group of Picture) structure, for each GOP structure size, or the entire sequence. A pattern or the like to be generated is assumed.

  Next, the operation of the prefilter color conversion matrix generation unit 101 shown in FIG. 3 will be described with reference to FIG. FIG. 4 is a flowchart showing the operation of the prefilter color conversion matrix generation unit 101 shown in FIG. First, the principal component analysis unit 1011 generates and outputs a color conversion matrix from the encoding target sequence by principal component analysis (step S21). Subsequently, the transposed matrix conversion unit 1012 performs a transposed matrix conversion on the color conversion matrix and outputs the result (step S22).

  Next, the triaxial rotation angle analysis unit 1013 performs a triaxial rotation angle analysis, and outputs a rotation angle around each axis and a reconfigurable color conversion matrix (step S23). Subsequently, the scaling value analysis unit 1014 converts the color space of each encoding target image in the encoding target sequence, calculates the maximum value of the absolute value at the time of conversion for each axis, calculates the scaling value, and outputs it. (Step S24).

  Next, the scaling difference calculation unit 1015 outputs a scaled color conversion matrix by dividing each row of the color conversion matrix by the scaling value (step S25). Then, the difference between the scaling value and the predetermined value is output as scaling information (step S26).

  Next, a detailed configuration of the prefilter processing unit 102 shown in FIG. 1 will be described with reference to FIG. FIG. 5 is a block diagram showing a detailed configuration of the prefilter processing unit 102 shown in FIG. The prefilter processing unit 102 inputs each encoding target image and color conversion matrix in the encoding target sequence, applies the color conversion matrix to each encoding target image in the encoding target sequence, and applies each prefilter applied The encoding target image is output. The prefilter execution unit 1021 inputs each encoding target image in the encoding target sequence and the color conversion matrix output by the prefilter color conversion matrix generation unit 101, and converts each RGB channel into a different color space. In this conversion, half of the pixel maximum value allowed by the bit depth of the sequence is subtracted from each RGB signal, multiplied by the color conversion matrix, and after scaling, half of the bit depth maximum value is added to 0. To a value clipped between the maximum pixel values allowed by the bit depth.

  Next, the operation of the prefilter processing unit 102 shown in FIG. 5 will be described with reference to FIG. FIG. 6 is a flowchart showing the operation of the prefilter processing unit 102 shown in FIG. First, the prefilter execution unit 1021 performs color conversion on each encoding target image (step S31). Then, the prefilter execution unit 1021 outputs each encoding target image after applying the prefilter (step S32).

  Here, the color space conversion will be described. When color space conversion is performed on the RGB signal of a general natural image, a high concentration appears in the gray scale of R = G = B in most cases, so the axis of R = G = B is assigned as the first principal component. In many cases, rotation is performed. For example, assuming that the axis of R = G = B is allocated on the x-axis representing the first principal component, the variance of the signal values on the yz-axis representing the second and third principal components is relatively small. In addition, for general natural images, the image is shot to the full range allowed by the bit depth by adjusting the ISO sensitivity and iris of the camera at the time of shooting so that the gradation of the subject looks the clearest with the highest contrast. Is normal. When these videos are targeted, in the case of an 8-bit sequence, the luminance values are widely dispersed, and most of them become the maximum value RGB (255, 255, 255) and the minimum value RGB (0, 0, 0). Accordingly, the scaling value of the first principal component is concentrated on √3 (≈1.73205) corresponding to the length of the diagonal component in the cube of one side. For this reason, the scaling value of the first principal component is more concentrated to 0 when the difference from √3 is encoded, so that the generated code amount can be suppressed compared to the case where the scaling value is encoded as it is. it can.

  Further, as a scaling value that does not significantly impair the original image reproducibility, for example, a ratio to the square root of the square sum of each coefficient value in the YCbCr conversion described above can be used. The ratio of the square root 0.6612 of the mean square of the sum of squares of the coefficient values used for derivation of Cb and Cr to 0.4612 and the square sum of the squares of the coefficient values of the rotation matrix by principal component analysis is approximately When the scaling value of the color difference component is encoded, the difference from this value is encoded. As a result, it is possible to describe with a smaller amount of code compared to the case where the scaling value is encoded as it is, and it is possible to suppress the generated code amount of the chrominance component with a low contribution rate. Improvement can be expected.

  The representative value of 2.2873 is an example calculated from the coefficient value of YCbCr conversion, and other values may be set. When experimenting with a plurality of patterns in HEVC Version 2 (Non-patent Document 1), it was confirmed that a favorable improvement in coding efficiency was confirmed at about 1.7818, so such a value may be used.

  Next, an image decoding device will be described. FIG. 7 is a block diagram showing the configuration of the image decoding apparatus in the embodiment of the present invention. The image decoding apparatus shown in FIG. 7 differs from a general image decoding apparatus represented by HEVC or the like in an entropy decoding unit 201 and a post filter processing unit 210. The configuration excluding the entropy decoding unit 201 and the post filter processing unit 210 is an example, and is not limited to this configuration.

  Here, the configuration of the image decoding apparatus shown in FIG. 7 will be described. The image decoding apparatus receives an encoded stream, converts the encoded stream into a decoded sequence, and outputs the decoded sequence as an output signal. The entropy decoding unit 201 receives an encoded stream and decodes prediction information, quantized transform coefficients, and in-loop filter information for each block. The inverse quantization unit 202 performs inverse quantization on the decoded post-quantization transform coefficient. The inverse transform unit 203 performs inverse orthogonal transform (IDCT) on the output after inverse quantization. The output signal of the inverse transform unit 203 is added and synthesized by the prediction block of the intra prediction unit 204 or the inter prediction unit 205 and the adder 211 and stored in the decoded image storage unit 206. The image in the decoded image storage unit 206 is used in the intra prediction unit 204. When the above-described processing is completed for all the blocks of the processing target frame, the in-loop filter processing unit 207 performs image processing filtering for removing the coding distortion of the decoded image, and the decoded image after filtering is stored in the frame buffer. Store in 208. The image in the frame buffer 208 is used in the inter prediction unit 205. The encoded information storage unit 209 stores various values such as various block sizes and prediction information, coefficient values after quantization, and in-loop filter overhead. Information in the encoded information storage unit 209 is cited by various processing units in the apparatus.

  Here, in addition to the image decoding apparatus described above, the entropy decoding unit 201 decodes color conversion information and scaling information necessary for the execution of the post filter processing unit 210. Subsequently, the post filter processing unit 210 is applied to the output image of the in-loop filter processing unit 207, and the image obtained by performing the inverse transformation of the color space is displayed on the display or stored in the recording device as the final output signal.

  Next, the operation of the image decoding apparatus shown in FIG. 7 will be described with reference to FIG. FIG. 8 is a flowchart showing the operation of the image decoding apparatus shown in FIG. First, the entropy decoding unit 201 reads an encoded stream (step S41). Subsequently, the entropy decoding unit 201 performs entropy decoding of the encoded stream, and decodes prediction information, coefficient values after quantization, information such as in-loop filter overhead, color conversion information, scaling information, and the like (step S42).

  Next, the inverse quantization unit 202 inversely quantizes and outputs the quantized coefficient value (step S43). In response to this, the inverse transform unit 203 performs inverse orthogonal transform on the coefficient value after inverse quantization. (Step S44). Subsequently, the adder 211 performs addition synthesis with the prediction block and stores it in the decoded image storage unit 206 (step S45). Then, it is determined whether or not all blocks have been processed (step S46). If there is an unprocessed block, the process returns to step S43.

  Next, the in-loop filter processing unit 207 applies an in-loop filter to the decoded image, and stores the decoded image in the frame buffer 208 (step S47). Then, it is determined whether or not all frames have been processed (step S48). If there is an unprocessed frame, the process returns to step S43.

  Next, the post filter processing unit 210 applies a post filter using the color conversion information and scaling information (step S49), and outputs a decoded sequence as an output signal (step S50).

  Next, a detailed configuration of the post filter processing unit 210 shown in FIG. 7 will be described with reference to FIG. FIG. 9 is a block diagram showing a detailed configuration of the post filter processing unit 210 shown in FIG. The post filter processing unit 210 receives the decoded image, color conversion information, and scaling information, applies a color conversion matrix to the decoded image, and outputs a decoded image after applying the post filter. The post-filter color conversion matrix generation unit 2101 receives the color conversion information (three-axis rotation angle) and scaling information, and derives the original scaling value by subtracting the differential scaling value described in the scaling information from a predetermined value. The original scaling value is multiplied to each row of the color conversion matrix generated from the three-axis rotation angles obtained by inverting the sign. The post filter execution unit 2102 receives the color conversion matrix and the decoded image as input, and inversely converts the decoded image into the RGB color space. In this inverse transform processing, each signal of the decoded image is subjected to scaling by subtracting half of the maximum pixel value allowed by the bit depth of the sequence and multiplying by the color transformation matrix, and then the maximum pixel value allowed by the bit depth. Half of the values are added and a value clipped between 0 and the maximum bit depth is output.

  Next, the operation of the post filter processing unit 210 shown in FIG. 9 will be described with reference to FIG. FIG. 10 is a flowchart showing the operation of the post filter processing unit 210 shown in FIG. First, the post-filter color conversion matrix generation unit 2101 subtracts the differential scaling value written in the scaling information from a predetermined value to generate a scaling value (step S61). Then, the post-filter color conversion matrix generation unit 2101 generates a color conversion matrix from the color conversion information (triaxial rotation angle), and integrates the scaling values corresponding to each row of the color conversion matrix (step S62). Subsequently, the post filter execution unit 2102 performs color inverse transformation on the decoded image (step S63), and outputs the decoded image after application of the post filter (step S64).

  The image encoding and image video decoding processes described above can be realized by a computer and a software program, and the program can be provided by being recorded on a computer-readable recording medium or provided through a network. Is also possible.

  FIG. 11 is a diagram illustrating a hardware configuration example of a system in which the above-described image encoding device is configured by a computer and a software program. This system includes a CPU A0 that executes a program, a memory A1 such as a RAM that stores programs and data accessed by the CPU A0, and an image signal input unit that inputs an image signal to be encoded from a camera or the like. A2 may be a storage unit that stores an image signal by a disk device or the like), and a program storage in which an image encoding program A4 that is a software program that causes the CPU A1 to execute processing for encoding an input image by the above-described method is stored. The encoded data generated by executing the image encoding program A4 loaded into the memory A1 by the device A3 and the CPU A0, for example, an encoded data output unit A5 (by a disk device or the like) for outputting via a network. It may be a storage unit that stores encoded data) and connected by a bus. There.

  FIG. 12 is a diagram illustrating a hardware configuration example of a system in which the above-described image decoding apparatus is configured by a computer and a software program. This system includes a CPU B0 that executes a program, a memory B1 such as a RAM that stores programs and data accessed by the CPU B0, and encoded data that is input with encoded data encoded by an image encoding device. An input unit B2 (which may be a storage unit that stores multiplexed encoded data by a disk device or the like) and an image decoding program B4 that is a software program that causes the CPU B1 to execute a process of decoding the encoded data by the above-described method. Decoding that outputs the decoded image obtained by decoding the encoded data by executing the stored program storage device B3 and the image decoding program B4 loaded into the memory B1 by the CPU B0 to a playback device or the like The image output unit B5 is connected by a bus.

  As described above, an encoding method for converting an RGB signal into a different color space using a rotation matrix, and a decoding method for returning a decoded signal to an RGB signal by converting the color space using a reverse rotation matrix , When encoding the rotation angles of the three axes orthogonal to the notation of the rotation matrix and the scaling values of each axis, the scaling values of one or more axes in the coordinate system after color space conversion are respectively set to predetermined values. By taking this difference, it is possible to describe with a smaller code amount compared to the case where the three-axis scaling value is encoded as it is. For this reason, the generated code amount of the color difference component can be suppressed and the encoding efficiency can be improved.

  You may make it implement | achieve the image coding apparatus and image decoding apparatus in embodiment mentioned above with a computer. In that case, a program for realizing this function may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be read into a computer system and executed. Here, the “computer system” includes an OS and hardware such as peripheral devices. The “computer-readable recording medium” refers to a storage device such as a flexible medium, a magneto-optical disk, a portable medium such as a ROM and a CD-ROM, and a hard disk incorporated in a computer system. Furthermore, the “computer-readable recording medium” dynamically holds a program for a short time like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line. In this case, a volatile memory inside a computer system serving as a server or a client in that case may be included and a program held for a certain period of time. Further, the program may be for realizing a part of the functions described above, and may be a program capable of realizing the functions described above in combination with a program already recorded in the computer system. It may be realized using hardware such as PLD (Programmable Logic Device) or FPGA (Field Programmable Gate Array).

  As mentioned above, although embodiment of this invention has been described with reference to drawings, the said embodiment is only the illustration of this invention, and it is clear that this invention is not limited to the said embodiment. is there. Accordingly, additions, omissions, substitutions, and other changes of the components may be made without departing from the technical idea and scope of the present invention.

  In color space conversion, when the matrix information is clearly indicated by using the rotation angle, it can be used for the purpose of improving the coding efficiency by suppressing the generated code amount.

  101 ... Pre-filter color conversion matrix generation unit, 1011 ... Principal component analysis unit, 1012 ... Transpose matrix conversion unit, 1013 ... Triaxial rotation angle analysis unit, 1014 ... Scaling value analysis unit DESCRIPTION OF SYMBOLS 1015 ... Scaling difference calculation part, 102 ... Pre filter process part, 1021 ... Pre filter execution part, 112 ... Entropy encoding part, 201 ... Entropy decoding part, 210 ... Post Filter processing unit, 2101 ... Post-filter color conversion matrix generation unit, 2102 ... Post-filter execution unit

Claims (11)

An image encoding method performed by an image encoding apparatus that converts an image input as an RGB signal into a signal of a different color space using a matrix and encodes the converted signal,
Expressing the matrix with orthogonal three-axis rotation angles and encoding the rotation angles of each axis;
Encoding a scaling value of each axis in the coordinate axes after color space conversion;
Code that encodes difference information from a predetermined value with respect to one or more scaling values of each axis, and uses a value obtained by dividing the value on each axis that has undergone color space conversion by the scaling value as an encoding target image An image encoding method comprising the steps of:   2. The image encoding method according to claim 1, wherein a value of one axis among the predetermined values is √3 or an integer value at the time of an integer calculation according to the value.   A value obtained by dividing the value on each axis subjected to the color space conversion by the scaling value, and a value that exceeds the maximum value of the bit depth of the input image and a value that is less than the minimum value for the encoding target image. The image encoding method according to claim 1, wherein clipping is performed. Image decoding that multiplies a decoded signal by a scaling value of each axis included in the input encoded stream, generates a matrix from orthogonal three-axis rotation angles included in the encoded stream, and converts the decoded signal into an RGB signal An image decoding method performed by an apparatus,
An image comprising: a step of deriving an original scaling value by subtracting the difference scaling value from the predetermined value, wherein information to be decoded is a differential scaling value from a predetermined value with respect to the scaling value Decryption method.   5. The image decoding method according to claim 4, wherein, of the predetermined values, the value of one axis is √3 or an integer value at the time of an integer calculation according to the same. An image encoding device that converts an image input as an RGB signal into a signal of a different color space using a matrix and encodes the converted signal,
Means for expressing the matrix by orthogonal three-axis rotation angles, and encoding the rotation angles of the respective axes;
Means for encoding the scaling value of each axis in the coordinate axes after color space conversion;
Code that encodes difference information from a predetermined value with respect to one or more scaling values of each axis, and uses a value obtained by dividing the value on each axis that has undergone color space conversion by the scaling value as an encoding target image And an image encoding device. Image decoding that multiplies a decoded signal by a scaling value of each axis included in the input encoded stream, generates a matrix from orthogonal three-axis rotation angles included in the encoded stream, and converts the decoded signal into an RGB signal A device,
An image characterized by comprising: means for deriving an original scaling value by subtracting the difference scaling value from the predetermined value, wherein information to be decoded is a differential scaling value from a predetermined value with respect to the scaling value Decoding device.   An image encoding program for causing a computer to execute the image encoding method according to any one of claims 1 to 3.   An image decoding program for causing a computer to execute the image decoding method according to claim 4 or 5.   A computer-readable recording medium on which the image encoding program according to claim 8 is recorded.   The computer-readable recording medium which recorded the image decoding program of Claim 9.

Images