JP2013090253A - Image processor and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve efficiency of in-screen prediction while suppressing increase in a calculation amount.SOLUTION: An image processor for performing in-screen prediction using plural prediction directions comprises: a sorting unit for sorting plural reference pixels into plural groups on the basis of possibility of being referred to; a filter determination unit for determining different filters for respective sorted groups; a filtering unit which filters the pixel values of the reference pixels included in each group using the filter determined by the filter determination unit; and a generation unit for generating a predictive image of the in-screen prediction using the pixel values filtered by the filtering unit.

Description

  The present invention relates to an image processing apparatus and a program for performing intra prediction using a plurality of prediction directions.

  In recent years, with the advance of multimedia, there are many opportunities to handle moving images. Image data has a large amount of information, and a moving image, which is a set of images, has an information amount exceeding 1 Gbit / sec in high-definition (1920 × 1080) broadcasting, which is the mainstream of television broadcasting.

  Therefore, the moving image data is H.264. H.264 / AVC (Advanced Video Coding) and standardization work HEVC (High Efficiency Video Coding) and the like are used for transmission and storage by compressing the amount of information.

  H., an encoding standard technology. H.264 / AVC uses a tool such as orthogonal transform, inter-screen prediction, intra-screen prediction, arithmetic coding, deblocking filter, and the like to achieve compression of about 1/100 of a moving image. Intra prediction, which is one of the tools, performs intra prediction by determining an optimal prediction mode from a plurality of prediction modes using a pixel value adjacent to the processing target block as a reference pixel.

  H. As a technique added to improve image quality in H.264 / AVC, there is 8 × 8 intra-screen prediction (Non-Patent Document 1).

  FIG. 3 is a diagram for explaining H.264 / AVC 8 × 8 intra-screen prediction. FIG. Pixels A to Y shown in FIG. 1 indicate reference pixels, and white circle pixels without characters indicate pixels of a processing target block. Here, before performing intra prediction, a low-pass filter of [1/4, 1/2, 1/4] is applied to the reference pixels A to Y. Hereinafter, the low-pass filter is also referred to as LPF.

  For example, when referring to the reference pixel C, a pixel value of B × 0.25 + C × 0.5 + D × 0.25 is used. For the reference pixels A and Y at both ends, [1/4, 3/4] filters are used. For example, when referring to the reference pixel A, a pixel value of B × 0.25 + A × 0.75 is used.

  Even in the HEVC system, which is being standardized, H.264 / AVC 8 × 8 intra-screen prediction uses reference pixel filtering not only for block size but before intra-screen prediction, so that standardization is being promoted.

  In this HEVC method, a reference pixel filtering method has been proposed in which filtering is performed when the absolute value of the difference between the pixel values before and after the pixel of interest is equal to or smaller than a threshold value, and filtering is not performed when the absolute value is larger than the threshold value (non-null). Patent Document 2).

  In the proposed filtering method, for example, when referring to the pixel D shown in FIG. 1, if the value of abs (CE) is equal to or less than the threshold, the filtered pixel value is used, and abs (CE). When the value of exceeds the threshold, the value of D is used as it is.

Supervised by Satoshi Okubo, “Revised Third Edition H.264 / AVC Textbook”, Impress R & D, p. 264-268, January 1, 2009 Jane Zhao, `` On intra prediction (JCTVC-E437) '', Joint Collaborative Team on Video Coding (JCT-VC) of ITU-T SG16 WP3 and ISO / IEC JTC1 / SC29 / WG11 5th Meeting: Geneva, CH, 16-23 March, 2011

  Here, the purpose of filtering the reference pixels for intra prediction is to reduce noise contained in the reference pixels. When filtering is not performed, if noise is included in the reference pixel, the noise is extended in the prediction direction of intra prediction. In this case, since a striped pattern that does not exist in the processing target block (predicted block) is generated by intra prediction, the efficiency of intra prediction is reduced.

  Further, when an edge exists in the prediction direction of the intra prediction, the value of the edge is changed by applying the LPF, so that the efficiency of the intra prediction is reduced. This is because an edge different from the edge that originally existed is generated by intra prediction.

  Thus, the noise reduction effect and the fact that the efficiency of intra prediction is reduced when there is an edge is a two-sided relationship.

  For this reason, in Non-Patent Document 2, the efficiency of intra prediction is improved by determining noise and edge using a pixel value adjacent to a reference pixel.

  However, Non-Patent Document 2 has a problem in that the amount of calculation increases because determination is performed using a pixel value adjacent to the reference pixel for each reference pixel.

  Therefore, the present invention has been made in view of the above problems, and an object thereof is to provide an image processing apparatus and a program capable of improving the efficiency of intra prediction while suppressing an increase in calculation amount.

  An image processing apparatus according to an aspect of the present invention is an image processing apparatus that performs intra prediction using a plurality of prediction directions, and classifies a plurality of reference pixels into a plurality of groups based on a possibility of being referred to. Filtering for filtering the pixel values of the reference pixels included in each group using a filter, a filter determination unit that determines a different filter for each classified group, and a filter determined by the filter determination unit And a generation unit that generates a predicted image of the intra prediction using the pixel value after filtering by the filtering unit.

  The classification unit may classify the plurality of reference pixels based on the number of reference prediction directions.

  Further, the classification unit may classify the plurality of reference pixels depending on whether or not they are adjacent to the processing target block.

  The classification unit may further classify reference pixels adjacent to the processing target block or pixels not adjacent to the processing target block into a plurality of groups.

  The filter determination unit may determine a low-pass filter having a wider pass band as the reference pixel in the group is more likely to be referred to.

  The filter determination unit determines a noise removal filter for a first group including a reference pixel that is most likely to be referred to, and for a group other than the first group, a reference within the group. You may determine to a low pass filter with a wide pass band, so that a possibility that a pixel will be referred to is high.

  A program according to another aspect of the present invention includes a classification step for classifying a plurality of reference pixels into a plurality of groups based on a possibility of reference, and a filter determination for determining a different filter for each classified group. A filtering step for filtering the pixel values of the reference pixels included in each group using the determined filter, and a generation step for generating a prediction image for intra prediction using the pixel values after filtering And let the computer run.

  According to the present invention, it is possible to improve the efficiency of in-screen prediction while suppressing an increase in the amount of calculation.

H. The figure for demonstrating the prediction in H.264 / AVC 8x8. 1 is a block diagram illustrating an example of a schematic configuration of an image encoding device according to Embodiment 1. FIG. The block diagram which shows an example of a structure of an intra estimation part. The figure which shows the prediction direction of 8 * 8 in-screen prediction encoding. H. The figure which shows the prediction direction which refers to the reference pixel Y in H.264 / AVC. H. The figure which shows the prediction direction which refers the reference pixel U in H.264 / AVC. H. The figure which shows the prediction direction which refers the reference pixel Q in H.264 / AVC. H. The figure which shows the prediction direction which refers to the reference pixel M in H.264 / AVC. The figure which shows an example which classify | categorizes a reference pixel into two. The figure which shows an example which classify | categorizes a reference pixel into three. The figure which shows an experimental result. The flowchart which shows an example of the prediction process in a screen in Example 1. FIG. FIG. 6 is a block diagram illustrating an example of a configuration of an image encoding device according to a second embodiment.

  Hereinafter, embodiments will be described in detail with reference to the accompanying drawings.

[Example 1]
<Configuration>
FIG. 2 is a block diagram illustrating an example of a schematic configuration of the image encoding device 100 according to the first embodiment. In the example illustrated in FIG. 2, the image encoding device 100 includes a prediction error signal generation unit 101, an orthogonal transform unit 102, a quantization unit 103, an entropy encoding unit 104, an inverse quantization unit 105, an inverse orthogonal transform unit 106, and a decoding The image generation unit 107, the deblocking filter unit 108, the decoded image storage unit 109, the intra prediction unit 110, the inter prediction unit 111, the motion vector calculation unit 112, the encoding control and header generation unit 113, and the predicted image selection unit 114 are included. An outline of each part will be described below.

  The prediction error signal generation unit 101 acquires block data obtained by dividing an encoding target image of input moving image data into blocks of 16 × 16 pixels, for example.

  The prediction error signal generation unit 101 generates a prediction error signal from the block data and the block data of the prediction image output from the prediction image selection unit 114. The prediction error signal generation unit 101 outputs the generated prediction error signal to the orthogonal transformation unit 102.

  The orthogonal transform unit 102 performs orthogonal transform processing on the input prediction error signal. The orthogonal transform unit 102 outputs a signal separated into horizontal and vertical frequency components by the orthogonal transform process to the quantization unit 103.

  The quantization unit 103 quantizes the output signal from the orthogonal transform unit 102. The quantization unit 103 reduces the code amount of the output signal by quantization, and outputs this output signal to the entropy encoding unit 104 and the inverse quantization unit 105.

  The entropy encoding unit 104 performs entropy encoding on the output signal from the quantization unit 103 and outputs the result. Entropy coding is a method of assigning variable-length codes according to the appearance frequency of symbols.

  The inverse quantization unit 105 dequantizes the output signal from the quantization unit 103 and then outputs the output signal to the inverse orthogonal transform unit 106. The inverse orthogonal transform unit 106 performs an inverse orthogonal transform process on the output signal from the inverse quantization unit 105 and then outputs the output signal to the decoded image generation unit 107. By performing decoding processing by the inverse quantization unit 105 and the inverse orthogonal transform unit 106, a signal having the same level as the prediction error signal before encoding is obtained.

  The decoded image generation unit 107 adds the block data of the image subjected to motion compensation by the inter prediction unit 111 and the prediction error signal decoded by the inverse quantization unit 105 and the inverse orthogonal transform unit 106. The decoded image generation unit 107 outputs the decoded image block data generated by the addition to the deblocking filter unit 108.

  The deblocking filter unit 108 applies a filter for reducing block distortion to the decoded image output from the decoded image generation unit 107 and outputs the filtered image to the decoded image storage unit 109.

  The decoded image storage unit 109 stores the input block data of the decoded image as new reference image data, and outputs the data to the intra prediction unit 110, the inter prediction unit 111, and the motion vector calculation unit 112.

  The intra prediction unit 110 generates a predicted image from reference pixels that have already been encoded for the processing target block of the encoding target image. Details of the intra prediction unit 110 will be described later with reference to FIG.

  The inter prediction unit 111 performs motion compensation on the reference image data acquired from the decoded image storage unit 109 with the motion vector provided from the motion vector calculation unit 112. Thereby, block data as a motion-compensated reference image is generated.

  The motion vector calculation unit 112 obtains a motion vector using the block data in the encoding target image and the reference image acquired from the decoded image storage unit 109. The motion vector is a value indicating a spatial deviation in units of blocks obtained using a block matching technique for searching for a position most similar to the processing target block from the reference image in units of blocks. The motion vector calculation unit 112 outputs the obtained motion vector to the inter prediction unit 111.

  The block data output from the intra prediction unit 110 and the inter prediction unit 111 are input to the predicted image selection unit 114.

  The predicted image selection unit 114 selects one of the block data acquired from the intra prediction unit 110 and the inter prediction unit 111 as a predicted image. The selected prediction image is output to the prediction error signal generation unit 101.

  The encoding control and header generation unit 113 performs overall encoding control and header generation. The encoding control and header generation unit 113 notifies the intra prediction unit 110 of presence / absence of slice division, notifies the deblocking filter unit 108 of presence / absence of deblocking filter, and provides a reference image to the motion vector calculation unit 112. Notification of restrictions, etc.

  The encoding control and header generation unit 113 uses the control result, for example, H.264. H.264 / AVC header information is generated. The generated header information is output to the entropy encoding unit 104, and is output as a stream together with image data and motion vector data.

<Configuration of intra prediction unit>
Next, the configuration of the intra prediction unit 110 will be described. FIG. 3 is a block diagram illustrating an example of the configuration of the intra prediction unit 110. The intra prediction unit 110 illustrated in FIG. 3 includes a classification unit 201, a filter determination unit 202, a filtering unit 203, and a generation unit 204.

  The classification unit 201 classifies the plurality of reference pixels used in the intra prediction into a plurality of groups based on the possibility that the reference pixels are referred to. The classification unit 201 classifies the reference pixels based on, for example, the number of prediction directions to be referenced and whether or not they are adjacent to the processing target block. Details of the reference pixel classification will be described later with reference to FIGS.

  When the plurality of reference pixels are classified into a plurality of groups, the classification unit 201 outputs classification information indicating to which group the reference pixel belongs to the filter determination unit 202.

  When the filter determination unit 202 acquires the classification information, the filter determination unit 202 determines a different filter for each classified group. For example, the filter determination unit 202 determines an LPF having a wider passband as the reference pixel included in the group is more likely to be referred to. Further, the filter determination unit 202 may determine the noise removal filter as a group including reference pixels that are most likely to be referred to. Details of the filter determination unit 202 will also be described later.

  The filter determination unit 202 outputs filter information indicating which filter is determined for which group to the filtering unit 203.

  The filtering unit 203 acquires the pixel value of the reference pixel from the decoded image storage unit 109. Further, the filtering unit 203 acquires filter information from the filter determination unit 202 and grasps which filter is used for which reference pixel.

  The filtering unit 203 filters the pixel values of the reference pixels included in each group using the filter determined by the filter determining unit 202. The filtering unit 203 outputs the filtered pixel value to the generation unit 204.

  The generation unit 204 uses the filtered pixel values acquired from the filtering unit 203 to generate a predicted image for intra-screen prediction. The generation unit 204 includes, for example, a predicted image generation unit 205 and a prediction mode determination unit 206.

  The predicted image generation unit 205 generates a predicted image for each prediction mode corresponding to the prediction direction of intra-screen prediction using the filtered pixel value as a predicted value.

  The prediction mode determination unit 206 determines a prediction mode with the smallest difference between the prediction image for each prediction mode and the processing target block. The prediction mode determination unit 206 outputs the prediction image of the determined prediction mode to the prediction image selection unit 114.

<Classification of reference pixels>
Next, reference pixel classification by the classification unit 201 will be described. As an example of the reference pixel, H.264 is used. A description will be given using an H.264 / AVC 8 × 8 block (see FIG. 1), but the block size is not limited to this, and may be a block size such as 4 × 4 or 16 × 16. Further, the present invention can be applied to both luminance and color difference blocks. Further, the present invention is not limited to a square block, and can be applied to a rectangular block such as 8 × 16.

  First, H. A prediction direction in the case of H.264 / AVC will be described. FIG. 4 is a diagram illustrating a prediction direction of 8 × 8 intra prediction encoding. In the example shown in FIG. 4, the prediction direction corresponding to each prediction mode is shown. The gray circles shown in FIG. 4 indicate reference pixels, and the white circles indicate processing target pixels. The arrow indicates the prediction direction.

  For example, the prediction mode 0 indicates that the reference pixel P is a predicted value of pixels (N0, N1,..., N7) in the same column.

  In the following, although the prediction direction shown in FIG. 4 is used, the classification method described below can be similarly applied even when there are other prediction directions without being limited to the prediction direction shown in FIG.

(Classification based on the number of prediction directions)
For example, the classification unit 201 classifies reference pixels based on the number of referenced prediction directions. This classification method is based on the idea that when the number of prediction directions to be referred to is large and there is an edge in any one direction, the efficiency of intra prediction is reduced by LPF.

  FIG. It is a figure which shows the prediction direction which refers to the reference pixel Y in H.264 / AVC. As shown in FIG. 5, the reference pixel Y is referred to in the case of the prediction direction of the prediction mode 3 shown in FIG.

  FIG. It is a figure which shows the prediction direction which refers to the reference pixel U in H.264 / AVC. As shown in FIG. 6, the reference pixel U is referred to in the prediction directions of the prediction modes 3 and 7 shown in FIG. 4.

  FIG. It is a figure which shows the prediction direction which refers the reference pixel Q in H.264 / AVC. As shown in FIG. 7, the reference pixel Q is referred to in the case of the prediction directions of the prediction modes 0, 3, and 7 shown in FIG.

  FIG. It is a figure which shows the prediction direction which refers to the reference pixel M in H.264 / AVC. As shown in FIG. 8, the reference pixel M is referred to in the prediction directions of the prediction modes 0, 3, 4, 5, 6, and 7 shown in FIG. 4.

  For example, with respect to the reference pixel M, when an edge exists in any of the six directions shown in FIG.

  For the reference pixel U, when an edge exists in one of the two directions shown in FIG. 6, the effect of intra prediction is reduced when LPF is applied, but it is not referenced in other prediction directions. The effect of the presence or absence of LPF is small.

  Here, assuming that the direction of the edge existing in the image is random, the probability that the prediction direction of the intra prediction matches the direction of the edge is 3 for the reference pixel M than for the reference pixel U. (= 6 (number of prediction directions of reference pixel M) / 2 (number of prediction directions of reference pixel U)) times higher.

Therefore, the classification unit 201 can classify reference pixels as follows based on the number of referenced prediction directions.
Group 1: V, W, X, Y
Group 2: A, R, S, T, U
Group 3: H, I, Q
Group 4: B, C, D, F, J
Group 5: E, G, L, N, P
Group 6: K, M, O
Group 1 has a number of referenced prediction directions of 1, group 2 has a number of referenced prediction directions of 2, group 3 has a number of referenced prediction directions of 3, group 4 The number of referenced prediction directions is 4, group 5 has 5 referenced prediction directions, and group 6 has 6 referenced prediction directions.

  Moreover, the classification | category part 201 may classify a some group collectively from the groups 1-6. For example, the classification unit 201 may group 1 and 2 into one group, group 3 and 4 into one group, and groups 5 and 6 into one group.

  Thereby, it is possible to classify the reference pixels in consideration of the probability that an edge in the direction matching the prediction direction exists.

  Note that the number of groups to be classified differs depending on the block size and prediction direction of the encoding standard technique, but if the block size and what kind of prediction direction is determined, the classification unit 201 should uniquely classify the reference pixels. Can do.

(Classification based on whether or not they are adjacent)
For example, the classification unit 201 classifies reference pixels based on whether or not they are adjacent to the processing target block. This classification method is based on the idea that reference pixels that are directly adjacent to the processing target block are referred to more prediction directions than reference pixels that are not directly adjacent.

FIG. 9 is a diagram illustrating an example of classifying the reference pixels into two. In the example illustrated in FIG. 9, the classification unit 201 classifies a group 2 that is directly adjacent to the processing target block and a group 1 that is not directly adjacent.
Group 1: Reference pixels not directly adjacent to the processing target block Group 2: Reference pixels directly adjacent to the processing target block FIG. 10 is a diagram illustrating an example of classifying the reference pixels into three categories. In the example illustrated in FIG. 10, the classification unit 201 classifies groups 2 and 3 adjacent to the processing target block and group 1 that is not directly adjacent to the processing target block.
Group 1: Reference pixel group not directly adjacent to the processing target block 2: Reference pixel group directly adjacent to the vicinity of the upper left of the processing target block 3: Directly adjacent to the processing target block, and the reference pixel group 2 other than the group 2 is, for example, Assume that the reference pixels are H, I, and J shown in FIG. The classification unit 201 may further classify the reference pixels directly adjacent to the processing target block and / or the reference pixels not directly adjacent to the processing target block into a plurality of groups.

  Thus, the reference pixels can be easily classified into a plurality of groups based on the possibility of reference.

  The classification unit 201 may classify the reference pixels by any one of the classification method based on the number of prediction directions and the classification based on whether or not they are adjacent. Further, the classification unit 201 can select and classify which classification is performed based on the encoding rate or the like.

<Filter determination>
Next, filter determination by the filter determination unit 202 will be described. The filter determination unit 202 determines a filter having a smaller change in edge component as the reference pixel in the group is more likely to be referenced.

(LPF)
For example, the filter determination unit 202 determines the filter so that the LPF having a wider passband becomes higher as the reference pixel in the group is more likely to be referred to. For example, the filter determining unit 202 determines that the filter has a narrower passband as the group number classified by the classifying unit 201 is smaller.

  For example, it is assumed that the classification unit 201 performs classification based on the number of prediction directions. In this case, the filter determination unit 202 determines LPFs of [1/4, 1/2, 1/4] for groups 1 and 2 and [1/8, 3/4] for groups 3 and 4. , 1/8] LPF, and for groups 5 and 6, [1/16, 7/8, 1/16] LPF.

  For example, it is assumed that the classification unit 201 performs classification into three groups based on whether or not they are adjacent to each other. In this case, the filter determination unit 202 determines [1/4, 1/2, 1/4] LPF for group 1 and [1/8, 3/4, 1/8] for group 2. The LPF of [1/16, 7/8, 1/16] for group 3 is determined. The LPF is merely an example, and the number of taps and the passband may be changed as appropriate.

  That is, an LPF having a wide passband is used for a reference pixel that is highly likely to be referenced. An LPF with a wide passband has a smaller change in the pixel value of interest than an LPF with a narrow passband. Therefore, a reference pixel that is highly likely to be referenced can reduce a decrease in prediction efficiency even when an edge exists.

  On the other hand, a reference pixel having a low possibility of being referenced uses an LPF having a narrow passband because the probability that an edge matching the applied prediction direction exists is low. Thereby, a sufficient noise reduction effect can be expected.

(LPF + noise removal filter)
For example, the filter determination unit 202 determines a noise removal filter for a group including a reference pixel that is most likely to be referred to, and may reference a reference pixel in the group for other groups. The filter is determined so as to be an LPF with a wider passband as the value of is higher.

  The filter determination unit 202 determines, for example, a noise removal filter for the group with the largest group number classified by the classification unit 201, and for other groups, a filter with a narrower passband as the group number is smaller. Decide to be.

  Examples of the noise removal filter include a median filter, a moving average filter, an ε filter, and an MTM (Modified Trimmed Mean) filter. For example, a median filter having a small number of filter taps is suitable as a noise removal filter from the viewpoint of calculation amount.

  For example, it is assumed that the classification unit 201 performs classification based on the number of prediction directions. In this case, the filter determination unit 202 determines LPFs of [1/4, 1/2, 1/4] for groups 1 and 2 and [1/8, 3/4] for groups 3 and 4. , 1/8] LPF and, for groups 5 and 6, a 3-tap median filter.

  For example, it is assumed that the classification unit 201 performs classification into three groups based on whether or not they are adjacent to each other. In this case, the filter determination unit 202 determines [1/4, 1/2, 1/4] LPF for group 1 and [1/8, 3/4, 1/8] for group 2. ], And for group 3, a 3-tap median filter is determined. The LPF is merely an example, and the number of taps and the passband may be changed as appropriate.

  In general, since the noise removal filter has a smaller edge value change than the LPF, the same effect as the LPF having a wide passband can be obtained. In addition, although the noise removal filter may have a larger calculation amount than the LPF, for example, when a 3-tap median filter is used, the increase in the calculation amount from the LPF is small. As described above, the filter determination unit 202 prepares a plurality of sets of filters that increase the passband as the number of referenced prediction directions increases, and narrow the passband as the number of referenced prediction directions decreases. Keep it. The filter determination unit 202 can determine the filter set to be used if the type of the input moving image is given by confirming the effect of each filter set corresponding to various patterns in advance. Further, the filter determination unit 202 may adaptively change the filter set according to the determination of the pattern type of the input moving image.

  Therefore, in the first embodiment, it is not necessary to perform edge determination for each reference pixel, and by performing filtering according to the classification of the reference pixel, a combination of the effect of maintaining the edge and the effect of reducing noise is used. It is possible to improve the prediction performance of intra prediction.

<Experimental result>
Next, a description will be given of an encoding experiment performed by the software encoder HM3.0 used in the HEVC standardization work and the proposed method in which the method of the first embodiment is implemented in HM3.0.

  The system of the first embodiment is classified into two groups depending on whether or not it is directly adjacent to the processing target block (see FIG. 9), and [1/4, 1/2, 1/4] LPF is assigned to group 1, [1/8, 3/4, 1/8] LPF is applied to group 2.

In the experiment, each of the images A to D is encoded, and HM3.0 is used as an anchor, and the performance of the proposed method is represented by BD-RATE.
Image A: City landscape image B: Indoor person image C: Running horse image D: Indoor person image BD-RATE is "G. Bjontegaard," Calculation of average PSNR differences between RD-Curves, "ITU -T SG16 Q.6 Document, VCEG-M33, April 2001 ”is an index for comparing the encoding performance proposed. When the value of BD-RATE is 0, the performance of the proposed method is equal to that of the anchor. The smaller the value of BD-RATE, the better the performance of the proposed method than the anchor.

  FIG. 11 is a diagram showing experimental results. As shown in FIG. 11, it can be seen that the proposed scheme has improved encoding performance over HM3.0. Therefore, in the first embodiment, it is possible to improve the prediction performance of intra prediction while suppressing an increase in calculation amount.

<Operation>
Next, the operation of the image encoding device 100 according to the first embodiment will be described. FIG. 12 is a flowchart illustrating an example of the intra-screen prediction process in the first embodiment.

  In step S101 illustrated in FIG. 12, the classification unit 201 classifies reference pixels into a plurality of groups based on the possibility of reference. For example, the classification unit 201 may determine a classification method according to the block size of the processing target block and classify by this classification method. As described above, the classification method may use either a classification based on the number of prediction directions or a classification based on whether or not adjacent to each other.

  In step S102, the filter determination unit 202 determines different filters according to the groups determined by the classification unit 201. For example, the filter determination unit 202 determines an LPF having a wider passband as the reference pixel in the group is more likely to be referenced.

  In step S103, the filtering unit 203 filters the pixel value of the reference pixel using the filter determined by the filter determining unit 202.

  In step S104, the generation unit 204 generates a prediction image corresponding to a plurality of prediction modes, and determines an optimal prediction mode. The generation unit 204 outputs the predicted image of the determined prediction mode to the predicted image selection unit 114.

  As described above, according to the first embodiment, it is possible to improve the prediction efficiency of intra prediction while suppressing an increase in calculation amount.

[Example 2]
FIG. 13 is a block diagram illustrating an example of the configuration of the image encoding device 300 according to the second embodiment. The image encoding device 300 is an example of a device in which the image encoding process described in the first embodiment is implemented by software.

  As illustrated in FIG. 13, the image encoding device 300 includes a control unit 301, a main storage unit 302, an auxiliary storage unit 303, a drive device 304, a network I / F unit 306, an input unit 307, and a display unit 308. These components are connected to each other via a bus so as to be able to transmit and receive data.

  The control unit 301 is a CPU that controls each device, calculates data, and processes in a computer. The control unit 301 is an arithmetic device that executes a program for image encoding processing including filter processing stored in the main storage unit 302 or the auxiliary storage unit 303. The control unit 301 receives data from the input unit 307 and the storage device, calculates and processes the data, and outputs the data to the display unit 308 and the storage device.

  The control unit 301 can realize the filter processing described in each embodiment by executing a program of image encoding processing including filter processing.

  The main storage unit 302 is a ROM (Read Only Memory), a RAM (Random Access Memory), or the like. The main storage unit 302 is a storage device that stores or temporarily stores programs and data such as OS (Operating System) and application software that are basic software executed by the control unit 301.

  The auxiliary storage unit 303 is an HDD (Hard Disk Drive) or the like, and is a storage device that stores data related to application software and the like.

  The drive device 304 reads the program from the recording medium 305, for example, a flexible disk, and installs it in the storage device.

  A predetermined program is stored in the recording medium 305, and the program stored in the recording medium 305 is installed in the image encoding device 300 via the drive device 304. The installed predetermined program can be executed by the image encoding device 300.

  The network I / F unit 306 is a peripheral having a communication function connected via a network such as a LAN (Local Area Network) or a WAN (Wide Area Network) constructed by a data transmission path such as a wired and / or wireless line. This is an interface between the device and the image encoding device 300.

  The input unit 307 includes a keyboard having cursor keys, numeric input, various function keys, and the like, a mouse and a slice pad for selecting keys on the display screen of the display unit 308, and the like. The input unit 307 is a user interface for a user to give an operation instruction to the control unit 301 or input data.

  The display unit 308 is configured by a CRT (Cathode Ray Tube), an LCD (Liquid Crystal Display), or the like, and performs display according to display data input from the control unit 301.

  The decoded image storage unit 109 illustrated in FIG. 2 is realized by, for example, the main storage unit 302 or the auxiliary storage unit 303, and the configuration other than the decoded image storage unit 109 illustrated in FIG. This can be realized by the main storage unit 302.

  The program executed by the image encoding device 300 has a module configuration including each unit described in the first embodiment. As actual hardware, when the control unit 301 reads a program from the auxiliary storage unit 303 and executes it, one or more of the above-described units are loaded onto the main storage unit 302, and one or more of the units are main. It is generated on the storage unit 302.

  Thus, the in-screen prediction process described in the first embodiment may be realized as a program for causing a computer to execute. The filter processing described above can be realized by installing this program from a server or the like and causing the computer to execute it.

  It is also possible to record the program in the recording medium 305 and cause the computer or portable terminal to read the recording medium 305 on which the program is recorded, thereby realizing the filter processing described above. The recording medium 305 is a recording medium that records information optically, electrically, or magnetically, such as a CD-ROM, a flexible disk, or a magneto-optical disk, and information is electrically stored such as a ROM or flash memory. Various types of recording media such as a semiconductor memory for recording can be used. Further, the filtering process described in each of the above embodiments may be implemented in one or a plurality of integrated circuits.

  In the embodiment, the image coding apparatus has been described as an example. However, the present invention can be similarly applied to an image decoding apparatus that performs intra prediction using a plurality of prediction modes. The image encoding device and the image decoding device are collectively referred to as an image processing device. In the above embodiment, H. Although H.264 / AVC has been described as an example, any image processing technique that performs intra prediction using a plurality of prediction directions with respect to a reference pixel can be applied.

  Each embodiment has been described in detail above. However, the present invention is not limited to the specific embodiment, and various modifications and changes other than the above-described modification are possible within the scope described in the claims. .

100, 300 Image coding apparatus 101 Predicted image generation unit 102 Orthogonal transformation unit 103 Quantization unit 104 Entropy coding unit 105 Inverse quantization unit 106 Inverse orthogonal transformation unit 107 Decoded image generation unit 108 Deblocking filter unit 109 Decoded image storage unit 110 Intra Prediction Unit 111 Inter Prediction Unit 112 Motion Vector Calculation Unit 113 Coding Control and Header Generation Unit 114 Predictive Image Selection Unit 201 Classification Unit 202 Filter Determination Unit 203 Filtering Unit 204 Generation Unit 205 Predictive Image Generation Unit 206 Prediction Mode Determination Unit 301 Control unit 302 Main storage unit 303 Auxiliary storage unit 304 Drive device 306 Network I / F unit 307 Input unit 308 Display unit

Claims (7)

An image processing apparatus that performs intra-screen prediction using a plurality of prediction directions,
A classification unit that classifies a plurality of reference pixels into a plurality of groups based on a possibility of being referred to; and
A filter determination unit that determines a different filter for each classified group;
A filtering unit that performs filtering on pixel values of the reference pixels included in each group using the filter determined by the filter determination unit;
Using the pixel value after filtering by the filtering unit, a generation unit that generates a prediction image of the intra-screen prediction;
An image processing apparatus comprising: The classification unit includes:
The image processing apparatus according to claim 1, wherein the plurality of reference pixels are classified based on the number of prediction directions to be referred to. The classification unit includes:
The image processing apparatus according to claim 1, wherein the plurality of reference pixels are classified according to whether or not they are adjacent to a processing target block. The classification unit includes:
The image processing apparatus according to claim 3, wherein reference pixels adjacent to the processing target block or pixels not adjacent to the processing target block are further classified into a plurality of groups. The filter determination unit
5. The image processing apparatus according to claim 1, wherein the image processing apparatus determines a low-pass filter having a wider pass band as the reference pixel in the group is more likely to be referred to. The filter determination unit
There is a possibility that a noise removal filter is determined for the first group including a reference pixel that is most likely to be referenced, and a reference pixel in the group is referred to for a group other than the first group. The image processing apparatus according to claim 1, wherein the image processing apparatus determines a low-pass filter having a wider pass band as the height is higher. A classification step of classifying a plurality of reference pixels into a plurality of groups based on a reference possibility;
A filter determination step for determining different filters for each classified group;
A filtering step of filtering the pixel values of the reference pixels included in each group using the determined filter;
A program for causing a computer to execute a generation step of generating a predicted image for intra prediction using pixel values after filtering.

Images