JP2007104645A - Compressing encoder, compressing encoding method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a compressed image compatible with the target image quality while the image data is encoded by the compressing encoding method in higher speed with less quantity of computation with noise being removed. <P>SOLUTION: A coder quantity controller 22 curtails data to a code train to which rearrangement and bit shift are implemented as shown in Fig. 10 in view of obtaining the predetermined noise removing effect. Curtailment of data is sequentially conducted from the right-end bit. For example, the bit data is sequentially deleted from the bit data of No.0 of VHL 4 shown in Fig. 10 to the bit data of No.0 of YHH5 in the lower direction. When it is assumed that the target noise removing effect can be attained by curtailing the bit data up to YHH1, the data at the scattering point in the relevant Fig. 10 is curtailed. In the case where the desired noise removing effect cannot be obtained even after the data is curtailed up to the data of YHH1, the bit data is sequentially deleted to the lower direction from the bit data of No.0 of VLL4. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a compression encoding apparatus and a compression encoding method used in an image compression / decompression technique.

  As a next-generation high-efficiency encoding system for image data, the ISO (International Organization for Standardization) and ITU-T (International Telecommunication Union Telecommunication Standardization Sector) have developed the JPEG 2000 (Joint Photographic Experts Group 2000) system. The JPEG2000 system has superior functions compared with the current mainstream JPEG (Joint Photographic Experts Group) system, adopts DWT (Discrete Wavelet Transform) as orthogonal transform, and uses bit for entropy coding. It is characterized in that a method called EBCOT (Embedded Block Coding with Optimized Truncation) that performs plain coding is adopted.

  FIG. 23 is a functional block diagram showing a schematic configuration of a compression encoding apparatus 100 that performs compression encoding of an image based on the JPEG2000 system. Hereinafter, the compression encoding procedure of the JPEG2000 system will be outlined with reference to FIG.

  The image signal input to the compression encoding apparatus 100 is subjected to DC level conversion as necessary by the DC level shift unit 102 and then output to the color space conversion unit 103. Next, the color space conversion unit 103 converts the color space of the signal input from the DC level shift unit 102. Here, for example, the RGB signal input to the color space conversion unit 103 is converted into a YCbCr signal (a signal composed of a luminance signal Y and color difference signals Cb and Cr).

  Next, the tiling unit 104 divides the image signal input from the color space conversion unit 103 into a plurality of area components called “tiles” having a rectangular shape, and outputs the result to the DWT unit 105. The DWT unit 105 performs integer type or real number type DWT on the image signal input from the tiling unit 104 in units of tiles, and outputs a conversion coefficient obtained as a result. In the DWT, a one-dimensional filter that divides a two-dimensional image signal into a high frequency component (high frequency component) and a low frequency component (low frequency component) is applied in the order of vertical and horizontal directions. The JPEG 2000 basic method employs an octave division method that recursively divides a band only in a band component that is divided in the vertical direction and the horizontal direction in the low frequency range. The number of times the band is recursively divided is called a decomposition level.

  FIG. 24 is a schematic diagram showing a two-dimensional image 120 that has been subjected to decomposition level 3 DWT according to the octave division method. At decomposition level 1, the two-dimensional image 120 is divided into four band components HH1, HL1, LH1, and LL1 (not shown) by sequentially applying the above-described one-dimensional filter in the vertical direction and the horizontal direction. The Here, “H” indicates a high frequency component, and “L” indicates a low frequency component. For example, HL1 is a band component composed of a high frequency component H in the horizontal direction and a low frequency component L in the vertical direction at the decomposition level 1. Generalizing the notation, “XYn” (X and Y are either H or L; n is an integer equal to or greater than 1) is defined as the horizontal band component X and the vertical band component Y at the decomposition level n. A band component consisting of

  At decomposition level 2, the low frequency component LL1 is band-divided into HH2, HL2, LH2, and LL2 (not shown). Further, at the decomposition level 3, the low frequency component LL2 is band-divided into HH3, HL3, LH3 and LL3. FIG. 24 shows the band components HH1 to LL3 generated as described above. In FIG. 24, an example of the third-order decomposition level is shown. However, in the JPEG2000 system, generally third-order to eighth-order decomposition levels are employed.

  Next, the quantization unit 106 has a function of performing scalar quantization on the transform coefficient output from the DWT unit 105 as necessary. The quantization unit 106 also has a function of performing bit shift processing that prioritizes the image quality of a region of interest (ROI) by the ROI unit 107. Note that, when performing reversible (lossless) transformation, scalar quantization in the quantization unit 106 is not performed. In the JPEG2000 system, two types of quantization means are prepared: scalar quantization in the quantization unit 106 and post-quantization (truncation) described later.

  As a typical usage method of ROI, there is a Max-shift method specified in an optional function of JPEG2000.

  The Maxshift method specifies an ROI portion in an arbitrary form and compresses the portion with high image quality, while compressing the non-ROI portion with low image quality. Specifically, first, wavelet transformation is performed on the original image to obtain wavelet coefficient distributions, and among these distributions, the value Vm of the wavelet coefficient having the largest coefficient distribution corresponding to the non-ROI portion is obtained. Keep it. Then, the number of bits S such that S> = max (Vm) is obtained, and only the wavelet coefficient of the ROI portion is shifted by S bits in the increasing direction. For example, if the value of Vm is “255” in decimal (ie, “11111111” in binary), S = 8 bits, and the value of Vm is “128” in decimal (ie, binary). Similarly, in the case of “10000000”), S = 8 bits, and in this case, the wavelet coefficient of the ROI portion is shifted by S = 8 bits in the increasing direction. As a result, the compression ratio can be set lower for the ROI portion than for the non-ROI portion, and high-quality compressed data can be obtained for the ROI portion.

  Next, the transform coefficient output from the quantization unit 106 is subjected to block-based entropy coding by the coefficient bit modeling unit 108 and the arithmetic coding unit 109 in accordance with the above-described EBCOT, and the code amount control unit 110 The rate is controlled. Specifically, the coefficient bit modeling unit 108 divides the band component of the input transform coefficient into areas called “code blocks” of about 16 × 16, 32 × 32, and 64 × 64, and further each code block. , It is decomposed into a plurality of bit planes composed of a two-dimensional array of bits.

FIG. 25 is a schematic diagram illustrating a two-dimensional image 120 that has been decomposed into a plurality of code blocks 121, 121, 121,... FIG. 26 is a schematic diagram showing _n bit planes 122 _{0 to} 122 _n−1 (n: natural number) constituting the code block 121. As shown in FIG. 26, when the binary value 123 of the conversion coefficient at one point in the code block 121 is “011... 0”, the bits constituting the binary value 123 are respectively bit planes 122 _{n−. 1, 122 n-2, 122} n-3, ..., is decomposed to belong to 122 _0. Bit plane 122 _n-1 in the figure represents the most significant bit plane consisting only of the most significant bit (MSB) of the transform coefficient, and bit plane 122 ₀ is the least significant bit plane consisting only of the least significant bit (LSB). Represents.

Further, the coefficient bit modeling unit 108 performs context determination of each bit in each bit plane 122 _k (k = 0 to n−1), and as shown in FIG. 27, the significance (determination) of each bit is determined. According to the result, the bit plane 122k is decomposed into three types of encoding passes, that is, an SIG pass (SIGnificance propagation pass), an MR pass (Magnitude Refinement pass), and a CL pass (CLeanup pass). The context determination algorithm for each coding pass is defined by EBCOT. According to it, “significant” means a state in which the coefficient of interest is known to be non-zero in the encoding process so far, and “not significant” means that the coefficient value is zero. , Or a state that may be zero.

  The coefficient bit modeling unit 108 performs an SIG path (significant coefficient coding pass with significant coefficients around), an MR path (significant coefficient coding path), and a CL path (SIG path, MR path not corresponding to the rest). Bit plane encoding is performed in three types of encoding passes). Bit plane coding is performed by scanning the bits of each bit plane in units of 4 bits from the most significant bit plane to the least significant bit plane and determining whether or not a significant coefficient exists. The number of bit planes composed only of insignificant coefficients (0 bits) is recorded in the packet header, and actual encoding is started from the bit plane in which significant coefficients first appear. The encoding start bit plane is encoded only by the CL pass, and the bit planes lower than the bit plane are sequentially encoded by the above three types of encoding passes.

FIG. 28 shows an RD curve representing the relationship between rate (code amount; R) and distortion (D). In this RD curve, R ₁ is a rate before bit plane encoding, R ₂ is a rate after bit plane encoding, D ₁ is distortion before bit plane encoding, and D ₂ is distortion after bit plane encoding. , Respectively. A, B, and C are labels representing the above-described encoding pass. In order to perform efficient encoding, among the paths from the start point P ₁ (R ₁ , D ₁ ) to the end point P ₂ (R ₂ , D ₂ ), the path of the convex curve CBA is used. However, it is preferable to adopt a concave-curved ABC path. In order to realize such a concave curve, it is known that encoding may be performed from the MSB plane toward the LSB plane.

  Next, the arithmetic coding unit 109 performs arithmetic coding in units of coding passes on the coefficient sequence from the coefficient bit modeling unit 108 based on the context determination result using the MQ coder. There is also a mode in which the arithmetic encoding unit 109 performs a bypass process in which a part of the coefficient sequence input from the coefficient bit modeling unit 108 is not arithmetically encoded.

  Next, the code amount control unit 110 controls the final code amount by performing post-quantization that truncates the lower bit planes of the code string output by the arithmetic encoding unit 109. Then, the bit stream generation unit 111 generates a bit stream in which the code string output from the code amount control unit 110 and additional information (header information, layer configuration, scalability information, quantization table, etc.) are multiplexed, and a compressed image Output as.

  In the compression coding apparatus having the above configuration, as a method for compressing the data amount of image data, for example, rate / distortion optimization (RD optimization) using a rate control method in the code amount control unit 110 (Refer to Non-Patent Document 1).

David S. Taubman and Michael W. Marcellin, "JPEG2000 IMAGE COMPRESSION FUNDAMENTALS, STANDARDS AND PRACTICE," Kluwer Academic Publishers

  However, in this method, (1) it is necessary to calculate the distortion amount with respect to the rate one by one in each coding pass, and it is necessary to estimate the optimal solution at a certain coding rate, which increases the amount of calculation and real time. There is a problem that (2) a memory for storing the distortion amount calculated in each encoding pass is required.

  In particular, for quantization and code amount control that directly affects the operation performance of the compression coding apparatus, a method for realizing processing while maintaining high image quality efficiently at high speed is desired.

  For example, when it is desired to obtain a skin-beautifying effect in human photography, in the conventional method, a noise removal filter (skin-beautification filter) is disposed in front of the compression encoding device, and noise removal processing is performed on the captured image by the noise removal filter. After that, image compression processing is performed by the compression encoding device. However, in this method, (1) since the image compression process is performed after the noise removal process, even if noise is removed by the noise removal process, new distortion occurs due to the subsequent image compression process. 2) Due to the two-stage configuration of the noise removal filter and the compression encoding device, there is a problem that the processing becomes complicated.

  An object of the present invention in view of the above problems is to compress and encode image data with a small amount of calculation and at high speed. Another object is to obtain a compressed image that matches the target image quality while removing noise.

  A compression encoding apparatus according to a first aspect of the present invention is a compression encoding apparatus that compresses and encodes an image signal, and recursively divides the image signal into a high-frequency component and a low-frequency component by wavelet transform, and A wavelet transform unit that generates and outputs transform coefficients of band components of the image, an image quality control unit that obtains a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient, and entropy-encodes the transform coefficients An entropy encoding unit, and the encoded data output from the entropy encoding unit are rearranged in a predetermined scanning order based on the quantization step size to generate a code sequence, and the encoded data in the code sequence And a code amount control unit that performs rate control to discard a part of the code string so that the total image quality becomes a target image quality.

  A compression encoding apparatus according to a second aspect of the present invention is a compression encoding apparatus that compresses and encodes an image signal, and recursively band-divides the image signal into a high frequency component and a low frequency component by wavelet transform, and A wavelet transform unit that generates and outputs a transform coefficient of the band component, an image quality control unit that obtains a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient, and a region of interest in the image signal A region-of-interest setting unit to be set; and a quantization unit that performs quantization and rearrangement on the transform coefficient based on the quantization step size and rearrangement and bit shift based on the setting information on the region of interest; An entropy encoding unit that entropy encodes the transform coefficient input from the quantization unit, and encoded data output from the entropy encoding unit. As the body volume of the target image quality, characterized in that it comprises a code amount control unit for performing rate control of truncating a portion of the encoded data.

  A compression encoding apparatus according to a third aspect of the present invention is a compression encoding apparatus for compressing and encoding an image signal, and recursively band-divides the image signal into a high-frequency component and a low-frequency component by wavelet transformation. A wavelet transform unit that generates and outputs transform coefficients of band components of the image, a region of interest setting unit that sets a region of interest in the image signal, a quantization unit that quantizes the transform coefficient, and the quantization unit An entropy encoding unit that entropy-encodes the transform coefficient, and a code string that is subjected to rearrangement and bit shift based on the setting information of the region of interest for the encoded data output from the entropy encoding unit And a code amount control unit that performs rate control for truncating a part of the code string so that the entire capacity of the encoded data becomes a target image quality.

  According to a fourth aspect of the present invention, there is provided a compression coding apparatus according to the third aspect, wherein the quantization parameter indicating the target image quality is divided by the norm of the filter coefficient to obtain a quantization step size. A control unit, wherein the quantization unit quantizes the transform coefficient based on the quantization step size, and the code amount control unit arranges the encoded data based on the quantization step size. It is characterized by changing.

  A compression coding apparatus according to a fifth invention is the compression coding apparatus according to any one of the second to fourth inventions, wherein the region-of-interest setting unit gives priority to each set region of interest, The shift amount of the bit shift based on the setting information of the region of interest is determined according to the priority.

  A compression encoding apparatus according to a sixth aspect of the present invention is the compression encoding apparatus according to any one of the first to fifth aspects, wherein the code amount control unit determines a truncation target for the rate control in bit plane units. It is characterized by that.

  A compression coding apparatus according to a seventh invention is the compression coding apparatus according to any one of the first to fifth inventions, wherein the code amount control unit determines the rate control truncation target in a path unit. It is characterized by.

  According to an eighth aspect of the present invention, there is provided a compression coding apparatus according to any one of the first, second, fourth to seventh aspects of the invention, wherein the image quality control unit includes a norm of filter coefficients and human vision. The quantization step size obtained by multiplying the specified quantization parameter by a value obtained by multiplying energy weighting facotor, which is a predetermined numerical value determined based on characteristics, and weighting in consideration of human visual characteristics It is characterized by calculating | requiring.

  A compression encoding apparatus according to a ninth aspect is the compression encoding apparatus according to any one of the first, second, fourth to eighth aspects, wherein the image quality control unit has a predetermined quantization step size. When smaller than a numerical value, the quantization step size is a value obtained by multiplying a power of 2 at which the quantization step size is equal to or greater than the predetermined numerical value.

  A compression encoding apparatus according to a tenth aspect of the invention is the compression encoding apparatus according to the ninth aspect of the invention, wherein when the quantization step size is a value obtained by multiplying a power of 2 in the image quality control unit, When performing the rearrangement based on the quantization step size, the transform coefficient or the encoded data is bit-shifted by the number of bits corresponding to the exponent of the power of 2.

  A compression encoding apparatus according to an eleventh aspect of the invention is the compression encoding apparatus according to the first aspect of the invention, wherein the code amount control unit can change the viewing distance, and the code string is changed according to the change of the viewing distance. It is characterized by bit shifting.

  A compression encoding apparatus according to a twelfth aspect of the invention is the compression encoding apparatus according to the second aspect of the invention, wherein the quantization unit can change the viewing distance, and further converts the transform coefficient according to the change in the viewing distance. It is characterized by bit shifting.

  A compression coding apparatus according to a thirteenth invention is the compression coding apparatus according to the third invention, wherein the code amount control unit can change a viewing distance, and the code string is changed according to a change in the viewing distance. Further, bit shift is performed.

  A compression encoding method according to a fourteenth aspect of the present invention is a compression encoding method for compressing and encoding an image signal, and (a) the image signal is recursively divided into a high frequency component and a low frequency component by wavelet transform. A step of generating transform coefficients of a plurality of band components; (b) a step of obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient; and (c) the transform coefficient And (d) rearranging the encoded data encoded in the step (c) in a predetermined scanning order based on the quantization step size to generate a code string, And a step of performing rate control for truncating a part of the code string so that the entire capacity of the encoded data becomes a target image quality.

  A compression encoding method according to a fifteenth aspect of the present invention is a compression encoding method for compressing and encoding an image signal. (A) The image signal is recursively divided into a high frequency component and a low frequency component by wavelet transformation. A step of generating transform coefficients of a plurality of band components; (b) a step of obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient; and (c) the image signal. A region of interest is set; and (d) quantization and rearrangement based on the quantization step size and rearrangement and bit shift based on the region of interest setting information are performed on the transform coefficient. A step, (e) a step of entropy encoding the transform coefficient subjected to the processing of step (d), and (f) an overall capacity of the encoded data encoded in step (e) becomes the target image quality. And performing rate control of truncating a portion of sea urchin the encoded data, characterized in that it comprises a.

  A compression encoding method according to a sixteenth aspect of the present invention is a compression encoding method for compressing and encoding an image signal, wherein (a) the image signal is recursively divided into a high frequency component and a low frequency component by wavelet transformation. Generating a plurality of band component transform coefficients, (b) setting a region of interest in the image signal, (c) quantizing the transform coefficients, and (d) the process ( c) entropy-encoding the transform coefficients quantized in (c), and (e) performing reordering and bit shifting based on the setting information of the region of interest on the encoded data encoded in the step (d). And a rate control step of cutting off a part of the code sequence so that the entire capacity of the encoded data becomes a target image quality.

  A compression encoding method according to a seventeenth aspect of the invention is the compression encoding method according to the sixteenth aspect of the invention, wherein the step of performing the quantization includes dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient to quantize the quantization parameter. A step of obtaining a quantization step size, and a step of quantizing the transform coefficient based on the quantization step size, wherein the step of rearranging the encoded data includes the encoded data based on the quantization step size The step of rearranging is included.

  A compression encoding method according to an eighteenth aspect of the present invention is the compression encoding method according to any one of the fifteenth to seventeenth aspects, wherein the step of setting the region of interest gives a priority to each set region of interest. A step of performing a bit shift based on the setting information of the region of interest includes a step of performing a bit shift by a predetermined number of bits determined in accordance with the priority.

  A compression encoding method according to a nineteenth aspect of the present invention is the compression encoding method according to any one of the fourteenth to eighteenth aspects of the present invention, wherein the step of performing rate control determines a truncation target for the rate control in bit plane units. Including a step of performing.

  A compression encoding method according to a twentieth aspect of the present invention is the compression encoding method according to any one of the fourteenth to eighteenth aspects of the invention, wherein the step of performing rate control determines a truncation target for the rate control in units of paths. Including a process.

  A compression coding method according to a twenty-first invention is the compression coding method according to any one of the fourteenth, fifteenth, seventeenth to twentieth inventions, wherein the step of obtaining the quantization step size is a norm of a filter coefficient. And a value obtained by multiplying energy weighting facotor, which is a predetermined numerical value determined based on human visual characteristics, by dividing the specified quantization parameter and weighting in consideration of human visual characteristics A step of obtaining a quantization step size.

  A compression encoding method according to a twenty-second aspect of the invention is the compression encoding method according to any of the fourteenth, fifteenth, seventeenth to twenty-first aspects, wherein the step of obtaining the quantization step size comprises the quantization step. When the size is smaller than a predetermined numerical value, including a step of setting the quantization step size to a value obtained by multiplying a power of 2 at which the quantization step size is equal to or larger than the predetermined numerical value. To do.

  A compression encoding method according to a twenty-third aspect is the compression encoding method according to the twenty-second aspect, wherein the step of rearranging based on the quantization step size multiplies the quantization step size by a power of 2. The transform coefficient quantized with the quantization step size or the encoded data is bit-shifted by the number of bits corresponding to the exponent of the power of 2, and the code string is And a generating step.

  A program according to a twenty-fourth invention is a program for causing a microprocessor to compress and encode an image signal, and recursively divides the image signal into a high frequency component and a low frequency component by wavelet transform, and A wavelet transform unit that generates and outputs transform coefficients of band components, an image quality control unit that obtains a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient, and entropy-encodes the transform coefficients The encoded data output from the entropy encoding unit and the entropy encoding unit are rearranged in a predetermined scanning order based on the quantization step size to generate a code sequence, and the encoded data As a code amount control unit that performs rate control to discard a part of the code string so that the overall capacity becomes the target image quality, Characterized in that the functioning of the serial microprocessor.

  A program according to a twenty-fifth aspect of the invention is a program for causing a microprocessor to compress and encode an image signal, and recursively divides the image signal into a high frequency component and a low frequency component by wavelet transformation, and A wavelet transform unit that generates and outputs transform coefficients of band components, an image quality control unit that obtains a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient, and sets a region of interest in the image signal A region of interest setting unit, a quantization unit that performs quantization and rearrangement based on the quantization step size and rearrangement and bit shift based on the setting information of the region of interest, with respect to the transform coefficient, An entropy encoding unit for entropy encoding the transform coefficient input from the quantization unit, and an entropy encoding unit. So that the entire capacity of the encoded data force becomes equal to the target quality, the code amount control unit for performing rate control of truncating a portion of the encoded data, characterized in that to function the microprocessor.

  A program according to a twenty-sixth aspect of the invention is a program for causing a microprocessor to compress and encode an image signal, and recursively divides the image signal into a high frequency component and a low frequency component by wavelet transformation, and A wavelet transform unit that generates and outputs transform coefficients of band components, a region of interest setting unit that sets a region of interest in the image signal, a quantization unit that quantizes the transform coefficient, and the quantization unit An entropy encoding unit that entropy-encodes the transform coefficient, and a code string that is subjected to rearrangement and bit shift based on the setting information of the region of interest for the encoded data output from the entropy encoding unit As a code amount control unit that performs rate control for truncating a part of the code string so that the entire capacity of the encoded data becomes a target image quality, Characterized in that the functioning of the serial microprocessor.

  A program according to a twenty-seventh aspect is the program according to the twenty-sixth aspect, wherein the micro-processing unit further includes an image quality control unit that obtains a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient. When the microprocessor functions as the quantization unit, the microprocessor is caused to function so as to quantize the transform coefficient based on the quantization step size, and the microprocessor controls the code amount control unit. , The encoded data is rearranged based on the quantization step size.

  A program according to a twenty-eighth aspect is the program according to any one of the twenty-fifth to twenty-seventh aspects, wherein priority is given to each set region of interest when the microprocessor functions as the region-of-interest setting unit. When the bit shift based on the setting information of the region of interest is performed by the microprocessor, the bit shift is performed by a predetermined number of bits determined according to the priority. And

  A program according to a twenty-ninth invention is the program according to any one of the twenty-fourth to the twenty-eighth inventions, wherein when the microprocessor functions as the code amount control unit, the rate control truncation target is a bit plane unit. It is characterized by functioning to determine.

  A program according to a thirtieth aspect is the program according to any one of the twenty-fourth to the twenty-eighth aspects, wherein when the microprocessor functions as the code amount control unit, the rate control truncation target is determined on a path basis. It is made to function like this.

  A program according to a thirty-first invention is the program according to any of the twenty-fourth, twenty-fifth, twenty-seventh to thirtieth inventions, wherein when the microprocessor functions as the image quality control unit, a norm of a filter coefficient and a human Quantization obtained by dividing the specified quantization parameter by a value obtained by multiplying a predetermined numerical value determined based on the visual characteristics of the image by the weighting factor, and weighting in consideration of human visual characteristics It is characterized by functioning to obtain the step size.

  A program according to a thirty-second invention is the program according to any of the twenty-fourth, twenty-fifth, twenty-seventh to thirty-first inventions, wherein when the microprocessor functions as the image quality control unit, the quantization step size is When smaller than a predetermined numerical value, the quantization step size is caused to function as a value obtained by multiplying a power of 2 at which the quantization step size is greater than or equal to the predetermined numerical value.

  A program according to a thirty-third invention is the program according to the thirty-second invention, wherein the quantization step size is a power of two in the image quality control unit when the microprocessor performs rearrangement based on the quantization step size. When the value is obtained by multiplying, the conversion coefficient or the encoded data is made to function so as to be bit-shifted by the number of bits corresponding to the exponent of the power of 2.

  According to the first, fourteenth and twenty-fourth inventions, it is possible to obtain a noise removal effect on the entire screen with a small amount of calculation while controlling the compression rate of data by quantization according to the target image quality.

  According to the second, fifteenth and twenty-fifth inventions, it is possible to easily generate a compressed image suitable for the target image quality having a noise removal effect on the region of interest.

  According to the third, sixteenth and twenty-sixth aspects, it is possible to easily generate a compressed image suitable for the target image quality having a noise removing effect on the region of interest. The processing can also be realized in a compression encoding apparatus that cannot use the Max-Shift method, which is an optional function of JPEG2000.

  According to the fourth, seventeenth, and twenty-seventh inventions, the quantum compression speed is reduced with a small amount of computation compared to the conventional technique that requires processing for estimating the optimum solution while controlling the compression rate of the data by quantization according to the target image quality. Can be performed.

  According to the fifth, eighteenth and twenty-eighth aspects, by setting the priority to each of the plurality of regions of interest, it is possible to generate a compressed image that matches the target image quality while maintaining the image quality of the desired region. it can.

  According to the sixth, nineteenth and twenty-ninth aspects, the code amount can be finely and efficiently rate-controlled in bit plane units according to the target image quality.

  According to the seventh, twentieth and thirtieth inventions, it is possible to finely and efficiently rate control the code amount in units of paths in accordance with the target image quality.

  According to the eighth, twenty-first, and thirty-first aspects, it is possible to easily generate a compressed image having a high display image quality suitable for human visual evaluation.

  According to the ninth, twenty-second, and thirty-second inventions, an apparatus that efficiently quantizes according to the target image quality can be easily realized.

  According to the tenth, twenty-third and thirty-third inventions, a compressed image suitable for the target image quality can be efficiently generated while maintaining the image quality.

  According to the eleventh aspect of the present invention, it is possible to control so that an image encoded with a compression quantization value also has a target image quality with a noise removal effect.

  According to the twelfth aspect, it is possible to control so that an image encoded with a compression quantization value also has a target image quality with a noise removal effect.

  According to the thirteenth aspect, it is possible to control so that an image encoded with a compression quantization value also has a target image quality with a noise removal effect.

(First embodiment)
{Compression encoder}
FIG. 1 is a functional block diagram showing a schematic configuration of a compression coding apparatus 1 according to an embodiment of the present invention. After an overview of the configuration and functions of the compression encoding apparatus 1, the quantization method and encoding method according to the present embodiment will be described in detail.

  The compression coding apparatus 1 includes a DC level shift unit 10, a color space conversion unit 11, a tiling unit 12, a DWT unit 13, a quantization unit 14, a coefficient bit modeling unit 20, an arithmetic coding unit (entropy coding unit). 21, a code amount control unit 22, an image quality control unit 23, and a bit stream generation unit 17.

  Note that all or part of each of the processing units 10 to 14, 17, and 20 to 23 constituting the compression encoding apparatus 1 may be configured by hardware or a program that causes a microprocessor to function. May be.

  The image signal input to the compression encoding device 1 is subjected to DC level conversion as necessary by the DC level shift unit 10 and then output to the color space conversion unit 11. The color space conversion unit 11 performs color space conversion on the input signal and outputs it. According to the JPEG2000 system, RCT (Reversible Component Transformation) for reversible conversion and ICT (Irreversible Component Transformation) for irreversible conversion are prepared as color space conversion, and one of them can be selected as appropriate. Thereby, for example, the input RGB signal is converted into a YCbCr signal or a YUV signal.

  Next, the tiling unit 12 divides the image signal input from the color space conversion unit 11 into a plurality of region components called “tiles” having a rectangular shape, and outputs the region components to the DWT unit 13. It is not always necessary to divide the image signal into tiles, and the image signal for one frame may be output to the next functional block as it is.

  Next, the DWT unit 13 performs integer type or real type DWT on the image signal input from the tiling unit 12 in units of tiles, thereby converting the image signal into a high frequency component and a low frequency component according to the octave division method. And band recursively. As a result, transform coefficients of a plurality of band components (subbands) HH 1 to LL 3 as shown in FIG. 24 are generated and output to the quantization unit 14. Specifically, for real type DWT, a filter of 9 × 7 tap, 5 × 3 tap or 7 × 5 type is used, and for integer type DWT, 5 × 3 tap or 13 × 7 tap, etc. The filter is used. Further, the processing of these filters may be executed by a convolution operation, or may be executed by a lifting scheme that is more efficient than the convolution operation.

  The quantization unit 14 has a function of performing scalar quantization on the transform coefficient input from the DWT unit 13 according to the quantization parameter determined by the image quality control unit 23. The quantization unit 14 also has a function of performing a predetermined bit shift process. A method of quantization and bit shift processing by the quantization unit 14 will be described later.

  Next, the transform coefficient QD output from the quantization unit 14 is subjected to block-based entropy coding by the coefficient bit modeling unit 20 and the arithmetic coding unit 21, and the rate is controlled by the code amount control unit 22. The

Similar to the coefficient bit modeling unit 108 shown in FIG. 23, the coefficient bit modeling unit 20 divides the band component of the input transform coefficient QD into code blocks of about 32 × 32 or 64 × 64, and each code block Are broken down into a plurality of bit planes configured by two-dimensionally arranging each bit. As a result, each code block is decomposed into a plurality of bit planes 122 _{0 to} 122 _n-1 as shown in FIG.

  Next, the arithmetic encoding unit 21 arithmetically encodes the encoded data BD input from the coefficient bit modeling unit 20 and outputs the encoded data AD obtained as a result to the code amount control unit 22. Here, the arithmetic encoding unit 21 may perform bypass processing in which a part of the encoding target is not arithmetically encoded and the encoding target is included in the encoded data AD and output as it is. In this embodiment, arithmetic coding is adopted, but the present invention is not limited to this, and other types of entropy coding may be adopted.

  Next, the code amount control unit 22 has a function of controlling the rate of the encoded data AD input from the arithmetic encoding unit 21 based on an instruction from the image quality control unit 23. In other words, the code amount control unit 22 considers the target code amount (the code amount of the final compressed image), and obtains a desired noise removal effect (skin-beautifying effect in human photography) for the entire screen. It has a function of performing post-quantization in which the quantized data AD is discarded in order from the lowest priority in band component units, bit plane units, and path units. A method of rate control in the code amount control unit 22 will be described later.

  Then, the bit stream generation unit 17 generates a bit stream in which the encoded data CD output from the code amount control unit 22 and additional information (header information, layer configuration, scalability, quantization table, etc.) are multiplexed, Output to the outside as a compressed image.

{Quantization}
Next, the contents of the quantization process realized by the image quality control unit 23 and the quantization unit 14 shown in FIG. 1 will be described.

The image quality control unit 23 quantizes the transform coefficient input from the DWT unit 13 by the quantization unit 14 based on target image quality information (high quality, standard image quality, low image quality, resolution information, etc.) supplied from the outside. Has a function of determining the quantization step size Δ _b of the. The following describes the method of determining the quantization step size delta _b.

As shown in FIG. 24, the DWT unit 13 converts the original image into sub-bands (band components) of “XYn” (X and Y are either the high frequency component H or the low frequency component L. n is the decomposition level). When divided, the quantization step size Δ _b used for quantization of each subband is set as in the following equation (1).

Here, Q _p is a positive number input according to the target image quality information, that is, a quantization parameter, and a smaller value is input as the image quality is higher. The quantization parameter Q _p may be specified by a user by directly inputting a numerical value. For example, a predetermined instruction word such as “high image quality, standard image quality, low image quality” indicating target image quality information may be used in advance. And a predetermined table in which the numerical value of the quantization parameter Q _p is associated, and the target image quality after compression of the image data desired by the user is designated by an instruction word, whereby the quantization associated in the table It may be an aspect in which the value of the parameter Q _p is read and used.

Q _b is a quantization coefficient in each subband, and is expressed by the following equation (2) as a norm of the synthesis filter coefficient.

Here, the weighting coefficient Gb of the subband _b is calculated according to the following equation (3).

In the above equation (3), s _b [n] indicates a one-dimensional synthesis filter coefficient of subband b. The symbol ‖x‖ indicates a norm related to the vector x.

According to the equations (4.39) and (4.40) described in Non-Patent Document 1 described above, the one-dimensional synthesis filter coefficient s _{L [1]} [n] of the low-frequency component L1 at the decomposition level 1 The one-dimensional synthesis filter coefficient s _{H [1]} [n] of the high frequency component H1 at the same decomposition level is calculated according to the following equation (4).

Here, in the above equation (4), g ₀ [n] represents a low-pass filter coefficient of a forward conversion filter for dividing an image signal into bands, and g ₁ [n] represents a high-pass filter coefficient thereof.

Further, the one-dimensional synthesis filter coefficient s _{L [d]} [n] of the low-frequency component Ld at the decomposition level d (d = 1, 2,..., D) and the one-dimensional synthesis filter of the high-frequency component Hd at the same decomposition level. The coefficient s _{H [d]} [n] is calculated according to the following equation (5).

  Then, the square of the norm of the one-dimensional synthesis filter coefficient of the low frequency component Ld at the decomposition level d is calculated according to the following equation (6).

  The norm square of the one-dimensional synthesis filter coefficient of the high-frequency component Hd can also be calculated in the same manner as in the above equation (6).

Table 1 shows the calculation result of the square of the norm of the one-dimensional synthesis filter coefficient. In the table, n indicates a decomposition level. For example, GL ₁ indicates a calculation result at a decomposition level 1 of the low-frequency component L.

Next, the two-dimensional synthesis filter coefficients of the band components LLD, HLd, LHd, and HHd at the decomposition level d (d = 1, 2,..., D; D is an integer) are expressed by the product of the one-dimensional synthesis filter coefficients. The two-dimensional weight coefficient G _{b of} the band component b can also be expressed by a product of the one-dimensional weight coefficient. Specifically, the two-dimensional synthesis filter coefficient and the two-dimensional weighting coefficient are calculated according to the following equation (7).

  In the above formula (7), the subscript LL [D] indicates the subband LLD, and HL [d], LH [d], and HH [d] indicate the subbands HLd, LHd, and HHd, respectively.

Square root of the weighting factor G _b is the norm. Tables 2 and 3 below show the calculation results regarding the two-dimensional weighting coefficient G _b obtained from Table 1. Table 2 shows the norm square value of each band component of the (9, 7) filter (9 × 7 tap filter), and Table 3 shows the norm value corresponding to Table 2.

For example, for all of the luminance signal Y and the color difference signals U and V, the luminance signal Y obtained using the above equations (1) and (2) from the values shown in Table 3 with the quantization parameter Q _p = 16. , the quantization step size delta _b of the color difference signals U and V are as in Table 4.

Note that the quantization parameter Q _p used to determine the quantization step size Δ _b for each of the luminance signal Y and the color difference signals U and V does not necessarily have the same value, and depends on the content of the image data. Different values may be used. For example, when it is desired to enhance color components, the quantization parameter Q _p used for the color difference signals U and V is made smaller than the luminance signal Y. The initialization parameter Q _p may be used.

The image quality control unit 23, thus to determine the quantization step size delta _b, the and notifies the quantization unit 14. Then, the quantization unit 14, for each subband, performs quantization in accordance with the notified quantization step size delta _b.

However, if the value of the quantization step size delta _b is less than 1 is used after multiplying the power of 2 so that the value of 1 or more. For example, results of calculation in the manner described above, but the quantization step size delta _b subband LL5 obtained is 0.47163, when quantizing the actual image data is multiplied by 2 ² This value Quantization is performed with a quantization step size Δ _b = 1.88652. Similarly, in subband HL5, quantization step size Δ _b = 0.93204 is multiplied by 2 to perform quantization with quantization step size Δ _b = 1.886408. Thus the quantization step size delta _b, that it has a function of converting into a predetermined number by the performance of the quantizer to achieve the quantization, it is possible to simplify the structure of the quantizer It is also possible to achieve compression of the amount of data that is the original purpose of quantization. Note that the quantization step size Δ _b is set to a value of 1 or more. For example, if the quantizer uses a value of 1/2 or more by the function of the quantizer, the quantization step size is not limited. What is necessary is just to convert (DELTA) _b so that it may become 1/2 or more. In other words, if the lower limit value handled by the quantizer is ½ ^m , the quantization step size Δ _b may be used after being multiplied by a power of 2 so that it becomes ½ ^m or more. That's fine.

Further, the image quality control unit 23, in addition to the method described above, in consideration of human visual characteristics can also be determined quantization step size delta _b. The method is as follows.

  In Chapter 16 of Non-Patent Document 1 described above, weighted mean squared error (WMSE) based on CSF (Contrast Sensitivity Function of human visual system) is described. Using this, the above equation (2) is corrected to the following equation (8) in order to improve human visual evaluation of the image data after compression coding.

Here, in the above equation (8), W _{b [i]} ^csf is called “energy weighting factor” of subband b [i], and the recommended value of W _{b [i]} ^csf is “INTERNATIONAL STANDARD ISO / IEC 15444-1 ITU-T RECOMMENDATION T.800 Information technology-JPEG 2000 image coding system: Core coding system "(hereinafter referred to as non-patent document 2). 2 to 4 show numerical values of “energy weighting factor” described in Non-Patent Document 2. FIG.

  2 to 4, “level” and “Lev” indicate the decomposition level, “Comp” indicates the luminance component Y and the color difference components Cb and Cr, respectively, and “Viewing distance” is 1000 and 1700. , 2000, 3000, 4000 examples are shown. "Viewing distance 1000", "Viewing distance 1700", "Viewing distance 2000", "Viewing distance 3000", and "Viewing distance 4000" are 10 displays or printed materials of 100 dpi, 170 dpi, 200 dpi, 300 dpi, and 400 dpi, respectively. The viewing distance when viewed from an inch away.

For example, the image data of a color, a specific method for obtaining the quantization step size delta _b below. As for the color space, it is assumed that a color input image composed of RGB signals is converted into color space data in the YUV422 or YUV420 format by the color space conversion unit 11.

  For image data in the YUV422 or YUV420 format, the color difference signals U and V have a data amount of 1/2 and 1/4, respectively, compared to the luminance signal Y. The wavelet plane obtained by applying DWT to the luminance signal Y can be represented as shown in FIG. 5, but the data amount is ½ that the DWT is performed once in the horizontal direction with respect to the wavelet plane shown in FIG. Assuming that it is equivalent to the applied one, the scattered dot portions in FIG. 6 are the wavelet planes of the color difference signals U and V in the YUV422 format. Similarly, if it is assumed that the data amount is 1/4, it is equivalent to the case where the DWT is performed once in the horizontal direction and the vertical direction on the wavelet plane shown in FIG. Are the wavelet planes of the color difference signals U and V in the YUV420 format.

  In the YUV422 format, as shown in FIG. 6, it is assumed that the horizontal component is filtered once more than the vertical component. Therefore, the two-dimensional synthesis filter coefficient and the two-dimensional weight coefficient are expressed by the following equation (7) as follows: It can be expressed as (9).

  Similarly, in the YUV420 format, as shown in FIG. 7, it is assumed that the horizontal component and the vertical component are filtered one by one, so the above equation (7) is expressed as the following equation (10). be able to.

  Therefore, when the norms of the color difference signals in the YUV422 and YUV420 formats are obtained from the values shown in Table 1 using the above formulas (9) and (10), Tables 5 and 6 are obtained.

Next, the _{energy weighting factor W b [i]} csf , according to the description of Non-Patent Document 1, the _{energy weighting factor W b [i]} csf one-dimensional horizontal and vertical subband b [i] each The product of the band component energy weighting factors is expressed by the following equation (11).

  The energy weighting factor related to the luminance signal Y in the YUV422 or YUV420 format image data can be obtained by the above equation (11). In the case of the YUV444 format, both the luminance signal and the color difference signal are obtained by the above equation (11).

  As for the color difference signals U and V in the YUV422 format, it is assumed that the horizontal component is filtered once more than the vertical component as described above. Therefore, the energy weighting factor is expressed by the following equation (12) with respect to the above equation (11). ).

  Similarly, for the color difference signals U and V in the YUV420 format, it is assumed that the horizontal component and the vertical component are filtered one by one, so that the energy weighting factor is expressed by the following equation with respect to the above equation (11). It can be expressed as (13).

  Tables 7 to 9 show values of energy weighting factors of the color difference signals U and V of the viewing distance 1000, the viewing distance 1700, and the viewing distance 3000 obtained from the description of Non-Patent Document 2. Here, including the following table, Cb and Cr in the table indicate the color difference signals U and V, respectively.

  The energy weighting factors relating to the image data in the YUV422 and YUV420 formats obtained from the values shown in Tables 7 to 9 using the above formulas (11) to (13) are shown in Tables 10 to 12 and Tables 13 to 13. It is Table 15.

By substituting the norm values of Tables 5 and 6 obtained in this way into the above equations (1) and (2), the normal quantization step size Δ _b is obtained as the norm values of Tables 5 and 6; By substituting the values of energy weighting factors in Tables 10 to 15 into the above formulas (1) and (8), the quantization step size Δ _b obtained by visual weighting in consideration of human visual characteristics can be obtained. .

For example, with respect to all of the luminance signal Y and the color difference signals U and V, the luminance obtained when visual weighting of a viewing distance of 3000 is performed on color image data in the YUV422 format with the quantization parameter Q _p = 16. signal Y, a quantization step size delta _b of the color difference signals U and V, the value of the norm shown in Table 5, the value of the energy weighting factor shown in Table 12, and the above equation (1), it is determined using (8) . The results are shown in Tables 16-18.

Note that the quantization parameter Q _p used to determine the quantization step size Δ _b for each of the luminance signal Y and the color difference signals U and V does not necessarily have the same value, and depends on the content of the image data. Different values may be used. For example, when it is desired to enhance color components, the quantization parameter Q _p used for the color difference signals U and V is made smaller than the luminance signal Y. The initialization parameter Q _p may be used. Similarly, different values of energy weighting factor may or may not be used.

The image quality control unit 23, thus to determine the quantization step size delta _b, the and notifies the quantization unit 14. Then, the quantization unit 14, for each subband, performs quantization in accordance with the notified quantization step size delta _b. In this case, smaller than 1 the quantization step size delta _b, to use that after one or more values by multiplying the power of 2, is the same as described above.

  As described above, in this embodiment, by controlling the image quality by quantization, it is possible to perform strict control according to the target image quality having a noise removal effect (skin-beautifying effect). At this time, since complicated processing for obtaining an optimal solution is not required, high-speed processing can be performed with a small amount of calculation. Further, in consideration of human visual characteristics, a compressed image having high display image quality after compression can be generated.

{Sort and bit shift}
Quantization unit 14 shown in FIG. 1, based on the quantization step size delta _b, to sort and bit shift processing of the quantized data.

In favor of the data subjected to the quantization with a smaller quantization step size delta _b allocate more code amount by performing a bit shift processing so as not to be affected by truncation by the rate control, according to the coding process Image quality degradation can be avoided or suppressed.

The following describes the processing of sorting and bit shifting based on the quantization step size delta _b.

First, when a value of a predetermined quantization parameter Q _p is designated as the target image quality, the image quality control unit 23 calculates the quantization step size Δ _b as described above based on this value, and quantizes this Notification to the unit 14.

At the quantization unit 14, the quantization step size delta _b is notified, based on this value, DWT unit 13 is quantized as described above the image data after the DWT.

Then, the quantization unit 14, the data after quantization, according to the magnitude of the quantization step size delta _b, rearranged in the order (ascending order) the value is small.

As described above, the data quantized using the quantization step size Δ _b converted to be 1 or more is rearranged based on the quantization step size Δ _b before conversion. It performs a process of bit shift to the left data by the number of bits corresponding to the exponent of 2 multiplied when converting the quantization step size delta _b. Specific processing modes are as follows.

For example, in Table 4, but the quantization step size delta _b subband LL5 is 0.47163, when quantizing the actual image data, the quantization step to 1.88652 multiplied by the 2 ² This value Quantization is performed with the size Δ _b . Therefore, data of the sub-band LL5, corresponding to 2 ² of index multiplied for conversion of the quantization step size delta _b, it is shifted left two bits. Similarly, in subband HL5, quantization step size Δ _b = 0,93204 is multiplied by 2 to perform quantization with quantization step size Δ _b = 1.886408. Therefore, the encoded data AD of the subband HL5 is shifted to the left by 1 bit corresponding to the exponent of 2 multiplied. That is, when the quantized by multiplied by 2 ^m quantization step size delta _b, by left-shifting the relevant data by the amount of the index m, is to adjust the priority of the data.

Based on the quantization step size delta _b shown in Table 4, showing the transformation coefficient which has been subjected to processing such bit shift in FIG. Figure, * code string lay the indicia, the value of the quantization step size delta _b indicates a transformation during quantization, numbers 0 and lay to each bit of the code sequence, ..., 9, the bit Indicates the bit plane number to which. Here, LSB number = 0 and MSB number = 9.

Then, each transform coefficient, sorted in ascending order of the quantization step size delta _b used for quantization (in ascending order). In Figure 8, the value of the portion of the quantization step size delta _b indicated by the arrow is not the ascending order, replace them. FIG. 9 shows a code string that has been rearranged in this way. The arrows in FIG. 9 indicate code strings whose positions have been changed from those in FIG.

Further, the same processing is performed when calculating the quantization step size delta _b consider the case and visual weighting of the color image.

For example, as described above, with the quantization parameter Q _p = 16, the luminance signal Y, the color difference signals U and V when the visual weighting of the viewing distance 3000 is performed on the color image data in the YUV422 format. The quantization step size [Delta] _b is as shown in Tables 16-18.

Here, what quantization step size delta _b in Table 16 Table 18 is smaller than 1 is used in quantization on multiplied by the powers of 2 as described above. Then, the data quantized with the converted quantization step size Δ _b is left-shifted by the number of bits corresponding to the exponent of the power of 2 multiplied by the original quantization step size Δ _b .

For color images, as the luminance signal Y, but the data for each of the color difference signals U and V are present, these data also without distinction for each signal, all data is the quantization step size delta _b becomes ascending Sort. The conversion coefficient obtained as a result is shown in FIG. In the figure, YLL5 indicates data of the subband LL5 of the luminance signal Y.

  In this way, the above-described bit shift and rearrangement processing is performed on all data of the luminance signal Y and the color difference signals U and V.

  The transform coefficient QD output from the quantization unit 14 is subjected to block-based entropy coding by the coefficient bit modeling unit 20 and the arithmetic coding unit 21, and the rate is controlled by the code amount control unit 22.

{Code amount control}
The code amount control unit 22 performs rate control of the encoded data AD processed by the coefficient bit modeling unit 20 and the arithmetic encoding unit 21 after the quantization unit 14 performs quantization.

  The code amount control unit 22 performs data truncation so that a desired noise removal effect is obtained with respect to a code string that has been rearranged and bit-shifted as shown in FIG. Data truncation is performed in order from the rightmost bit. For example, the bit data of number 0 of VHL4 shown in FIG. 10 are sequentially deleted in the order of bit data of number 0 of YHH5. Then, if the target noise removal effect can be obtained by truncating the bit data up to YHH1, the data at the corresponding dot portion in FIG. 10 is discarded. At this time, if the desired noise removal effect cannot be obtained even if the data of YHH1 is rounded down, the bit data of VLL4 number 0 is moved downward, the bit data of ULL4 number 0, and the number 0 of YLH5 Delete bit data ... in order.

  The bit data (bit plane) can be decomposed into SIG path, MR path, and CL path as shown in FIG. The rate control by the code amount control unit 22 may be performed in units of paths in addition to the mode performed in units of bit planes as described above.

  In this case, the code amount control unit 22 sequentially deletes the CL path from the CL path with the number 0 of VHL4 in the order of the CL path with the number 0 of YHH5. If the desired noise removal effect cannot be obtained even if the CL path of number 0 of YHH1 is deleted, then return to VHL4, MR path of number 0 of VHL4, MR path of number 0 of YHH5 downward ..., MR path of YHH1 with number 0 is deleted in order. The bit data is continuously deleted until a desired noise removal effect is obtained.

  If the target image quality intended by the user has already been obtained at the stage of input to the code amount control unit 22, the above-described rate control need not be performed.

As described above, according to the compression coding apparatus 1 according to the present embodiment, the rearrangement and the bit shift are performed according to the value of the quantization step size Δ _b , and the bit data or path of each subband is subjected to desired noise removal. Until the effect is obtained (that is, until the desired target image quality is obtained), rate control is performed by deleting from the lower bits. Thereby, a desired noise removal effect can be acquired with respect to the whole screen.

  Moreover, since it is not necessary to calculate the amount of distortion in each coding pass for rate / distortion optimization processing as in the past, it is possible to realize high-efficiency rate control with high real-time performance and significantly reduced overhead. .

  Note that the rearrangement and the bit shift processing based on the quantization step size do not necessarily have to be executed by the quantization unit 14, and may be executed by the code amount control unit 22, for example. In this case, the code amount control unit 22 rearranges and bit-shifts the encoded data AD based on the quantization step size Δb, and then truncates the data. In this case, even in a compression coding apparatus in which the quantization unit 14 does not have a function related to bit shift, the present embodiment can be realized only by changing the function and operation of the code amount control unit 22.

(Second Embodiment)
{Compression encoder}
FIG. 11 is a functional block diagram showing a schematic configuration of the compression encoding apparatus 1 according to the second embodiment of the present invention. An ROI unit 15 is added to the compression encoding apparatus 1 according to the first embodiment shown in FIG.

  Note that all or part of each of the processing units 10 to 15, 17, and 20 to 23 constituting the compression encoding apparatus 1 may be configured by hardware or a program that causes a microprocessor to function. May be.

  In the following, the configuration and operation of compression encoding in the present embodiment will be described focusing on differences from the first embodiment.

  The quantization unit 14 has a function of performing scalar quantization on the transform coefficient input from the DWT unit 13 according to the quantization parameter determined by the image quality control unit 23. The quantization unit 14 also has a function of performing bit shift processing for obtaining a noise removal effect on the region of interest set by the ROI unit 15 (hereinafter referred to as ROI (Region Of Interest) portion). Yes. The ROI part setting method by the ROI unit 15 and the bit shift processing method by the quantization unit 14 in consideration of the ROI part will be described later.

  The code amount control unit 22 has a function of controlling the rate of the encoded data AD input from the arithmetic encoding unit 21 based on an instruction from the image quality control unit 23. That is, the code amount control unit 22 considers the target code amount (the code amount of the final compressed image), and obtains the noise removal effect for the ROI portion by converting the encoded data AD into band component units or It has a function of performing post-quantization of truncating from the lowest priority in bit plane units and path units. A method of rate control in the code amount control unit 22 will be described later.

{Setting of ROI part}
The ROI unit 15 shown in FIG. 11 sets an area in the image data where a noise removal effect by rate control, which will be described later, is desired, as the ROI part.

  Below, the setting method of a ROI part is demonstrated.

  The ROI unit 15 gives a mask signal for designating an important part as an ROI part to the image data subjected to wavelet transform. For example, in the image data (original image) 30 of a person as shown in FIG. 12, when only the face part below the amount for which a skin beautifying effect is desired is set as the ROI part, as shown in the white part in FIG. A single mask area 31 is set and designated as the ROI portion.

  The mask area 31 can be designated corresponding to the original image 30 by using a pointing input device such as a mouse while viewing the original image 30 on the screen of the display device. Alternatively, the mask area 31 may be automatically specified by analyzing image data, for example, extracting a predetermined area including a skin color portion.

  FIG. 13A shows an example in which a single ROI portion is designated for the original image 30. For example, as shown in a white portion in FIG. 13B, the left eye portion, the right eye portion, and the nose portion are shown. A plurality of areas such as a part including a mouth may be designated as the ROI part. In this case, these portions are defined by different mask signals.

  Then, the remaining portion obtained by removing the mask region 31 for all ROI portions becomes a non-ROI portion 32 (FIGS. 13A and 13B).

  Here, when a plurality of mask signals are given, priority can be given to the plurality of mask signals. The higher the priority, the lower the information amount, for example, the bit rate, and the higher the skin beautifying effect.

  Then, the mask area 31 is developed on the wavelet plane to generate a mask signal (FIG. 14).

  Here, the method of converting the mask signal into a portion corresponding to the wavelet plane depends on the number of taps of the wavelet transform filter.

  For example, as shown in FIG. 15, a reversible 5 × 3 filter (a filter in which the number of taps on the decomposition-side low-pass filter is 5 taps and the number of taps on the decomposition-side high-pass filter is 3 taps) Assuming that the even-numbered (2n-th) pixel data of the original image is designated as the ROI portion, the n-th data of the low-pass filter (low-frequency side) 33 and the high-pass filter (high-frequency filter) Side) 34, the (n-1) th and nth data are assumed to be ROI portions, and the mask signal is developed on the wavelet plane. When the odd-numbered (2n + 1) -th pixel data of the original image is designated as the ROI portion, the n-th and (n + 1) -th data of the low-pass filter (low-frequency side) 33 and the high-pass filter (high The (n−1) th, nth, and (n + 1) th data of the (region side) 34 is the ROI part, and the mask signal is developed on the wavelet plane. FIG. 15 shows only the correspondence between the original image and the wavelet plane of the first hierarchy, but the same recursive development is performed for the development of a deeper hierarchy.

  Alternatively, for example, as shown in FIG. 16, in the wavelet transform calculation process, a Deubesy 9 × 7 filter (a filter in which the number of taps in the decomposition-side low-pass filter is 9 taps and the number of taps in the decomposition-side high-pass filter is 7 taps) When the even-numbered (2n-th) pixel data of the original image is designated as the ROI part, the (n−1) -th and n-th of the low-pass filter (low-frequency side) 33 And the (n + 1) th data and the (n-2) th, (n-1) th, nth and (n + 1) th data of the high-pass filter (high frequency side) 34 are ROI parts, The mask signal is developed on the wavelet plane. When the odd-numbered (2n + 1) -th pixel data of the original image is designated as the ROI part, the (n−1) -th, n-th, (n + 1) -th and low-pass filters (low-frequency side) 33 The (n + 2) th data and the (n-2) th, (n-1) th, nth, (n + 1) th and (n + 2) th data of the high-pass filter (high frequency side) 34 are the ROI portion. As an example, the mask signal is developed on the wavelet plane. FIG. 16 shows only the correspondence between the original image and the wavelet plane of the first hierarchy, but the same recursive development is performed for the development of a deeper hierarchy.

  15 and FIG. 16, when the non-ROI part due to the correspondence with certain pixel data of the original image and the ROI part due to the correspondence with other pixel data of the original image overlap, The mask signal is developed on the wavelet plane as being the ROI portion. Further, the mask signal conversion method has been described for each of the case of a reversible 5 × 3 filter and the case of a debusy (Daubechies) 9 × 7 filter. A description is given taking a x7 filter as an example.

  A white portion 31a in FIG. 14A is a region where the mask region (ROI portion) 31 shown in FIG. 13A is developed on the wavelet plane as described above (hereinafter referred to as “development mask region”). A mask signal corresponding to the developed mask area 31a is generated and applied to the wavelet transformed image data. Reference numeral 32a in FIG. 14A indicates a region in which the non-ROI portion 32 is expanded on the wavelet plane (hereinafter referred to as “development non-mask region”). Further, in the portion where the mask region (ROI portion) 31 and the non-ROI portion 32 overlap, the mask region (ROI portion) 31 is selected and assigned. That is, a higher priority is selected and assigned.

  When a plurality of ROI portions are set as the mask region 31 as shown in FIG. 13B, there are a plurality of mask regions 31a developed on the wavelet plane as shown in FIG. 14B. Thus, when there are a plurality of mask regions 31a developed on the wavelet plane, priority can be given here. As described above, when a plurality of ROI portions 31 are set for the original image, a development mask region 31 a is generated on the wavelet plane for each ROI portion 31. All the development mask areas 31a corresponding to one ROI part 31 are given the same priority based on the priority of the ROI part 31, and finally the priority is set to all the development mask areas 31a. To do.

  When a plurality of ROI portions 31 are set for the original image, a plurality of development mask regions 31a may overlap in a portion that has passed through the low pass filter (low frequency side) 33 on the wavelet plane. In this case, the priority is determined for the overlapping portion as being the development mask region 31a having a higher priority among the plurality of overlapping development mask regions 31a.

  In the wavelet-transformed image data, a portion not applied to any mask signal becomes a developed non-mask area 32a. The priority of the development non-mask area 32a is lower than that of all the development mask areas 31a.

  In this way, it is possible to set a plurality of ROI parts and set the priority for each ROI part.

  Information on the set ROI portion is input to the quantization unit 14.

{Sort and bit shift}
The quantization unit 14 illustrated in FIG. 11 performs rearrangement of the quantized data and bit shift processing based on the quantization step size Δ _b and the setting information of the ROI part notified from the ROI unit 15 described above.

In favor of the data subjected to the quantization with a smaller quantization step size delta _b allocate more code amount to perform the bit shift processing so as not to be affected by truncation by the rate control, the first This is the same as the embodiment.

  Hereinafter, the bit shift process based on the setting information of the ROI part will be described.

As described above, according to the present invention, it is possible to perform processing when a black-and-white image and a color image are further taken into consideration when visual weighting is considered. In the following, the quantization parameter Q _p = 16 shown in FIG. The details of the process will be described using a color image obtained by visual weighting of a viewing distance 3000 in color image data in the YUV422 format.

  First, the ROI unit 15 sets the ROI part and the non-ROI part as described above, and notifies the quantization unit 14 of the setting information. In addition to information indicating whether each pixel data is an ROI part or a non-ROI part, the setting information includes information on the priority when priority is set for a plurality of ROI parts.

  Based on the notified setting information of the ROI part, the quantization unit 14 further divides the transform coefficient in the state of FIG. 10 into data of the ROI part and the non-ROI part.

In Figure 10, the transform coefficients are subjected to processing and sorting bit shift on the basis of the quantization step size delta _b as described above, are shown for each sub-band shown in FIG. As can be seen from the correspondence between FIG. 5 and FIG. 14, the conversion coefficients of each subband shown in FIG. 10 actually include data of both the ROI portion 31 a and the non-ROI portion 32 a shown in FIG. 14. ing.

  Here, in this embodiment, as shown in FIG. 17, it is assumed that four ROI portions 36 to 39 and a non-ROI portion 40 are set on the image data. Further, it is assumed that priorities are set in the order of the ROI portion 36, the ROI portion 37, the ROI portion 38, and the ROI portion 39 for each ROI portion. The ROI unit 15 develops the ROI part on the wavelet plane as described above based on the ROI part and the priority information.

  However, the ROI portion is developed for each code block obtained by dividing the image data developed on the wavelet plane by applying DWT into rectangular regions of a predetermined size. Specifically, for example, when each subband is divided into code blocks of a predetermined size such as 32 pixels in the vertical direction and 32 pixels in the horizontal direction, whether each code block is the ROI portion or not, and the ROI portion. Is assigned a priority and is expanded on the wavelet plane.

  When a plurality of ROI parts, or both ROI parts and non-ROI parts are included in one code block, the ROI part or ROI having a high priority for the code block. Select and assign each part. That is, a higher priority is selected and assigned. Or you may set a new priority according to the ratio of the ROI part contained in one code block, and the non-ROI part.

  As a result, for the ROI portion and priority setting shown in FIG. 17, a wavelet plane as shown in FIG. 18 is developed. In FIG. 18, the rectangles constituting each subband indicate code blocks.

  The setting information related to the ROI part set in this way is input to the quantization unit 14 shown in FIG.

  FIG. 10 shows each subband shown in FIG. 5. As can be seen from the correspondence between FIG. 5 and FIG. 18, the data of each subband shown in FIG. All the data of the four ROI portions 36a to 39a and the non-ROI portion 40a are included.

  Based on the setting information of the ROI portion notified from the ROI unit 15, the quantization unit 14 converts the code string shown in FIG. 10 into the data of the four ROI portions 36 to 39 and the non-ROI as shown in FIG. The data is separated into the data of the portion 40.

  Further, the quantization unit 14 bit-shifts each data of the ROI portions 36 to 39 rightward by a predetermined number of bits. At this time, when a plurality of ROI portions are set, the shift amount is set to be larger as the priority is higher according to the priority.

For example, in FIG. 19, for the data of the non-ROI portion 40, the ROI portion 39 with the lowest priority is 1 bit (V _ml = -1), and the ROI portion 38 with the next highest priority is 2 bits (V _ml = -2), the next highest priority ROI portion 37 is 3 bits (V _ml = -3), and the highest priority ROI portion 36 is 4 bits (V _ml = -4). ing.

  The shift amount of the data of the ROI portions 36 to 39 and the data of the non-ROI portion 40 may be a predetermined number of bits, or may be a target noise removal effect or an image after compression coding. The mode may be changed to an arbitrary value according to the image quality of data.

  Further, the data of the ROI portions 36 to 39 and the data of the non-ROI portion 40 are not limited to the mode of arranging the data in order from the high priority portion as shown in FIG. A mode of arranging them in order may be used. For example, in the code amount control described later, since the lower bit plane includes higher harmonic components such as noise components in the code amount control described later, rate control for the lower bit plane is effective for achieving the target image quality. It is determined in consideration of the relationship between the order of data deletion by rate control at this time and the priority of the portions 36 to 40, the contents of the image data of the portions 36 to 40, and the like.

  The transform coefficient QD output from the quantization unit 14 is subjected to block-based entropy coding by the coefficient bit modeling unit 20 and the arithmetic coding unit 21, and the rate is controlled by the code amount control unit 22.

  In JPEG2000 Part 1 standardized in Non-Patent Document 2, only when the Max-Shift method is used, the quantizing unit 14 is allowed to shift in two types, an ROI part and a non-ROI part. Yes. Therefore, as shown in FIG. 29, the shift amount is right-shifted by the maximum value of the ROI portion. The shape of the ROI does not need to be in units of code blocks, and may be shifted in units of coefficients on the wavelet plane of the ROI portion 31a and the non-ROI portion 32a shown in FIG.

  Therefore, in order for the quantization unit 14 to set the priority of an arbitrary shift amount with respect to a plurality of ROIs, it is necessary to extend JPEG2000 Part1 standardized in Non-Patent Document 2. The mask shape data of each subband and the shift amount indicating the priority of each ROI are stored as extended data. The shape data of the mask of each subband may be a code block unit or a wavelet plane coefficient unit. When the mask shape data is stored in units of code blocks, the amount of data to be stored can be reduced.

{Code amount control}
The code amount control unit 22 performs rate control of the encoded data AD processed by the coefficient bit modeling unit 20 and the arithmetic encoding unit 21 after the quantization unit 14 performs quantization.

  As shown in FIG. 19, the code amount control unit 22 performs a target noise removal effect on a code string that has been rearranged and bit-shifted (that is, a target image quality is obtained). Truncate the data. Data truncation is performed in order from the rightmost bit. For example, the bit data of number 0 of VHL4 in the code string of the ROI portion 36 shown in FIG. 19 is deleted in the order of bit data of number 0 of YHH5 in the downward direction. Then, if the target noise removal effect is achieved by truncating the bit data up to YHH1, the data at the corresponding dot portion in FIG. 19 is discarded. At this time, if the target image quality cannot be obtained even if the data of YHH1 is rounded down, the bit data of number 0 of ULL4, the number of bit data of ULL4, and the number of YLH5 are subsequently shifted from the VLL4 number 0 bit data in the ROI portion 36. Bit data of 0 ... are deleted in order. Bit data deletion is continued until the target image quality is obtained.

  The bit data (bit plane) can be decomposed into SIG path, MR path, and CL path as shown in FIG. The rate control by the code amount control unit 22 may be performed in units of paths in addition to the mode performed in units of bit planes as described above.

According to the compression encoding device 1 according to this embodiment performs sorting and bit shifting according to the value of the quantization step size delta _b, further sorting and bit shifting based on the setting information of the ROI The rate control is performed by deleting the bit data or pass of each performed subband from the lower bits until a desired noise removal effect is obtained (that is, until a desired target image quality is obtained). As a result, it is possible to strictly control the target image quality with a noise removal effect.

  Moreover, since it is not necessary to calculate the amount of distortion in each coding pass for rate / distortion optimization processing as in the past, it is possible to realize high-efficiency rate control with high real-time performance and significantly reduced overhead. . Image compression makes it possible to generate an image from which noise has been removed.

  In addition, data is truncated in order from the ROI part with the highest priority. For this reason, it is possible to preferentially remove information including high-frequency components such as noise components from the ROI information with a high priority by rate control, and to achieve a target image quality that has a noise removal effect for the ROI portion with a high priority. Thus, it is possible to perform compression encoding while being controlled.

  For example, by controlling the rate of human skin or the like as an ROI portion, wrinkles or spots on the skin can be removed as noise components, so that a skin beautifying effect can be obtained.

  FIG. 20 is a block diagram showing a modification of the present embodiment. The difference from the example shown in FIG. 11 is that information on the ROI part set by the ROI unit 15 is input to the code amount control unit 22 instead of the quantization unit 14.

  In the example shown in FIG. 11, data rearrangement and bit shift processing based on the quantization step size is performed at the quantization stage, and data rearrangement and bit shift processing based on the setting information of the ROI portion is performed. ing. On the other hand, in the example shown in FIG. 20, the quantization unit 14 performs only the quantization based on the quantization step size. Then, after block-based entropy coding is performed by the coefficient bit modeling unit 20 and the arithmetic coding unit 21 output from the quantization unit 14, the quantization step size is determined when the code amount control unit 22 performs rate control. The rearrangement and bit shift processing based on the data, and the data rearrangement and bit shift processing based on the setting information of the ROI portion are performed. Since the processing of data rearrangement and bit shift based on the setting information of the ROI part is performed by the code amount control unit 22 instead of the quantization unit 14, it is the same as the example shown in FIG. Description is omitted.

  In the example shown in FIG. 20, all the rearrangement and bit shift processes are performed by the code amount control unit 22. Therefore, even in a compression coding apparatus in which the quantization unit 14 does not have a function related to bit shift, the present embodiment can be realized only by changing the function and operation of the code amount control unit 22.

(Third embodiment)
In the first embodiment, after quantization using the quantization step size Δ _b , data is truncated so that the target image quality with a noise removal effect (skin-beautifying effect) is obtained based on the code amount control process. . The quantization step size Δ _b used for quantization of each subband is expressed by Equation (1), Equation (2) using the norm of the synthesis filter coefficient, or “energy weighting factor” described in Non-Patent Document 2. And the equation (8) using the product of the norms of the synthesis filter coefficients.

On the other hand, in the present embodiment, after quantization using the quantization step size Δ _b , a “Viewing distance” that differs in the numerical value of “energy weighting factor” is applied in the code amount control process. An example of the above will be described.

  Hereinafter, the configuration and operation of the compression coding apparatus according to the present embodiment will be described focusing on differences from the first embodiment.

The quantization step size Δ _b ′ using a different “Viewing distance” in the code amount control process is represented by the following equation (14).

Here, Q _p ′ is a positive number input according to the target image quality information in the code amount control process, that is, a quantization parameter, and a smaller value is input as the image quality is higher. Q _b ′ is the expression (2) using the norm of the synthesis filter coefficient, or the expression (8) using the product of the numerical value of “energy weighting factor” described in Non-Patent Document 2 and the norm of the synthesis filter coefficient. It is the same.

The relationship between Δ _b ′ and Δ _b is defined as in the following equation (15).

Here, α is a weight of the quantization step size Δ _b ′ in the code amount control process with respect to the quantization step size Δ _b used for quantization of each subband.

For example, in the first embodiment, for all of the luminance signal Y and the color difference signals U and V, the quantization parameter Q _p = 16 is used, and the visual distance of 3000 in the viewing image (viewing distance) 3000 in the color image data in the YUV422 format. luminance signal Y, a quantization step size delta _b of the color difference signals U and V when performing weighting is shown in Table 16 to Table 18. In the code amount control processing, visual weighting of viewing distance 1000 was performed on the color image data in the YUV422 format with the quantization parameter Q _p ′ = 16 for all of the luminance signal Y and the color difference signals U and V. Α such that the quantization step size Δ _b ′ of the luminance signal Y and the color difference signals U and V is expressed by the following equation (16).

The weight α of the quantization step size Δ _b ′ is determined by the quantization step size Δ _b notified from the image quality control unit 23 by the code amount control unit 22 and the target image quality information input to the code amount control unit 22.

{Sort and bit shift}
The first embodiment is based on the quantization step size Δ _b ′ using the different “Viewing distance” and the weight α of the quantization step size obtained in this way in the code amount control process. Similar to the above, rearrangement and bit shift are performed.

  The code amount control unit 22 illustrated in FIG. 1 determines the number of bits S to be bit-shifted using the quantization step size weight α calculated in the code amount control unit 22 as shown in the following equation (17). .

Here, ROUND is a function for rounding the decimal point, and rounding down, rounding up, and rounding can be considered. Shifts to the left if the value is positive, and to the right if the value is negative. An example of the number of bits S to be bit-shifted when rounding off with a quantization step size Δ _b of Viewing distance 3000 in Expression (16) and a quantization step size Δ _b ′ of Viewing distance 1000 of 1000 19 to Table 21.

Next, the respective code strings are rearranged in the ascending order of Δ _b ′ (ascending order) as in the first embodiment, based on the quantization step size Δ _b ′ obtained by the code amount control unit 22. Tables 22 to 24 show quantization step sizes Δ _b ′ corresponding to the bit shift number S in Tables 19 to 21. The quantization step size Δ _b ′ is used only for reordering.

In the present embodiment, the image quality control unit 23 obtains the quantization step size delta _b, and notifies the quantization unit 14. Then, the quantization unit 14, for each subband, performs quantization in accordance with the notified quantization step size delta _b. In this case, smaller than 1 the quantization step size delta _b, to use that after one or more values by multiplying the power of 2, is the same as described above.

  Therefore, the actual shift amount is the sum of the shift amount and the shift amount of Expression (17).

{Code amount control processing}
Next, processing contents of the code amount control unit 22 shown in FIG. 1 will be described. As shown in FIG. 21, the difference from the first embodiment is only the difference in the shift amount due to the number of bits S to be bit-shift determined based on the weight α of the quantization step size. As shown in FIG. 21, for example, YHH1 is shifted to the left by 8 bits, and UHH1 is shifted to the left by 4 bits.

  As a result, since the “Viewing distance” can be freely changed by the code amount control unit 22, an image encoded with the compression quantization value can also be reduced by the code amount control unit 22. It is possible to strictly control to achieve a target image quality with a skin-beautifying effect.

  Similar to the first embodiment, all the rearrangement and bit shift processes may be performed by the code amount control unit 22. In this case, even in a compression coding apparatus in which the quantization unit 14 does not have a function related to bit shift, the present embodiment can be realized only by changing the function and operation of the code amount control unit 22.

(Fourth embodiment)
In the second embodiment, after quantizing using the quantization step size Δ _b , a bit shift is performed by assigning a priority to each ROI, thereby performing a noise removal effect (skin-beautifying effect) based on the code amount control process. The data was truncated to achieve a certain target image quality.

On the other hand, in this embodiment, as in the third embodiment, after quantization using the quantization step size Δ _b , “Viewing”, which differs in the numerical value of “energy weighting factor” in the code amount control process An embodiment in the case where “distance (viewing distance)” is applied to each ROI will be described.

  Hereinafter, the configuration and operation of compression encoding in the present embodiment will be described focusing on differences from the first to third embodiments. The setting of the compression encoding device, quantization, and ROI part is the same as in the second embodiment.

{Sort and bit shift}
The quantization unit 14 illustrated in FIG. 11 or 20 performs the rearrangement and bit shift processing of the input encoded data AD based on the quantization step size notified from the image quality control unit 23. At this time, the transform coefficient quantized using the quantization step size transformed so as to be 1 or more is bit-shifted to the left by the number of bits corresponding to the exponent of power of 2 multiplied at the time of transformation.

For example, in the same manner as in the third embodiment, the quantization parameter Q _p = 16 and the image data in the YUV422 format color is quantized by visual weighting of viewing distance 3000. In the code amount control process, the quantization parameter Q _p ′ = 16, and the rearrangement and bit shift process based on the quantization step size obtained by visually weighting the viewing distance 1000 in the color image data in the YUV422 format Is performed, a code string as shown in FIG. 22 is obtained.

  Next, the quantization unit 14 shown in FIG. 11 or the code amount control unit 22 shown in FIG. 20 is based on the setting information of the ROI part notified from the ROI unit 15 to the code string shown in FIG. Sorting and bit shifting are performed. By performing processing so that the code amount of the ROI part is cut off more than that of the non-ROI part, the noise removal effect (skin-beautifying effect) of the ROI part by the encoding process can be expected.

First, the quantization unit 14 shown in FIG. 11 or the code amount control unit 22 shown in FIG. 20 performs the code subjected to the rearrangement and bit shift processing based on the quantization step size Δ _b ′ as described above. Based on the setting information of the ROI portion notified from the ROI unit 15, the conversion column is divided into data of the ROI portion and the non-ROI portion.

  As can be seen from the correspondence between FIG. 5 and FIG. 18, in fact, the data of each subband shown in FIG. 10 includes all the data of the four ROI portions 36a to 39a and the non-ROI portion 40a shown in FIG. It is included.

  The quantization unit 14 shown in FIG. 11 or the code amount control unit 22 shown in FIG. 20 converts the code string shown in FIG. 10 into the code sequence shown in FIG. 19 based on the setting information of the ROI part notified from the ROI unit 15. As shown, the data of the four ROI parts 36 to 39 and the data of the non-ROI part 40 are separated.

Since the quantization step size Δ _b ′ using a different “Viewing distance” is used in the code amount control process for each ROI set region, as in the third embodiment, the expression The number of bits S to be bit-shifted using the quantization step size weight α in (17) is applied to each ROI.

In FIG. 19, priority is given to each ROI part, and the whole is just bit-shifted. Similar to FIG 21, also for each ROI portions, the code amount control processing "the Viewing distance (viewing distance)" used in the quantization step size delta _b different figures "energy weighting factor""Viewing distance ( (Viewing distance) "can be changed.

  In FIG. 22, the code amount control processing is performed at the viewing distance 1000 for the ROI portions 36 and 37 and at the viewing distance 3000 for the ROI portions 38 and 39 and the non-ROI portion 40.

{Code amount control}
The code amount control unit 22 performs data truncation on the code string rearranged as shown in FIG. 22 so as to obtain a target image quality with a noise removal effect (skin-beautifying effect). Data truncation is performed in order from the rightmost bit. For example, the bit data of number 0 of VHL4 in the ROI portion 36 shown in FIG. 22 is deleted in the order of bit data of number 0 of YHH5 in the downward direction. Then, if the target image quality is obtained by truncating the bit data up to YHH3, the corresponding portion of data is discarded. At this time, if the target image quality is not achieved even if the data up to YHH3 is cut off, then the bit data of number 0 of VLL4 in the ROI portion 36 is deleted in the order of bit data of number 0 of UL4 downward. go. The deletion of bit data is continued in order from the bit located on the right side until the target image quality is achieved.

  Similarly to the third embodiment, in the fourth embodiment, the “Viewing distance” can be freely set by the quantization unit 14 shown in FIG. 11 or the code amount control unit 22 shown in FIG. Since it can be changed, it is possible to strictly control an image encoded with a quantization value for compression so as to obtain a target image quality with a noise removal effect (skin-beautifying effect).

  Similar to the second embodiment, all the rearrangement and bit shift processes may be performed by the code amount control unit 22. In this case, even in a compression coding apparatus in which the quantization unit 14 does not have a function related to bit shift, the present embodiment can be realized only by changing the function and operation of the code amount control unit 22.

It is a figure which shows schematic structure of the compression coding apparatus based on 1st Embodiment of this invention. It is a figure which shows the numerical table of Energy weighting factor. It is a figure which shows the numerical table of Energy weighting factor. It is a figure which shows the numerical table of Energy weighting factor. It is a figure which shows the wavelet plane of a luminance signal. It is a figure which shows the wavelet plane of the color difference signal of a YUV422 format. It is a figure which shows the wavelet plane of the color difference signal of YUV420 format. It is a figure which shows the bit shift process of a code sequence. It is a figure which shows the process of rearrangement of a code string. It is a figure which shows the process of rearrangement of a code string of YUV format, and a bit shift. It is a figure which shows schematic structure of the compression coding apparatus which concerns on 2nd Embodiment of this invention. It is a figure which shows the example of an original image. It is a figure which shows the single mask area | region set with respect to the original image of FIG. It is a figure which shows the state which expand | deployed the mask area | region of FIG. 13 on the wavelet plane. It is a figure which shows the correspondence of the mask area | region between the low-frequency side and high frequency side in an inverse wavelet 5x3 filter, and an input side. It is a figure which shows the correspondence of the mask area | region between the low-frequency side and high frequency side in an inverse wavelet 9x7 filter, and the input side. It is a figure which shows the example of a setting of the ROI part of image data. It is a figure which shows the state which expand | deployed the ROI setting information of FIG. 17 on the wavelet plane. FIG. 18 is a diagram illustrating code string rearrangement and bit shift processing based on the ROI setting information of FIG. 17. It is a figure which shows schematic structure of the compression coding apparatus which concerns on the modification of 2nd Embodiment. It is a figure which shows the rearrangement of a code sequence and the process of a bit shift which concern on 3rd Embodiment of this invention. It is a figure which shows the rearrangement of the code sequence and bit shift process which concern on 4th Embodiment of this invention. It is a figure which shows schematic structure of the compression encoding apparatus by a JPEG2000 system. It is a schematic diagram which shows the two-dimensional image divided | segmented according to the octave division system. It is a schematic diagram which shows the two-dimensional image decomposed | disassembled into the some code block. It is a schematic diagram which shows the several bit plane which comprises a code block. It is a schematic diagram which shows three types of encoding passes. It is a figure which shows the RD curve showing the relationship between a rate and distortion. It is a figure which shows the shift of the ROI part in a quantization part.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Compression encoding apparatus 10 DC level shift part 11 Color space conversion part 12 Tiling part 13 DWT part 14 Quantization part 15 ROI part 17 Bit stream production | generation part 20 Coefficient bit modeling part 21 Arithmetic coding part 22 Code amount control part 23 Image quality control unit

Claims (33)

A compression encoding device for compressing and encoding an image signal,
A wavelet transform unit that recursively divides an image signal into a high-frequency component and a low-frequency component by wavelet transform to generate and output transform coefficients of a plurality of band components;
An image quality control unit for obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient;
An entropy encoding unit for entropy encoding the transform coefficient;
The encoded data output from the entropy encoding unit is rearranged in a predetermined scanning order based on the quantization step size to generate a code string, and the total capacity of the encoded data in the code string is equal to the target image quality. A code amount control unit that performs rate control for truncating a part of the code string;
A compression encoding apparatus comprising: A compression encoding device for compressing and encoding an image signal,
A wavelet transform unit that recursively divides an image signal into a high-frequency component and a low-frequency component by wavelet transform to generate and output transform coefficients of a plurality of band components;
An image quality control unit for obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient;
A region-of-interest setting unit that sets a region of interest in the image signal;
A quantization unit that performs quantization and rearrangement on the transform coefficient based on the quantization step size, and rearrangement and bit shift based on the setting information of the region of interest;
An entropy encoding unit for entropy encoding the transform coefficient input from the quantization unit;
A code amount control unit that performs rate control for truncating a part of the encoded data so that the entire capacity of the encoded data output by the entropy encoding unit is a target image quality;
A compression encoding apparatus comprising: A compression encoding device for compressing and encoding an image signal,
A wavelet transform unit that recursively divides an image signal into a high-frequency component and a low-frequency component by wavelet transform to generate and output transform coefficients of a plurality of band components;
A region-of-interest setting unit that sets a region of interest in the image signal;
A quantization unit that quantizes the transform coefficient;
An entropy encoding unit for entropy encoding the transform coefficient input from the quantization unit;
A code string is generated by performing rearrangement and bit shift based on the setting information of the region of interest for the encoded data output from the entropy encoding unit, so that the entire capacity of the encoded data becomes the target image quality. A code amount control unit that performs rate control for truncating a part of the code string;
A compression encoding apparatus comprising: The compression encoding apparatus according to claim 3, further comprising:
An image quality control unit for obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient,
With
The quantization unit quantizes the transform coefficient based on the quantization step size,
The code amount control unit rearranges the encoded data based on the quantization step size. A compression encoding device according to any one of claims 2 to 4,
The region-of-interest setting unit gives priority to each set region of interest,
The compression coding apparatus according to claim 1, wherein a shift amount of the bit shift based on the setting information of the region of interest is determined according to the priority. A compression encoding device according to any one of claims 1 to 5,
The code amount control unit determines a truncation target for the rate control in units of bit planes. A compression encoding device according to any one of claims 1 to 5,
The code amount control unit determines a truncation target for the rate control in units of paths. A compression encoding device according to any one of claims 1, 2, and 4 to 7,
The image quality control unit divides the specified quantization parameter by a value obtained by multiplying a norm of a filter coefficient by an energy weighting facotor that is a predetermined value determined based on human visual characteristics, and A compression encoding apparatus characterized by obtaining the quantization step size weighted in consideration of visual characteristics. A compression encoding device according to any one of claims 1, 2, 4 to 8,
When the quantization step size is smaller than a predetermined numerical value, the image quality control unit calculates a value obtained by multiplying a power of 2 that makes the quantization step size equal to or larger than the predetermined numerical value. A compression encoding apparatus characterized by the above. The compression encoding apparatus according to claim 9, wherein
When the quantization step size is a value obtained by multiplying the power of 2 in the image quality control unit, when performing the rearrangement based on the quantization step size, the transform coefficient or the encoded data, 2. A compression encoding apparatus, wherein a bit shift is performed by the number of bits corresponding to an exponent of the power of 2. The compression encoding device according to claim 1,
The code amount control unit can change a viewing distance, and bit-shifts the code string in accordance with the change of the viewing distance. The compression encoding apparatus according to claim 2, wherein
The quantization unit is capable of changing a viewing distance, and further performs bit shift of the transform coefficient in accordance with the change of the viewing distance. The compression encoding device according to claim 3, wherein
The code amount control unit can change a viewing distance, and further performs bit shift of the code string in accordance with the change of the viewing distance. A compression encoding method for compressing and encoding an image signal,
(A) a step of recursively dividing the image signal into a high-frequency component and a low-frequency component by wavelet transform to generate conversion coefficients of a plurality of band components;
(B) dividing the quantization parameter indicating the target image quality by the norm of the filter coefficient to obtain a quantization step size;
(C) entropy encoding the transform coefficient;
(D) The encoded data encoded in the step (c) is rearranged in a predetermined scanning order based on the quantization step size to generate a code string, and the total capacity of the encoded data in the code string Performing a rate control for truncating a part of the code string so that a target image quality is achieved;
A compression encoding method comprising: A compression encoding method for compressing and encoding an image signal,
(A) a step of recursively dividing the image signal into a high-frequency component and a low-frequency component by wavelet transform to generate conversion coefficients of a plurality of band components;
(B) dividing the quantization parameter indicating the target image quality by the norm of the filter coefficient to obtain a quantization step size;
(C) setting a region of interest in the image signal;
(D) Applying quantization and rearrangement based on the quantization step size to the transform coefficient, and rearrangement and bit shift based on the setting information of the region of interest;
(E) entropy encoding the transform coefficient subjected to the process of step (d);
(F) performing a rate control for truncating a part of the encoded data so that the entire capacity of the encoded data encoded in the step (e) becomes a target image quality;
A compression encoding method comprising: A compression encoding method for compressing and encoding an image signal,
(A) a step of recursively dividing the image signal into a high-frequency component and a low-frequency component by wavelet transform to generate conversion coefficients of a plurality of band components;
(B) setting a region of interest in the image signal;
(C) quantizing the transform coefficient;
(D) entropy encoding the transform coefficient quantized in step (c);
(E) A code string is generated by performing reordering and bit shifting based on the setting information of the region of interest for the encoded data encoded in the step (d), and the total capacity of the encoded data is a target Performing rate control for truncating a part of the code string so as to obtain image quality;
A compression encoding method comprising: The compression encoding method according to claim 16, comprising:
The step of performing the quantization includes:
Dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient to obtain a quantization step size;
Quantizing the transform coefficient based on the quantization step size;
Including
The step of rearranging the encoded data includes:
Rearranging the encoded data based on the quantization step size;
A compression encoding method comprising: A compression encoding method according to any one of claims 15 to 17, comprising:
The step of setting the region of interest includes:
Giving priority to each set region of interest;
Including
The step of performing a bit shift based on the setting information of the region of interest includes
Bit shifting by a predetermined number of bits determined according to the priority,
A compression encoding method comprising: A compression encoding method according to any one of claims 14 to 18, comprising:
The step of performing the rate control includes:
Determining a rate control truncation target in bit plane units;
A compression encoding method comprising: A compression encoding method according to any one of claims 14 to 18, comprising:
The step of performing the rate control includes:
Determining the rate control truncation target in units of paths;
A compression encoding method comprising: A compression encoding method according to any one of claims 14, 15, and 17 to 20,
The step of obtaining the quantization step size includes:
Weighting that takes into account human visual characteristics by dividing the specified quantization parameter by the value obtained by multiplying the norm of the filter coefficient by energy weighting facotor, which is a predetermined value determined based on human visual characteristics Obtaining the quantization step size subjected to
A compression encoding method comprising: A compression encoding method according to any one of claims 14, 15, and 17 to 21,
The step of obtaining the quantization step size includes:
When the quantization step size is smaller than a predetermined numerical value, a value obtained by multiplying a power of 2 at which the quantization step size is equal to or larger than the predetermined numerical value is set as the quantization step size;
A compression encoding method comprising: The compression encoding method according to claim 22, comprising:
Reordering based on the quantization step size is:
When the quantization step size is a value obtained by multiplying by a power of 2, the transform coefficient or the encoded data quantized with the quantization step size corresponds to the exponent of the power of 2 Generating the code string by bit-shifting by the number of bits;
A compression encoding method comprising: A program for causing a microprocessor to compress and encode an image signal,
A wavelet transform unit that recursively divides an image signal into a high-frequency component and a low-frequency component by wavelet transform to generate and output transform coefficients of a plurality of band components;
An image quality control unit for obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient;
An entropy encoding unit for entropy encoding the transform coefficient;
The encoded data output from the entropy encoding unit is rearranged in a predetermined scanning order based on the quantization step size to generate a code string, and the total capacity of the encoded data in the code string is equal to the target image quality. As a code amount control unit that performs rate control to cut off a part of the code string,
A program for causing the microprocessor to function. A program for causing a microprocessor to compress and encode an image signal,
A wavelet transform unit that recursively divides an image signal into a high-frequency component and a low-frequency component by wavelet transform to generate and output transform coefficients of a plurality of band components;
An image quality control unit for obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient;
A region-of-interest setting unit that sets a region of interest in the image signal;
A quantization unit that performs quantization and rearrangement on the transform coefficient based on the quantization step size, and rearrangement and bit shift based on the setting information of the region of interest;
An entropy encoding unit for entropy encoding the transform coefficient input from the quantization unit;
As a code amount control unit that performs rate control for truncating a part of the encoded data so that the entire capacity of the encoded data output by the entropy encoding unit becomes a target image quality,
A program for causing the microprocessor to function. A program for causing a microprocessor to compress and encode an image signal,
A wavelet transform unit that recursively divides an image signal into a high-frequency component and a low-frequency component by wavelet transform to generate and output transform coefficients of a plurality of band components;
A region-of-interest setting unit that sets a region of interest in the image signal;
A quantization unit that quantizes the transform coefficient;
An entropy encoding unit for entropy encoding the transform coefficient input from the quantization unit;
A code string is generated by performing rearrangement and bit shift based on the setting information of the region of interest for the encoded data output from the entropy encoding unit, so that the entire capacity of the encoded data becomes the target image quality. In addition, as a code amount control unit that performs rate control for truncating a part of the code string,
A program for causing the microprocessor to function. The program according to claim 26, further comprising:
As an image quality control unit for obtaining a quantization step size by dividing a quantization parameter indicating a target image quality by a norm of a filter coefficient,
Allowing the microprocessor to function;
When the microprocessor functions as the quantization unit,
Function to quantize the transform coefficient based on the quantization step size;
When the microprocessor functions as the code amount control unit,
A program that functions to perform rearrangement of the encoded data based on the quantization step size. A program according to any one of claims 25 to 27,
When the microprocessor functions as the region of interest setting unit,
It works to give priority to each set region of interest,
When performing the bit shift based on the setting information of the region of interest in the microprocessor,
A program for causing a function to shift a bit by a predetermined number of bits determined according to the priority. A program according to any one of claims 24 to 28, wherein
When the microprocessor functions as the code amount control unit,
A program for causing a rate control truncation target to be determined in units of bit planes. A program according to any one of claims 24 to 28, wherein
When the microprocessor functions as the code amount control unit,
A program for causing a rate control truncation target to be determined in units of paths. A program according to any one of claims 24, 25, 27 to 30,
When the microprocessor functions as the image quality control unit,
Weighting that takes into account human visual characteristics by dividing the specified quantization parameter by the value obtained by multiplying the norm of the filter coefficient by energy weighting facotor, which is a predetermined value determined based on human visual characteristics A program that functions to obtain the quantization step size that has been subjected to. A program according to any one of claims 24, 25, 27 to 31,
When the microprocessor functions as the image quality control unit,
When the quantization step size is smaller than a predetermined numerical value, the value obtained by multiplying a power of 2 that makes the quantization step size equal to or larger than the predetermined numerical value is caused to function as the quantization step size. A program characterized by that. A program according to claim 32, wherein
When performing the reordering based on the quantization step size in the microprocessor,
When the quantization step size is a value obtained by multiplying the power of 2 by the image quality control unit, the transform coefficient or the encoded data is represented by the number of bits corresponding to the exponent of the power of 2. A program characterized by functioning to shift.

Images