JP2009088736A - Image encoder, image decoder, and image encoding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce deterioration of image quality suitably while reducing the amount of arithmetic processing. <P>SOLUTION: The image encoder comprises a section for encoding input data, a section for decoding the data which are encoded at the encoding section, and a section for reducing flicker for image data decoded at the decoding section. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image encoding technique and an image decoding technique for moving image data.

  Coding methods such as the MPEG (Moving Picture Experts Group) method have been established as a method for recording and transmitting moving image data such as TV signals with high efficiency. The MPEG-1 standard, MPEG-2 standard, MPEG-4 It is an international standard encoding method as a standard. As a method for further improving the compression ratio, H. The H.264 / AVC (Advanced Video Coding) standard is defined.

  In moving picture coding, a high compression rate is realized by techniques such as variable length coding, orthogonal transform such as discrete cosine transform (DCT), quantization of transform coefficients, motion compensation prediction, and the like. However, when quantization control is performed at a low rate, flicker (quantization distortion of luminance level) occurs and subjective image quality degradation occurs.

  Here, as a cause of occurrence of flicker, since a moving image generally has a correlation in the time direction, an image having no large change such as a scene change has a high ratio of selecting an inter-picture coding mode in a P picture. Is mentioned.

  Therefore, the coding distortion generated in the P picture is propagated to the subsequent P pictures until the next I picture, and the change of the coding distortion in the time direction is relatively small. On the other hand, since encoding is performed only in the intra-picture coding mode without using the inter-picture coding mode in the I picture, the coding distortion generated above is temporarily interrupted in the I picture.

  For this reason, the coding distortion changes at intervals of the I picture, and the change of the coding distortion is visually recognized as flicker.

  In this way, as one method for solving the occurrence of flicker in moving image encoding, in Patent Document 1, for a block in which a difference value of a block between frames is equal to or less than a predetermined value, a preceding decoded block and a pixel included in the block A method of selecting a prediction mode that minimizes a difference value from a value has been proposed.

JP 2005-323315 A

  However, when quantization control is performed by the flicker reduction method according to Patent Document 1, prediction error calculation, DCT conversion, quantization, generated code amount calculation, and inverse quantization are performed for each coding mode / block size. In addition, since it is necessary to repeatedly perform encoding / decoding processing such as inverse DCT transformation and distortion amount calculation, there has been a problem that the calculation processing for encoding is enormous.

  The present invention has been made in view of the above problems, and an object thereof is to suitably reduce image quality deterioration while further reducing the amount of calculation processing.

  One embodiment of the present invention may be configured as described in the claims, for example.

  According to the present invention, it is possible to suitably reduce image quality degradation while further reducing the amount of calculation processing.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram of an image coding apparatus according to the first embodiment of the present invention.

  1 includes a subtracter (100), a first DCT unit (101), a quantization unit (102), an inverse quantization unit (103), a first IDCT unit (104), an addition Device (105), deblocking unit (106), frame memory unit (107), variable length coding unit (113), buffer unit (114), and flicker reduction unit (108). The flicker reduction unit (108) includes a second DCT unit (109), an image correction unit (110), a second IDCT unit (111), and a third DCT unit (112).

  Next, the operation of the image coding apparatus shown in FIG. 1 will be described.

  The first DCT unit (101) orthogonally transforms the difference image signal output from the subtracter (100) for each image block and outputs the result to the quantization unit (102). The quantization unit (102) quantizes the input DCT transform data. Here, the first DCT unit (101) and the quantization unit (102) may be configured as one encoding unit. The quantized data is input to the variable length encoding unit (113). The variable length coding unit (113) performs variable length coding on the input data. The encoded data subjected to variable length encoding is temporarily stored in the buffer unit (114). The buffer unit (114) outputs the input encoded data according to the transmission line.

  On the other hand, the quantized data generated by the quantization unit (102) is input to the inverse quantization unit (103). The inverse quantization unit (103) inversely quantizes the quantized data. The inversely quantized image block data is input to the first IDCT unit (104). The first IDCT unit (104) performs inverse orthogonal transform on this data to restore it as a difference block. A local decoded image is generated from the restored difference block via the adder (105), and is input to the flicker reduction unit (108). Here, the inverse quantization unit (103), the first IDCT unit (104), and the adder (105) may be configured as one decoding unit. The flicker reduction unit (108) corrects the input local decoded image. The corrected image block data is stored in the frame memory (107) via the deblocking unit (106). The deblocking unit (106) performs a filter process for reducing block distortion generated during image coding.

  Next, a detailed operation of the flicker reducing unit (108) in FIG. 1 will be described. In the flicker reduction unit (108), first, the local decoded image generated by the adder (105) is input to the second DCT unit (109). The second DCT unit (109) orthogonally transforms the input local decoded image and outputs it to the image correction unit (110). The third DCT unit (112) performs orthogonal transformation (112) on the image data (122) of the adjacent block read from the frame memory unit (107) and outputs (123) to the image correction unit (110). The image correction unit (110) corrects the DCT conversion data (120) using the DCT conversion data (123) of the adjacent block read from the frame memory unit (107). The corrected DCT conversion data (121) is input to the second IDCT unit (111). The second IDCT unit (111) restores this data by inverse orthogonal transformation. The restored image block data is stored in the frame memory (107) via the deblocking unit (106).

Next, a detailed operation of the image correction unit (110) in FIG. 1 will be described. The image correction unit (110) corrects the DCT conversion data (120) using a correction value calculated from the DCT conversion data (123) of the adjacent block. This will be described with reference to FIG. In FIG. 2, D indicates a correction target block, and D (p, q) is the pth horizontal direction and the vertical direction based on the DC component (D (0, 0)) of the DCT conversion data in the correction target block D. Shows the q-th DCT coefficient value. D (p, q) corresponds to the DCT transformed data (120) in FIG. D ₁ is a block adjacent to D above, and D ₁ (p, q) represents DCT transformed data in the block D ₁ . Similarly, D ₂ (p, q) is the DCT transformed data of the block D ₂ adjacent to the left with respect to the correction target block D, and D ₀ (p, q) is adjacent to the upper left of the correction target block D. It shows the DCT transform data block D _0. U represents the number of horizontal coefficients in each DCT transform data, and v represents the number of vertical coefficients in the DCT transform. The image correction unit (110) first calculates the average value (AVE (p, q)) of each frequency component from the DCT transform data (123) of the upper / left / upper left adjacent block. A specific calculation example is shown in Equation 1. In Equation 1, D _i (p, q) represents adjacent image block data DCT-transformed by the third DCT unit (112). In Equation 1, n indicates the total number of adjacent image blocks.

  In FIG. 2, n = 3 and the operation when the DCT conversion data of the upper / left / upper left adjacent block is referred to are illustrated. However, the value of n does not necessarily need to be 3, and the amount of calculation, etc. It may be increased or decreased depending on conditions.

  The image correction unit (110) corrects the DCT conversion data D (p, q) to be corrected using AVE (p, q) obtained by Expression 1. A specific correction example using the calculation result of AVE (p, q) is shown in Formula 2.

以上のように、本発明の第１の実施例に係る画像符号化装置では、第２のＤＣＴ部（１０９）にて周波数変換された画像中のブロックに対して、画面内の隣接するブロックのＤＣＴ変換データを用いて補正を行うことによって、第２のＤＣＴ部にて周波数変換された画像中のブロックと画面内の隣接するブロックとの周波数成分の差異が低減可能であり、１度の符号化動作にて画質劣化を軽減することが可能となる。

As described above, in the image coding apparatus according to the first embodiment of the present invention, the adjacent block in the screen is compared with the block in the image frequency-converted by the second DCT unit (109). By performing correction using the DCT conversion data, it is possible to reduce the frequency component difference between the block in the image frequency-converted by the second DCT unit and the adjacent block in the screen. It is possible to reduce image quality degradation by the digitizing operation.

  Note that as a condition for performing the correction according to the above formula 2, for example, as shown in the formula 3, this correction is applied only to a DCT coefficient region that satisfies a condition that p and q are equal to or less than predetermined threshold values Th_p and Th_q. Therefore, it is possible to reduce the amount of calculation processing for performing the main correction and to apply the main correction only to a low frequency region where the image quality deterioration is more easily visible. It can also be reduced.

Further, when referring to the DCT conversion data of the adjacent block in the image correction unit (110), it is possible to set a constraint condition on the adjacent block to be referred to depending on the similarity between the correction target block and the adjacent block. Specifically, the coefficient values of the frequency components of the correction target block D and the adjacent block D _i are compared, and all coefficient values of 0 ≦ p ≦ (u−1) and 0 ≦ q ≦ (v−1) are obtained. On the other hand, as shown in Expression 4, the adjacent block D _i is used as a reference block only when the condition that the difference absolute value is equal to or less than a predetermined threshold Th_DCT (p, q) is satisfied.

  By providing the above conditions, for example, if the pattern of the correction target block and the adjacent block are significantly different, it is possible to remove the reference at the time of correction and perform image correction that refers only to the adjacent block with a similar pattern It becomes.

  FIG. 3 is a block diagram of an image coding apparatus according to the second embodiment of the present invention.

  3 includes a subtracter (500), a first DCT unit (501), a quantization unit (502), an inverse quantization unit (503), a first IDCT unit (504), an addition Device (505), deblocking unit (506), frame memory unit (507), variable length coding unit (513), buffer unit (514), and flicker reduction unit (508). The flicker reduction unit (508) includes a second DCT unit (509), an image correction unit (510), a second IDCT unit (511), a third DCT unit (515), and a frequency determination unit (516). ) And a switching unit (512).

  Next, the operation of the image coding apparatus shown in FIG. 3 will be described.

  The first DCT unit (501) orthogonally transforms the difference image signal output from the subtracter (500) for each image block and outputs the result to the quantization unit (502). The quantization unit (502) quantizes the input DCT transform data. The quantized data is input to the variable length coding unit (513). The variable length coding unit (513) performs variable length coding on the input data. The encoded data subjected to variable length encoding is temporarily stored in the buffer unit (514). The buffer unit (514) outputs the input encoded data in accordance with the transmission line.

  On the other hand, the quantized data generated by the quantization unit (502) is input to the inverse quantization unit (503). An inverse quantization unit (503) inversely quantizes the quantized data. The inversely quantized image block data is input to the first IDCT unit (504). The first IDCT unit (504) performs inverse orthogonal transform on this data to restore it as a difference block. A local decoded image is generated from the restored difference block via the adder (505) and input to the flicker reduction unit (508). The flicker reduction unit (508) corrects the input local decoded image. The corrected image block data is stored in the frame memory (507) via the deblocking unit (506). The deblocking unit (506) performs a filter process for reducing block distortion generated during image encoding.

  Next, the operation of the flicker reduction unit (508) in FIG. 3 will be described. The flicker reduction unit (508) first outputs the local decoded image that has passed through the adder (505) to the second DCT unit (509). The second DCT unit (509) orthogonally transforms the input local decoded image and outputs it to the image correction unit (510). On the other hand, image block data adjacent to the local decoded image output from the adder (505) is read from the frame memory unit (507) and input to the third DCT unit (515). The third DCT unit (515) orthogonally transforms the input image block data and outputs (521) to the image correction unit (510). The image correction unit (510) corrects the DCT conversion data input from the second DCT unit (509) by using the DCT conversion data (521) read from the frame memory (507) and subjected to DCT conversion. Do. The corrected data is input to the second IDCT unit (511). The second IDCT unit (511) restores this data by inverse orthogonal transform, and outputs (526) to the switching unit (512). The switching unit (512) selects one of the local decoded image (525) before correction and the local decoded image (526) after correction by the switching control signal (524) from the frequency discriminating unit (516) to deblock. Part (506). The frequency discriminating unit (516) outputs the switching control signal (524) to the variable length coding unit (514).

  Next, the operation of the frequency discrimination unit (516) in FIG. 3 will be described with reference to FIG. The frequency discriminating unit (516) acquires the DCT coefficient data (522) of the correction target block output from the DCT unit (501) and inputs it to the DCT coefficient sum calculating unit (801). The DCT coefficient sum calculation unit (801) calculates the sum of the coefficients of the DCT coefficient data (522), and outputs (821) this to the correction determination unit (803). Further, the frequency discriminating unit (516) acquires the DCT coefficient data (523) of the correction target block output from the inverse quantization unit (503) and inputs it to the DCT coefficient sum calculating unit (802). The DCT coefficient sum calculation unit (802) calculates the sum of the coefficients of the DCT coefficient data (523), and outputs (822) this to the correction determination unit (803). The correction determination unit (803) determines whether or not to correct the corresponding block from the values of the DCT coefficient sum (821) and the DCT coefficient sum (822), and outputs a switching control signal (524).

  FIG. 6 shows an example of the determination operation in the correction determination unit (803) of FIG. As shown in FIG. 6, the DCT coefficient sum (DCT_sum1) (821) calculated for the DCT coefficient data (522) of the correction target block output from the DCT unit (501) is compared with a predetermined threshold (Th_sum1). When DCT_sum1 is less than Th_sum1, the switching control signal (flag_select) is set to 0 ((9-1), (9-2) in FIG. 6). On the other hand, when DCT_sum1 is equal to or greater than Th_sum1, a DCT coefficient sum (DCT_sum2) (822) calculated for DCT coefficient data (523) of the correction target block output from the quantization unit (503) is set to a predetermined threshold (Th_sum2). The value of the switching control signal (flag_select) varies depending on the comparison result. Specifically, when DCT_sum2 is less than Th_sum2, the switching control signal (flag_select) is set to 1 ((9-3) in FIG. 6). Conversely, when DCT_sum2 is equal to or greater than Th_sum2, switching control signal (flag_select) is set. Is set to 0 ((9-4) in FIG. 6). When the switching control signal (flag_select) is 0, the switching unit (512) in FIG. 3 selects and outputs the local decoded image (525) before correction, which is an output from the adder (505). On the other hand, when the switching control signal (flag_select) is 1, a corrected local decoded image (525) which is an output from the second IDCT unit (511) is selected and output.

  Details of the determination operation of FIG. 6 will be described below. When the sum (821) of the DCT coefficient data (522) before encoding is sufficiently small (less than the threshold (Th_sum1)), distortion due to quantization is small and correction is unnecessary, so that the switching control signal (flag_select) is set to 0. ((9-1), (9-2) in FIG. 6). On the other hand, when the sum of the DCT coefficient data (522) before encoding is equal to or greater than the threshold (Th_sum1), distortion due to quantization may occur, but the sum of the DCT coefficient data (523) after encoding (822) ) Is sufficiently large (threshold (Th_sum2) or more), the distortion due to quantization is considered to be small, so the switching control signal (flag_select) is set to 0 ((9-4) in FIG. 6), and the sign is reversed. When the sum (822) of the DCT coefficient data (523) after conversion is less than the threshold value (Th_sum2), it is considered that there is a lot of distortion due to quantization, so the switching control signal (flag_select) is set to 1 (FIG. 6 ( 9-3)).

  As described above, the image coding apparatus according to the second embodiment of the present invention is an image coding apparatus that performs image correction on a block in an image frequency-converted by the second DCT unit. By performing correction using an image of an adjacent block in the screen, subjective image quality degradation can be reduced. In addition, it is possible to more effectively reduce image quality degradation by specifying and correcting a region where image quality degradation is conspicuous using the DCT coefficient after DCT conversion and the DCT coefficient after inverse quantization for the correction target block. It becomes possible.

  FIG. 4 is a block diagram showing the configuration of an image decoding apparatus according to the third embodiment of the present invention.

  4 includes a variable length decoding unit (701), an inverse quantization unit (702), an IDCT unit (703), an adder (705), a flicker reduction unit (708), a frame memory unit ( 707) and a deblocking part (706). The flicker reduction unit (708) includes a DCT unit (709), an image correction unit (710), an IDCT unit (711), a second DCT unit (712), and a switching unit (713).

  Next, the operation of the image decoding apparatus shown in FIG. 4 will be described.

  The compressed and encoded image data is input to the variable length decoding unit (701). The variable length decoding unit (701) performs variable length decoding on the input compressed encoded data, and outputs the obtained quantized data to the inverse quantization unit (702). The inverse quantization unit (702) inversely quantizes the quantized data. The inversely quantized image block data is input to the first IDCT unit (703). The first IDCT unit (703) performs inverse orthogonal transform on this data to restore it as a difference block. A local decoded image is generated from the restored difference block via an adder (705) and input to a flicker reduction unit (708). A flicker reduction unit (708) corrects the input local decoded image. The detailed operation of the flicker reduction unit (708) will be described later. The corrected image block data is input to the second IDCT unit (711). The second IDCT unit (711) performs inverse orthogonal transform on this data and inputs the data to the switching unit (713). The data selected by the switching unit (713) is stored in the frame memory (707) via the deblocking unit (706). The deblocking unit (706) performs a filter process for reducing block distortion generated during image coding.

  Next, a detailed operation of the flicker reducing unit (708) in FIG. 4 will be described. The flicker reduction unit (708) first outputs the local decoded image that has passed through the adder (705) to the DCT unit (709). The DCT unit (709) orthogonally transforms the input local decoded image and outputs (721) to the image correction unit (710). The image block data (723) output from the frame memory unit (707) is input to the second DCT unit (712). The second DCT unit (712) orthogonally transforms the input image block data and outputs (724) to the image correction unit (710). The image correction unit (710) corrects the DCT conversion data input from the DCT unit (709) using the DCT conversion data (724) of the adjacent image block. Note that this correction method is equivalent to the correction method in the image coding apparatus, and as an example, is based on Equation 2 in the first embodiment. The corrected data is input to the IDCT section (711). The IDCT unit (711) restores this data by inverse orthogonal transform and outputs it to the switching unit (713). The switching unit (713) selects and outputs either the image data output from the adder (705) or the image data output from the IDCT unit (711). This is performed based on the switching control signal (720) output from the conversion unit (701).

As described above, in the image decoding apparatus according to the third embodiment of the present invention, correction is performed on the blocks in the image frequency-converted by the DCT unit using the images of adjacent blocks in the screen. By doing so, it is possible to reduce the difference in frequency components between the block in the image frequency-converted by the DCT unit and the adjacent block in the screen, and it is possible to reduce image quality degradation.

1 is a configuration block diagram of an image encoding device according to a first embodiment of the present invention. It is the figure which showed an example of the correction example of the frequency component of the image block data which concerns on 1st Example of this invention. It is a block diagram of the configuration of an image encoding device according to a second embodiment of the present invention. It is a block diagram of a configuration of an image decoding apparatus according to a third embodiment of the present invention. It is a figure which shows the structure of the frequency discrimination | determination part concerning the 2nd Example of this invention. It is a figure which shows an example of the determination operation | movement in the correction | amendment determination part which concerns on 2nd Example of this invention.

Explanation of symbols

101, 109, 112, 501, 509, 515, 709, 712 ... DCT unit 102, 502 ... quantization unit 103, 503, 702 ... inverse quantization unit 104, 111, 504, 511, 703, 711 ... IDCT unit 113 513: Variable length encoding unit 100, 500: Subtractors 105, 505, 705 ... Adders 108, 508, 708 ... Flicker reduction units 110, 510, 710 ... Image correction units 107, 507, 707 ... Frame memory unit 106 506 706 Deblocking units 114 514 Buffer units 512 713 Switching units 516 Frequency determination units 701 Variable length decoding units 801 802 DCT coefficient sum calculation units 803 Correction determination units

Claims (10)

An image encoding device for encoding input data,
An encoding unit for encoding the input data;
An inverse encoding unit that decodes encoded data encoded by the encoding unit;
An image encoding apparatus comprising: a flicker reducing unit that reduces flicker on image data decoded by the inverse encoding unit. The flicker reduction unit includes an image correction unit that corrects the decoded image data;
The image coding apparatus according to claim 1, further comprising: a switching unit that switches between the corrected image data and the decoded image data by a switching control signal from a frequency discrimination unit.   The image encoding device according to claim 2, wherein the image correction unit corrects the image data of the correction target block using image data of a block adjacent to the correction target block in the screen data.   The frequency discriminating unit determines whether or not to perform correction on the corresponding block from a DCT coefficient sum calculating unit for obtaining a sum of coefficients of the DCT coefficient data, and a value of the DCT coefficient sum and the DCT coefficient sum. The image encoding device according to claim 2, wherein An image decoding device for decoding encoded data,
A decoding unit for decoding the encoded data;
An image decoding apparatus comprising: a flicker reducing unit that reduces flicker on the image data decoded by the decoding unit. The flicker reducing unit is
An image correction unit that corrects the image data decoded by the decoding unit;
The image decoding apparatus according to claim 5, further comprising: a switching unit that switches between the corrected image data and the decoded image data by a switching control signal. An image encoding method for encoding input data, comprising:
An encoding step for encoding the input data;
A reverse encoding step of decoding the encoded data encoded by the encoding unit;
An image encoding method comprising: a flicker reduction step for reducing flicker on the image data decoded by the inverse encoding unit.   In the flicker reduction step, the decoded image data is corrected, and the correction is performed using the frequency information of the image data used in the encoding step and the frequency information of the image data used in the inverse encoding step. The image encoding method according to claim 7, wherein the image data is switched between the decoded image data and the decoded image data.   9. The image encoding method according to claim 8, wherein in the flicker reduction step, image data of a block to be corrected is corrected using an image of a block adjacent to the block to be corrected in the screen data.   9. The flicker reduction step obtains a sum of coefficients of DCT coefficient data, and determines whether or not to correct the corresponding block from the values of the DCT coefficient sum and the DCT coefficient sum. The image encoding method described in 1.

Images