Patent Application
20230199215
The present invention relates to a prediction image correcting device, an image encoding device, an image decoding device, and a program.
In video encoding technology, efficient encoding (compression) has been realized by performing transform, prediction and the like for each block obtained by dividing a present image frame (picture). There are two prediction methods: intra prediction and inter prediction.
The intra prediction is a method of generating a prediction image by referencing decoded neighboring blocks adjacent to an encoding or decoding target block (hereinafter referred to as a “target image block”). The inter prediction is a method of generating a prediction image by referencing decoded frames different from the present frame to which the target image block belongs.
Patent Literature 1 discloses a prediction image correcting device that predicts a target image block by inter prediction to generate an inter-predicted image and predicts the target image block by intra prediction to generate an intra-predicted image, and performs weighted averaging on the inter-predicted image and the intra-predicted image.
Specifically, the prediction image correcting device disclosed in Patent Literature 1 evaluates the continuity between the inter-predicted image corresponding to the target image block and a decoded neighboring block adjacent to the target image block, and when evaluating that it is discontinuous, corrects an edge region of the inter-predicted image by filter processing using the decoded neighboring block.
The prediction image correcting device disclosed in Patent Literature 1 applies the filter processing according to the evaluation of the continuity between the inter-predicted image and the decoded neighboring block regardless of the prediction accuracy of the inter prediction.
For that reason, in a case such as when there is an object boundary at the boundary between the inter-predicted image and the decoded neighboring block, even when the prediction accuracy at the edge region of the inter-predicted image is high, the filter processing is applied to the edge region, which causes a problem that the accuracy of the prediction image may be lowered due to the filter processing.
Thus, it is an object of the present invention to provide a prediction image correcting device, an image encoding device, an image decoding device, and a program that can correct a prediction image in an appropriate manner.
A prediction image correcting device according to a first feature includes: a predictor configured to predict a target image block obtained by dividing a present image frame by using a plurality of reference images to generate a prediction image corresponding to the target image block; a prediction accuracy evaluator configured to evaluate prediction accuracy of the prediction image based on a degree of similarity between the plurality of reference images used for generating the prediction image; and a corrector configured to perform correction processing on the prediction image by using a decoded neighboring block adjacent to the target image block, wherein the corrector is configured to control the correction processing based at least on an evaluation result of the prediction accuracy evaluator.
Note that the prediction using the plurality of reference images is typified by bi-prediction for inter prediction but is not limited thereto, and a similar technique can also be applied in the case of generating the prediction image by multiple reference, such as the IntraBC mode (intra block copy mode) of images used in the HEVC encoding scheme, for example.
An image encoding device according to a second feature includes the prediction image correcting device according to the first feature.
An image decoding device according to a third feature it includes the prediction image correcting device according to the first feature.
A program according to a fourth feature causes a computer to function as the prediction image correcting device according to the first feature.
According to the present invention, it is possible to provide a prediction image correcting device, an image encoding device, an image decoding device, and a program that can correct a prediction image in an appropriate manner.
An image encoding device and an image decoding device according to embodiments will be described with reference to the drawings. The image encoding device and the image decoding device according to the embodiments encode and decode a moving image typified by MPEG. The same or similar reference numerals will be assigned to the same or similar portions in the following description of the drawings.
The block divider 100 divides an input image frame (or picture) into block-like subregions, and outputs the image blocks to the subtractor 101 (and the inter predictor 108). For example, the size of an image block is 32×32 pixels, 16×16 pixels, 8×8 pixels, 4×4 pixels, or the like. The image block is a unit for which the image encoding device 1 performs encoding and the image decoding device 2 performs decoding, and such an image block is referred to as a target image block. Note that the shape of the image block is not limited to a square and may also be a rectangle.
The subtractor 101 calculates a prediction residual indicating the pixel-by-pixel difference between the target image block input from the block divider 100 and a prediction image corresponding to the target image block (a prediction image block). Specifically, the subtractor 101 calculates the prediction residual by subtracting each pixel value of the prediction image from each pixel value of the encoding target block, and outputs the calculated prediction residual to the transformer 102a. In the present embodiment, the prediction image has been corrected by the prediction image corrector 110, which will be described later, and is input to the subtractor 101 from the prediction image corrector 110.
The transformer 102a and the quantizer 102b configure a transformer/quantizer 102 which performs orthogonal transform processing and quantization processing in a block unit.
The transformer 102a performs an orthogonal transform of the prediction residual input from the subtractor 101 to calculate transform coefficients, and outputs the calculated transform coefficients to the quantizer 102b. The orthogonal transform refers to, for example, discrete cosine transform (DCT), discrete sine transform (DST), Karhunen-Loeve transform (KLT), or the like.
The quantizer 102b quantizes the transform coefficients input from the transformer 102a using a quantization parameter (Qp) and a quantization matrix to generate quantized transform coefficients. The quantization parameter (Qp) is a parameter to be applied in common to respective transform coefficients within a block and is a parameter which defines roughness of quantization. The quantization matrix is a matrix having a quantization value upon quantization of the respective transform coefficients as an element. The quantizer 102b outputs quantization control information, the generated quantized transform coefficient information, and the like, to the entropy encoder 103 and the inverse quantizer 104a.
The entropy encoder 103 performs entropy encoding on the quantized transform coefficients input from the quantizer 102b for data compression to generate encoded data (a bit stream), and outputs the encoded data to the outside of the image encoding device 1. The entropy encoding can use Huffman encoding, context-based adaptive binary arithmetic coding (CABAC), and the like. Note that information about prediction is input to the entropy encoder 103 from the intra predictor 107 and the inter predictor 108. The entropy encoder 103 also performs entropy encoding of this information.
The inverse quantizer 104a and the inverse transformer 104b configure an inverse quantizer/inverse transformer 104 which performs inverse quantization processing and inverse orthogonal transform processing in a block unit.
The inverse quantizer 104a performs inverse quantization processing corresponding to the quantization processing performed by the quantizer 102b. Specifically, the inverse quantizer 104a restores the transform coefficients by performing inverse quantization of the quantized transform coefficients input from the quantizer 102b using the quantization parameter (Qp) and the quantization matrix, and outputs the restored transform coefficients to the inverse transformer 104b.
The inverse transformer 104b performs inverse orthogonal transform processing corresponding to the orthogonal transform processing performed by the transformer 102a. For example, in a case where the transformer 102a performs discrete cosine transform, the inverse transformer 104b performs inverse discrete cosine transform. The inverse transformer 104b restores the prediction residual by performing an inverse orthogonal transform of the transform coefficients input from the inverse quantizer 104a and outputs the restored prediction residual to the combiner 105.
The combiner 105 combines the restored prediction residual input from the inverse transformer 104b with the prediction image input from the prediction image corrector 110 on a pixel-by-pixel basis. The combiner 105 reconstructs (decodes) the target image block by adding each pixel value of the restored prediction residual and each pixel value of the prediction image, and outputs a reconstruction image block, which is the reconstructed target image block, to the memory 106. Note that the reconstruction image block may be referred to as a decoded block.
The memory 106 stores the reconstruction image block input from the combiner 105. The memory 106 stores the reconstruction image block on a frame-by-frame basis.
The intra predictor 107 generates an intra-predicted image by referencing a decoded neighboring block that is adjacent to the target image block out of reconstruction image blocks (decoded blocks) stored in the memory 106. The intra predictor 107 selects an optimum intra prediction mode and performs intra prediction by using the selected intra prediction mode. A plurality of intra prediction modes corresponding to a plurality of intra prediction directions are prescribed. When a target pixel in the target image block is to be predicted by referencing a neighboring pixel adjacent to the target image block, the intra prediction direction refers to the direction of the neighboring reference pixel with respect to the target pixel. That is, the intra prediction mode (intra prediction direction) determines a neighboring reference pixel to be used for predicting each pixel in the target image block. The intra predictor 107 outputs the intra-predicted image (and/or neighboring reference pixel) to the prediction image corrector 110, as well as outputting information of the selected intra prediction mode to the entropy encoder 103.
The inter predictor 108 performs inter prediction for predicting the target image block by using a reconstruction image (decoded image) frame stored in the memory 106 as a reference image. Specifically, the inter predictor 108 calculates motion vectors by using a technique such as block matching, and generates an inter-predicted image based on the motion vectors. The inter predictor 108 selects an optimum inter prediction method from inter prediction using a plurality of reference images (typically, bi-prediction) and inter prediction using a single reference image (unidirectional prediction), and performs the inter prediction by using the selected inter prediction method. In the present embodiment, description will be made mainly for a case where the inter predictor 108 uses the bi-prediction will be mainly described, and the inter predictor 108 corresponds to a predictor that performs prediction by using a plurality of reference images. The inter predictor 108 outputs the generated inter-predicted image to the prediction image corrector 110, as well as outputting information about the selected inter prediction method and the motion vectors to the entropy encoder 103. The inter predictor 108 also outputs the plurality of reference images used for the inter prediction to the prediction accuracy evaluator 109.
The prediction accuracy evaluator 109 evaluates prediction accuracy at an edge region of the inter-predicted image based on the degree of similarity between the plurality of reference images used for generating the inter-predicted image. Specifically, if the inter predictor 108 performs the inter prediction by using a plurality of reference images, the degree of similarity between the plurality of reference images used for the inter prediction is calculated for each image portion consisting of one or more pixels, the prediction accuracy of the inter-predicted image is evaluated for each image portion based on the degree of similarity, and information of the evaluation result is output to the prediction image corrector 110. In the present embodiment, description will be made mainly for an example where the prediction accuracy evaluator 109 calculates the degree of similarity between the plurality of reference images used for the prediction on a pixel-by-pixel basis and evaluates the prediction accuracy of the inter-predicted image on a pixel-by-pixel basis. The details of the prediction accuracy evaluator 109 will be described later.
The prediction image corrector 110 corrects the inter-predicted image input from the inter predictor 108 and the intra-predicted image input from the intra predictor 107 by performing weighted averaging on them. Specifically, the prediction image corrector 110 corrects the edge region of the inter-predicted image by filter processing by using the decoded neighboring block (neighboring reference pixel) corresponding to the intra-predicted image. The prediction image corrector 110 also controls the filter processing based at least on the evaluation result of the prediction accuracy evaluator 109. To “control the filter processing” includes controlling whether to perform the filter processing and controlling filter intensity. In the present embodiment, description will be made mainly for an example where the prediction image corrector 110 controls whether to perform the filter processing based at least on the evaluation result of the prediction accuracy evaluator 109. For example, the prediction image corrector 110 performs the filter processing when the prediction accuracy evaluated by the prediction accuracy evaluator 109 is less than or equal to a first threshold and does not perform the filter processing when the prediction accuracy evaluated by the prediction accuracy evaluator 109 exceeds the first threshold. The prediction image corrector 110 outputs the filter-processed inter-predicted image to the subtractor 101 and the combiner 105 as the prediction image when the filter processing is performed, and outputs the inter-predicted image, as it is, to the subtractor 101 and the combiner 105 as the prediction image when the filter processing is not performed. Note that the “prediction accuracy” may be represented by “prediction inaccuracy”. With regard to the “prediction inaccuracy”, that the prediction accuracy is less than or equal to the first threshold may mean that the prediction inaccuracy is high, for example. That the prediction accuracy exceeds the first threshold may mean that the prediction inaccuracy is low. The details of the prediction image corrector 110 will be described later.
The entropy decoder 200 decodes the encoded data generated by the encoding device 1, and outputs the quantized transform coefficients to the inverse quantizer 201a. Also, the entropy decoder 200 decodes the encoded data, acquires the information about the prediction (intra prediction and inter prediction), and outputs the information about the prediction to the intra predictor 204 and the inter predictor 205.
The inverse quantizer 201a and the inverse transformer 201b configure an inverse quantizer/inverse transformer 201 which performs inverse quantization processing and inverse orthogonal transform processing in a block unit.
The inverse quantizer 201a performs inverse quantization processing corresponding to the quantization processing performed by the quantizer 102b of the image encoding device 1. The inverse quantizer 201a restores the transform coefficients by performing inverse quantization of the quantized transform coefficients input from the entropy decoder 200 by using the quantization parameter (Qp) and the quantization matrix, and outputs the restored transform coefficients to the inverse transformer 201b.
The inverse transformer 201b performs inverse orthogonal transform processing corresponding to the orthogonal transform processing performed by the transformer 102a of the image encoding device 1. The inverse transformer 201b restores the prediction residual by performing an inverse orthogonal transform of the transform coefficients input from the inverse quantizer 201a and outputs the restored prediction residual to the combiner 202.
The combiner 202 reconstructs (decodes) the original target image block by combining the prediction residual input from the inverse transformer 201b and the prediction image input from the prediction image corrector 207 on a pixel-by-pixel basis, and outputs the reconstruction image block to the memory 203.
The memory 203 stores the reconstruction image block input from the combiner 202. The memory 203 stores the reconstruction image block on a frame-by-frame basis. The memory 203 outputs the reconstruction image (decoded image) frames to the outside of the image decoding device 2 in the display order.
The intra predictor 204 generates an intra-predicted image by referencing the reconstruction image block stored in the memory 203 and performing intra prediction according to the intra prediction information (intra prediction mode) input from the entropy decoder 200. Specifically, the intra predictor 204 generates the intra-predicted image by referencing a neighboring reference pixel that is determined according to the intra prediction mode out of reconstruction image blocks (decoded blocks) stored in the memory 203. The intra predictor 204 outputs the intra-predicted image (and/or neighboring reference pixel) to the prediction image corrector 207.
The inter predictor 205 performs inter prediction for predicting the target image block by using a reconstruction image (decoded image) frame stored in the memory 106 as a reference image. The inter predictor 205 generates an inter-predicted image by performing the inter prediction according to the inter prediction information (motion vector information or the like) input from the entropy decoder 200, and outputs the inter-predicted image to the prediction image corrector 207. The inter predictor 205 also outputs a plurality of reference images used for the inter prediction to the prediction accuracy evaluator 206.
The prediction accuracy evaluator 206 performs an operation similar to that of the prediction accuracy evaluator 109 of the image encoding device 1. If the inter predictor 205 performs the inter prediction by using a plurality of reference images, the prediction accuracy evaluator 206 evaluates the prediction accuracy of the prediction image on a pixel-by-pixel basis by calculating the degree of similarity between the plurality of reference images on a pixel-by-pixel basis, and outputs information of the evaluation result to the prediction image corrector 207.
The prediction image corrector 207 performs an operation similar to that of the prediction image corrector 110 of the image encoding device 1. The prediction image corrector 207 corrects the edge region of the inter-predicted image by filter processing by using the decoded neighboring block (neighboring reference pixel) corresponding to the intra-predicted image. The prediction image corrector 207 performs the filter processing when the prediction accuracy evaluated by the prediction accuracy evaluator 206 is less than or equal to a first threshold and does not perform the filter processing when the prediction accuracy evaluated by the prediction accuracy evaluator 206 exceeds the first threshold. The prediction image corrector 207 outputs the filter-processed inter-predicted image to the combiner 202 as the prediction image when the filter processing is performed, and outputs the inter-predicted image, as it is, to the combiner 202 as the prediction image when the filter processing is not performed. The details of the prediction image corrector 207 will be described later.
As illustrated in
The detected portions are reference images. Information indicating relative positions of the reference images with respect to the target image block indicated with an arrow in the drawing, will be referred to as a motion vector. Information of the motion vector is entropy encoded along with the frame information of the reference images at the image encoding device 1. Meanwhile, the image decoding device 2 detects the reference images based on the information of the motion vector generated by the image encoding device 1.
As illustrated in
A prediction image is generated from the reference images 1 and 2. Note that, in general, prediction processing involves averaging the reference images 1 and 2 having different features but partially having similarity to generate a prediction image having the features of the individual reference images. However, the prediction image may also be generated by combinedly using more advanced processing, e.g., signal enhancement processing such as by a low-pass filter or a high-pass filter. Here, since the reference image 1 contains the circle design and the reference image 2 does not contain the circle design, if the prediction image is generated by averaging the reference images 1 and 2, the circle design in the prediction image would have half-reduced signals as compared to the reference image 1.
A difference between the prediction image obtained from the reference images 1 and 2 and the target image block (encoding target image) is the prediction residual. The prediction residual indicated in
A difference does not exist (at a portion which does not correspond to edges of the stars in the designs and at a background portion) at portions where a degree of similarity between the reference image 1 and the reference image 2 is high, and where prediction is performed with high accuracy. Meanwhile, a large difference exists at portions unique to the respective reference images, that is, at portions where the degree of similarity between the reference image 1 and the reference image 2 is significantly low. Thus, it can be known that prediction accuracy is low and a large difference (residual) exists at portions where the degree of similarity between the reference image 1 and the reference image 2 is significantly low.
If the transform coefficients degrade due to the prediction residual including a portion with a large difference and a portion with no difference being orthogonally transformed and quantized, such degradation of the transform coefficients propagates to the whole of the image (block) through inverse quantization and inverse orthogonal transform. Then, if the target image block is reconstructed by combining the prediction residual (restored prediction residual) restored by inverse quantization and inverse orthogonal transform with the prediction image, degradation of image quality also propagates to portions where prediction has been performed with high accuracy such as a portion which does not correspond to the edges of the stars in the designs and the background portion illustrated in
As illustrated in
The difference calculator 109a calculates the absolute difference values between the reference image 1 and the reference image 2 on a pixel-by-pixel basis, and outputs the calculated absolute difference values to the normalizer 109b. The absolute difference values are an example of values indicating the degree of similarity. It can be regarded that the degree of similarity is higher when the absolute difference values are smaller, and the degree of similarity is lower when the absolute difference values are larger. The difference calculator 109a may calculate the absolute difference values after performing filter processing on each reference image. The difference calculator 109a may calculate statistics such as a squared error to use the statistics as the degree of similarity.
The normalizer 109b normalizes the difference values of the individual pixels input from the difference calculator 109a by the absolute difference value of the pixel with the largest absolute difference value in the block (i.e., the maximum value of the absolute difference values in the block), and outputs normalized difference values, which are the normalized absolute difference values, to the adjuster 109c. In the present embodiment, the normalized difference values are used as weights for weighting each pixel of the restored prediction residual combined with the prediction image in the combiner 105.
The adjuster 109c adjusts the normalized difference values (weights) input from the normalizer 109b based on the quantization parameter (Qp), which defines the coarseness of quantization, and outputs these weights. Since the degree of degradation of the restored prediction residual increases as the coarseness of quantization increases, by adjusting the normalized difference values (weights) based on the quantization parameter (Qp), the adjuster 109c can perform weighting of the restored prediction residual by taking the degree of degradation into consideration.
An estimated prediction accuracy Rij of each pixel (ij) output by the prediction accuracy evaluator 109 can be expressed by the following equation (1), for example.
Rij=1−(abs(Xij−Yij)/maxD×Scale(Qp)) (1)
In the equation (1), Xij is the pixel value of a pixel ij in the reference image 1, Yij is the pixel value of a pixel ij in the reference image 2, and abs is a function for obtaining the absolute value.
Also, in the equation (1), maxD is the maximum value of the difference values abs(Xij-Yij) in the block. Although it is required to obtain the difference values of all pixels in the block to obtain maxD, in order to omit this processing, the maximum value in a neighboring block on which the processing has already been done or the like may instead be used, and the normalization of maxD may be performed by clipping at the used maximum value when there is a value greater than or equal to that value, for example. Alternatively, a table defining the correspondence between the quantization parameter (Qp) and maxD may be used to obtain maxD from the quantization parameter (Qp). Alternatively, a fixed value prescribed by the specifications may be used as maxD.
Further, in expression (1), Scale(Qp) is a coefficient to be multiplied in accordance with the quantization parameter (Qp). Scale(Qp) is designed so as to approach 1.0 in a case where Qp is greater and approach 0 in a case where Qp is smaller, and a degree of approach is adjusted by a system. Alternatively, it is also possible to use a fixed value defined in specifications in advance as Scale(Qp). Further, to simplify the processing, it is also possible to set a fixed value such as 1.0 designed in accordance with the system as Scale(QP).
The adjuster 109c outputs the estimated prediction accuracy Rij. Also, as Rij, the assignment of weights adjusted by a sensitivity function designed according to the system may be output. For example, the sensitivity may be adjusted by providing an offset according to control information such as QP, that is, for example, assuming abs(Xij-Yij)/maxD×Scale(Qp)=Rij, not only Rij=Clip(rij, 1.0, 0.0) but also Rij=Clip(rij+offset, 1.0, 0.0). Note that Clip(x, max, min) indicates a process of clipping at max if x exceeds max and at min if x is below min.
The estimated prediction accuracy Rij calculated in this manner takes a value within a range from 0 to 1.0. Basically, the estimated prediction accuracy Rij approaches 0 when the absolute difference value between the reference images at the pixel ij is larger (that is, the prediction accuracy is lower) and approaches 1 when the absolute difference value between the reference images at the pixel ij is smaller (that is, the prediction accuracy is higher). The prediction accuracy evaluator 109 outputs map information consisting of the estimated prediction accuracies Rij of the individual pixels ij in the block to the prediction image corrector 110 for each block.
Note that the prediction accuracy evaluator 109 may perform the evaluation (calculation of the estimated prediction accuracy Rij) only when the inter prediction using a plurality of reference images is applied, and may not perform the evaluation in other modes, for example, in unidirectional prediction and intra prediction processing without using a plurality of reference images.
Although the prediction accuracy evaluator 109 in the image encoding device 1 has been described, the prediction accuracy evaluator 206 in the image decoding device 2 is configured in a manner similar to the prediction accuracy evaluator 109 in the image encoding device 1. Specifically, the prediction accuracy evaluator 206 in the image decoding device 2 includes a difference calculator 206a, a normalizer 206b, and an adjuster 206c.
The continuity evaluator 110a evaluates the continuity between the inter-predicted image input from the inter predictor 108 and the decoded neighboring block input from the intra predictor 107. The decoded neighboring block is a decoded block adjacent to the target image block. The continuity evaluator 110a outputs the evaluation result of the continuity between the inter-predicted image and the decoded neighboring block to the filter determiner 110b.
Cont=|p2−2×p1+p0|+|q2−2×q1+q0| (2)
Note that the unit and evaluation index for the continuity evaluation are not limited to those in the equation (2) and can be replaced with another unit and evaluation index as long as they are prescribed by the system. For example, the unit for the continuity evaluation may be a side of the block boundary (continuity evaluation for the left boundary and continuity evaluation for the upper boundary are performed), and a pixel value difference (abs(p0−q0)) across the boundary may be used as the evaluation index for the continuity evaluation in order to simplify the processing.
The filter determiner 110b determines whether to perform filter processing based on the prediction accuracy evaluation result input from the prediction accuracy evaluator 109 and the continuity evaluation result input from the continuity evaluator 110a. The filter determiner 110b may perform the determination for each pixel of the block boundary between the inter-predicted image and the decoded blocks, or may perform the determination for each side of the block boundary. The filter determiner 110b may determine to change filter intensity (e.g., the number of taps and frequency response of the filter) based on the prediction accuracy evaluation result and the continuity evaluation result. Note that, if the prediction accuracy evaluator 109 performs the prediction accuracy evaluation on a pixel-by-pixel basis, the filter determiner 110b may use an average value of estimated prediction accuracies R of the individual pixels of the edge region (target region of the filter processing) of the inter-predicted image as the prediction accuracy evaluation result.
The filter determiner 110b determines to perform the filter processing when the prediction accuracy evaluated by the prediction accuracy evaluator 109 is less than or equal to a first threshold α and the continuity evaluated by the continuity evaluator 110a is less than or equal to a second threshold β. On the other hand, the filter determiner 110b determines not to perform the filter processing when the prediction accuracy evaluated by the prediction accuracy evaluator 109 exceeds the first threshold α and/or the continuity evaluated by the continuity evaluator 110a exceeds the second threshold β. Here, as the thresholds α and β, values fixed in the system may be set or values variable by a function using the quantization parameter (Qp) may be calculated and set. Also, if the estimated prediction accuracies R are not normalized, a value variable by a function using the maximum value of the estimated prediction accuracies in the target image block may be calculated and set as the threshold α.
When the filter determiner 110b determines to perform the filter processing, the filter processor 110c performs the filter processing on the edge region of the inter-predicted image. As illustrated in
[Math. 1]
q′
0=Clip3(q0−2×tc,q0+2×tc,(p1+2×p0+2×q0+2×q1+q2+4)>>2)
q′
1=Clip3(q1−2×tc,q1+2×tc,(p0+q0+q1+q2+2)>>2)
q′
2=Clip3(q2−2×tc,q2+2×tc,(p0+q0+q1+3×q2+2×q3+4)>>3) (3)
Here, clip(x, max, min) indicates a process of clipping at max if x exceeds max and at min if x is below min. Also, tc in the equation (3) is an adjustment offset for the process used for the clipping, and a value variable by a function using the quantization parameter (Qp) can be calculated and set.
Note that, although the prediction image corrector 110 in the image encoding device 1 has been described, the prediction image corrector 207 in the image decoding device 2 is configured in a manner similar to the prediction image corrector 110 in the image encoding device 1. Specifically, the prediction image corrector 207 in the image decoding device 2 includes a continuity evaluator 207a, a filter determiner 207b, and a filter processor 207c.
As illustrated in
In step S2, the continuity evaluator 110a evaluates the continuity between the inter-predicted image output by the inter predictor 108 and the decoded neighboring block adjacent to the inter-predicted image.
In step S3, the filter determiner 110b compares the prediction accuracy evaluated by the prediction accuracy evaluator 109 with the first threshold α and compares the continuity evaluated by the continuity evaluator 110a with the second threshold β.
When the prediction accuracy is less than or equal to the first threshold α and the continuity is less than or equal to the second threshold β (step S3: YES), the filter determiner 110b determines to perform the filter processing. In this case, in step S4, the filter processor 110c corrects the edge region of the inter-predicted image by filter processing by using the decoded neighboring block (neighboring reference pixel).
On the other hand, when the prediction accuracy exceeds the first threshold α and/or the continuity exceeds the second threshold β (step S3: NO), the filter determiner 110b determines not to perform the filter processing. In this case, the filter processor 110c does not perform the filter processing.
The image encoding device 1 and the image decoding device 2 according to the present embodiment control the filter processing based on the evaluation result of the prediction accuracy of the inter prediction. In this manner, when the prediction accuracy at the edge region of the inter-predicted image is high, the filter processing can be controlled not to be applied to the edge region, and this can prevent the lowering of the accuracy of the prediction image due to the filter processing. In other words, the filter processing can be applied to the edge region only when the prediction accuracy at the edge region of the inter-predicted image is low, and it is thus possible to perform the filter processing in an appropriate manner.
Also, in the present embodiment, whether to apply the filter processing is determined by operations common to the image encoding device 1 and the image decoding device 2. Specifically, the evaluation of the prediction accuracy of the inter prediction is performed in each of the image encoding device 1 and the image decoding device 2, and whether to apply the filter processing is determined based on the evaluation result. In this manner, it is not necessary to transmit a flag indicating whether to apply the filter processing from the image encoding device 1 to the image decoding device 2, and this can prevent the increase in code amount due to the flag.
In the continuity evaluation and the filter processing in the above-described embodiment, the intra prediction mode (intra prediction direction) applied to the decoded neighboring block may be taken into consideration. This can improve the accuracy of the continuity evaluation and the filter processing.
In the above-described embodiment, an example where the prediction image corrector 110 performs, as correction processing on the inter-predicted image, filter processing on the edge region of the inter-predicted image by using the decoded neighboring block.
Further, the correction processing is not limited to the filter processing, and weighted average processing on the inter-predicted image and the intra-predicted image may be used as the correction processing.
In the present modification, the intra predictor 107 and the prediction image corrector 110 of the image encoding device 1 illustrated in
The intra predictor 107 generates an intra-predicted image corresponding to the target image block by using the decoded neighboring block. For example, the intra predictor 107 generates the intra-predicted image from the decoded neighboring block in a predetermined intra prediction mode. The prediction image corrector 110 performs, as correction processing, weighted average processing on the inter-predicted image input from the inter predictor 108 and the intra-predicted image input from the intra predictor 107. Note that one prediction mode may be selected from a plurality of intra prediction mode candidates to generate the intra-predicted image on the transmission side. In this case, the entropy encoder 103 performs entropy encoding on an intra prediction mode flag indicating which intra prediction mode is used to generate the intra-predicted image on the transmission side.
Similarly, the intra predictor 204 and the prediction image corrector 207 of the image decoding device 2 illustrated in
For example, the intra predictor 204 generates the intra-predicted image from the decoded neighboring block in a predetermined intra prediction mode. The prediction image corrector 207 performs, as correction processing, weighted average processing on the inter-predicted image input from the inter predictor 205 and the intra-predicted image input from the intra predictor 204. Note that the intra prediction mode used for generating the intra-predicted image may be determined based on an intra prediction mode flag sent from the transmission side.
In the case of applying the arrangement of the present modification, “filter processing” in the above-described embodiment may be read as “weighted average processing”. In addition, besides the determination of whether to perform the weighted average processing, the weights for the weighted average processing may be adjusted based on the prediction accuracy evaluation result and the continuity evaluation result.
In the above-described embodiments, description has been made mainly for the use of bi-prediction as the prediction using a plurality of reference images. However, a technique called intra block copy can also be applied as the prediction using a plurality of reference images. In the intra block copy, a reference image in the same frame as the present frame is used for predicting the target image block in the present frame.
The foregoing may be provided by a program that causes a computer to perform each process performed by the image encoding device 1 and a program that causes the computer to perform each process performed by image decoding device 2. Also, the programs may be recorded in a computer readable medium. The programs can be installed on the computer by using the computer readable medium. Here, the computer readable medium in which the programs are recorded may be a non-transitory recording medium. The non-transitory recording medium is not particularly limited and may be a recording medium such as a CD-ROM or a DVD-ROM, for example. Also, circuits for performing the processes performed by the image encoding device 1 may be integrated to constitute the image encoding device 1 as a semiconductor integrated circuit (chipset, SoC). Similarly, circuits for performing the processes performed by the image decoding device 2 may be integrated to constitute the image decoding device 2 as a semiconductor integrated circuit (chipset, SoC).
The embodiments have been described in detail above with reference to the drawings. Specific configurations are not limited to the above-described configurations, and various design changes, and the like are possible within the scope not deviating from the gist.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2018-072451 | Apr 2018 | JP | national |
This application is a Continuation of U.S. patent application Ser. No. 17/062,306 filed Oct. 2, 2020, which is a continuation of PCT/JP2019/014364 filed on Mar. 29, 2019, which claims priority to Japanese Application No. 2018-072451 filed on Apr. 4, 2018. The entire contents of these applications are incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 17062306 | Oct 2020 | US |
| Child | 18172974 | US | |
| Parent | PCT/JP2019/014364 | Mar 2019 | US |
| Child | 17062306 | US |
