JP2007251923A - Image encoding device and image coding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image encoding device capable of limiting the number of objects in a prediction direction for prediction evaluation value calculation at prediction encoding to derive the optimal prediction direction from the obtained prediction direction evaluation values and efficiently reducing the quantity of calculations by limiting the prediction direction. <P>SOLUTION: This image encoding device comprises a prediction direction designation means 101 for designating a prediction direction for a target encoding block, a first prediction evaluation value calculation means 102 for calculating a prediction evaluation value in the designated prediction direction, a prediction direction extraction means 103 for determining a preferred prediction direction based on the calculated prediction evaluation value and extracting the related prediction direction, a second prediction evaluation value calculation means 104 for calculating the prediction evaluation value in each extracted prediction direction, a prediction direction determination means 105 for determining the best prediction direction based on the calculated prediction evaluation value in each prediction direction and a prediction mode conversion means 106 for converting this best prediction direction to a prediction mode. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

The present invention relates to an image encoding device and an image encoding method.

  Predictive coding applied in a conventional image coding apparatus includes, for example, intra-prediction coding that performs prediction within a frame, and inter-prediction coding that performs prediction between consecutive frames in a moving image. It is also used in international standards and standards.

  FIG. 11 shows an intra prediction code having a block of 4 pixels × 4 lines described in, for example, Non-Patent Document 1 (hereinafter referred to as H.264 system, which may be referred to as MPEG-4AVC system). This shows the prediction mode of conversion. In the H.264 system, which is an international standard for image coding, there is a prediction mode called DC prediction in addition to the prediction mode shown in FIG. 11, but description of DC prediction not directly related to the present invention will be omitted. Here, the numbers indicating the prediction modes in the figure are assigned values 0 to 7, which are different from the values described in Non-Patent Document 1. Hereinafter, the direction of the arrow is referred to as a prediction direction and basically corresponds to the prediction mode. Let me explain.

  Similarly, the H.264 method has twice the performance in terms of image quality compared to the MPEG-2 method, which is an international standard for image coding, but because it performs coding processing at the same frame rate. Is said to require a calculation amount of 10 times or more. One of the reasons why the calculation amount of the H.264 system is larger than that of the MPEG-2 system is predictive encoding performed by this intra encoding.

  In the H.264 system, as shown in FIG. 12, encoding is performed in units called macroblocks composed of 16 pixels × 16 lines of pixels. Further, intra prediction is performed in block units (hereinafter referred to as encoding target blocks) in which the macroblock is divided into 16 4 pixels × 4 lines or 4 8 pixels × 8 lines. Hereinafter, an encoding target block of 4 pixels × 4 lines is referred to as a 4 × 4 block, and an encoding target block of 8 pixels × 8 lines is referred to as an 8 × 8 block.

  In the H.264 scheme, a prediction value is obtained for a block to be encoded in intra coding by using neighboring pixels that have been encoded, and an error signal is encoded with the prediction value, so-called prediction code. Do. In the MPEG-2 system, intra encoding is performed, but in the MPEG-2 system, encoding is performed as it is without using prediction for pixels of an encoding target block.

  In the H.264 system, intra prediction encoding in the case of a 4-pixel × 4-line encoding target block will be described. In the 4 × 4 block composed of pixels indicated by uppercase letters A to P in FIG. 11, prediction of the prediction direction is performed using peripheral pixels (pixels indicated by lowercase letters a to m in the figure) that have been encoded. A pixel value (prediction value) is calculated, and a prediction error signal between the encoding target pixel and the prediction pixel is encoded. For example, prediction pixels in the case of prediction directions 1 and 5 in the figure are shown in FIG.

  The prediction values in the prediction directions other than the prediction directions 1 and 5 are calculated from the surrounding pixels a to m that have been encoded for each encoding target pixel. Similarly, in the case of an 8 × 8 block, the prediction direction can be made to correspond to the surrounding pixels that have been encoded, and the prediction pixel in the prediction direction can be calculated from the surrounding pixels.

  A method for obtaining an optimal prediction direction in intra prediction will be described. As a generally performed method, there is a method in which a prediction evaluation value in each prediction direction is calculated, and a prediction direction that minimizes the prediction evaluation value is used for encoding as an optimal prediction direction. At this time, as the prediction evaluation value, a sum value obtained by adding the absolute difference value or the square difference value between all the encoding target pixels in the encoding target block and the prediction pixel (prediction value) is used. There are many. In order to perform such calculation of the prediction evaluation value for all prediction directions, a large amount of calculation is required.

  FIG. 14 is a flowchart illustrating a prediction mode selection processing procedure based on the prediction evaluation value in the prediction direction of intra prediction encoding. If the prediction direction number is n and the prediction evaluation value of the prediction direction is En, in FIG. 14, the prediction evaluation values (E0 to E7) of all prediction directions are calculated in step ST101, and step ST102. The procedure for determining the optimum prediction direction based on the prediction evaluation values of all the prediction directions and converting the optimum prediction direction to the corresponding prediction mode in step ST103 is shown.

  As described above, predictive coding is performed in intra coding in the H.264 scheme, but an optimal prediction mode is selected from eight types of prediction directions except for DC prediction. The calculation of prediction evaluation values for selection is performed for 16 blocks in the case of 4 × 4 blocks per macroblock and 4 blocks in the case of 8 × 8 blocks, and all predictions are performed in each block. Since the predicted evaluation value is calculated with respect to the direction, the calculation amount is enormous.

ITU-T Recommendation H.264 Recommendation Advanced video coding for generic audiovisual services 8.3 Intra prediction process, Figure 8-1 (2005/3)

  In the conventional image coding apparatus, as described in the section of the prior art, prediction coding is performed in intra coding. However, since the prediction evaluation values are calculated in all prediction directions, the amount of calculation is large. Has the problem of becoming enormous.

  The present invention has been made to solve the above-described problems. When predictive coding is performed, the number of prediction direction targets for calculating a prediction evaluation value is limited for a plurality of defined prediction directions. It is an object to determine an optimal prediction direction only by calculating a prediction evaluation value of a prediction direction limited to a prediction direction used for prior determination for the purpose and a prediction direction extracted by the determination. It is another object of the present invention to efficiently reduce the amount of calculation by limiting the prediction direction in which the prediction evaluation value is calculated.

  An image encoding apparatus according to the present invention is an image encoding apparatus that performs predictive encoding, and includes a prediction direction setting unit that sets at least two prediction directions for an encoding target block, and the prediction direction setting unit. A first prediction evaluation value calculation means for calculating a prediction evaluation value in at least two set prediction directions, and a prediction direction superior based on the prediction evaluation value for each prediction direction calculated by the first prediction evaluation value calculation means. A prediction direction extraction unit that extracts one or a plurality of prediction directions related to the dominant direction, and a second prediction evaluation that calculates a prediction evaluation value of each prediction direction extracted by the prediction direction extraction unit A value calculation unit, a prediction direction determination unit that determines an optimal prediction direction based on the prediction evaluation value of each prediction direction calculated by the first and second prediction evaluation value calculation units, and the prediction direction determination unit determines Shi Is obtained by a prediction mode conversion means for converting an optimum prediction direction to prediction mode corresponding to the prediction direction.

  According to the present invention, when performing predictive encoding, without calculating all prediction evaluation values for a plurality of defined prediction directions, the number of prediction direction targets for calculating prediction evaluation values is limited. An optimal prediction direction can be determined only by calculating the prediction direction used for the prior determination and the prediction evaluation value of the prediction direction limited to the prediction direction extracted by the determination. Moreover, there is an effect that the amount of calculation can be efficiently reduced by limiting the prediction direction in which the prediction evaluation value is calculated.

  A prediction mode selection process based on the prediction evaluation value in the prediction direction of intra prediction encoding in the image encoding device according to the present invention will be described. Here, as described in the prior art, as the prediction evaluation value, the difference absolute value or the difference square value between all the encoding target pixels in the encoding target block and the prediction pixel (prediction value) is added. Use the total value obtained. And the prediction direction which takes the minimum prediction evaluation value in the prediction direction which calculated the prediction evaluation value shall be an optimal prediction direction. Hereinafter, embodiments of the present invention will be described in detail.

Embodiment 1 FIG.
In Embodiment 1 of the present invention, a prediction evaluation value is calculated by extracting a prediction direction in the vicinity of a superior prediction direction estimated based on a ratio between prediction evaluation values of at least two prediction directions set at first, An image coding apparatus that selects a prediction mode that provides an optimal prediction direction from the calculated prediction evaluation values of the prediction direction will be described.

  FIG. 1 is a block diagram showing an image coding apparatus according to Embodiment 1 of the present invention. FIG. 2 is a flowchart showing an example of a prediction mode selection processing procedure based on the prediction evaluation value in the prediction direction of intra prediction encoding according to the image encoding device in Embodiment 1 of the present invention. The block configuration of FIG. 1 will be described in association with the processing steps of FIG.

  In FIG. 1, the prediction direction setting unit 101A sets at least two prediction directions for calculating a prediction evaluation value (step ST201). The first prediction evaluation value calculation means 102 receives the image data and the prediction direction set by the prediction direction setting means 101A as inputs, calculates and outputs a prediction evaluation value in the prediction direction (step ST202). The prediction direction extraction unit 103 receives the prediction evaluation value output from the first prediction evaluation value calculation unit 102 and extracts a prediction direction for calculating another prediction evaluation value based on the prediction evaluation value (step ST203). (Step ST204 and Step ST208). The second prediction evaluation value calculation means 104 receives the image data and the prediction direction extracted by the prediction direction extraction means 103, and calculates and outputs a prediction evaluation value in the prediction direction (step ST205 and step ST209). The prediction direction determination means 105 receives the prediction evaluation values output from the first prediction evaluation value calculation means 102 and the second prediction evaluation value calculation means 104 as input, and determines an optimal prediction direction (step ST206). The prediction mode conversion means 106 receives the prediction direction output from the prediction direction determination means 105, and outputs a prediction mode corresponding to the prediction direction (step ST207).

  In addition, the 1st prediction evaluation value calculation means 102 and the 2nd prediction evaluation value calculation means 104 shall match | combine and output a prediction evaluation value and a prediction direction. In addition, the first prediction evaluation value calculation means 102 and the second prediction evaluation value calculation means 104 are independent means for explanation, but one prediction evaluation value is output for the input prediction direction. The prediction evaluation value calculation unit 151 may be configured.

  Further, although the prediction direction determination unit 105 and the prediction mode conversion unit 106 are independent units for explanation, prediction is performed by adding a prediction evaluation value using a direction other than the prediction direction such as DC prediction as in Non-Patent Document 1. In the case of determining the mode, etc., the correlation between the prediction direction shown in FIG. 11 and the prediction mode shown in the non-patent document is clearly specified, and the prediction mode is determined only by the prediction direction. Moreover, as long as it can determine without rearranging, it is good also as the prediction mode determination part 152A comprised integrally.

  The prediction direction extracted by the prediction direction extraction unit 103 is not limited to one and may be a plurality.

  A prediction mode selection processing procedure in the first embodiment of the present invention shown in the flowchart of FIG. 2 will be described. The prediction evaluation value in the prediction direction indicated by the number n shown in FIG. The determination threshold Th0 is set to a value of 1 or more, but will be described as being 1 here.

  First, in step ST201, two prediction directions 1 and 5, that is, prediction evaluation values in the horizontal direction and the vertical direction are set to be calculated. In step ST202, prediction evaluation values E1 and E5 in the two set prediction directions are calculated.

  Next, in the determination 1 of step ST203, a dominant prediction direction is estimated according to the ratio E5 / E1 of the prediction evaluation values E1 and E5. Here, a simple comparison is made between the predicted evaluation values E1 and E5 from Th0 = 1. In the determination 1, if Th0> 1, it is not a simple comparison between E1 and E5, so it is determined based on the relationship of the inequality in step ST203.

  In the branch of “Yes” in which E1 is determined to be less than E5 by comparison with the determination threshold Th0 = 1 of determination 1, in step ST204, the prediction direction 1 becomes the dominant prediction direction, and the prediction direction near the prediction direction 1 Extract 7, 0, 2, 3. In step ST205, the prediction evaluation values E7, E0, E2, and E3 in the extracted prediction direction are calculated. In step ST206, the prediction direction that takes the minimum value of the calculated prediction evaluation values E7, E0, E1, E2, and E3 is determined as the optimal prediction direction. In step ST207, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Further, in the “No” branch in which E1 is determined to be equal to or greater than E5 by comparison with the determination threshold Th0 = 1 in determination 1, in step ST208, the prediction direction 5 becomes the dominant prediction direction, and the vicinity of the prediction direction 5 Prediction directions 3, 4, 6, and 7 are extracted. In step ST209, the prediction evaluation values E3, E4, E6, and E7 in the extracted prediction direction are calculated. In step ST206, the prediction direction that takes the minimum value of the calculated prediction evaluation values E3, E4, E5, E6, and E7 is determined as the optimal prediction direction. In step ST207, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Here, in the “No” branch of determination 1 in step ST203, if the determination threshold Th0 = 1, E1 = E5 may be satisfied, but E1 <E5 is not satisfied. Since the branch side of “Yes” in determination 1 takes the prediction direction 1, the branch direction of “No” gives priority to the prediction direction 5 and does not have six prediction evaluation values E1, but prediction evaluation values E3, E4, E5, E6, and E7 are used. Note that if the threshold value Th0> 1, then in the range of Th0 ≧ E5 / E1> 1, E1 <E5, so the optimal value is calculated from the five predicted evaluation values E1, E3, E4, E6, and E7. The prediction direction may be determined. However, if the determination threshold Th0 is a value close to 1, the difference between E1 and E5 is small, so that it may be determined to take E5 exceptionally.

  Here, as a result of extracting two prediction directions in both the clockwise direction and the counterclockwise direction in FIG. 11 with the dominant prediction direction estimated in the determination 1 as a reference, the prediction directions 3 and 7 overlap. For example, the prediction directions 0, 2, and 3 are extracted in step ST204, and the prediction directions 4, 6, and 7 are extracted in step ST208 so as not to overlap. I do not care.

  In the first embodiment of the present invention, the number of prediction directions to be set first and the number and directions of prediction directions extracted thereafter may be arbitrarily changed. Here, although the prediction direction set first is the horizontal direction and the vertical direction, a prediction direction that is orthogonal or a prediction direction that is not orthogonal may be set by a combination other than the prediction directions 1 and 5.

  As described above, according to the first embodiment of the present invention, instead of calculating prediction evaluation values for all prediction directions, an optimal prediction direction can be obtained by calculating prediction evaluation values for fewer prediction directions. And the amount of calculation can be reduced.

Embodiment 2. FIG.
In Embodiment 2 of the present invention, all the prediction directions are divided into a plurality of at least two prediction directions that are initially set, and the prediction direction that is superior based on the ratio of the prediction evaluation values of the at least two prediction directions is An image coding apparatus that extracts a prediction direction of an estimated segment, calculates a prediction evaluation value, and selects a prediction mode that provides an optimal prediction direction from the prediction evaluation values of the calculated prediction direction will be described.

  A block diagram showing an image coding apparatus according to Embodiment 2 of the present invention will be described with the same configuration as the block diagram of FIG. 1 shown in Embodiment 1 of the present invention. FIG. 3 is a flowchart showing an example of a prediction mode selection processing procedure based on the prediction evaluation value in the prediction direction of intra prediction encoding according to the image encoding device in Embodiment 2 of the present invention. The block configuration of FIG. 1 will be described in association with the processing steps of FIG.

  In FIG. 1, the prediction direction setting means 101A sets at least two prediction directions for calculating a prediction evaluation value (step ST301). The first prediction evaluation value calculation means 102 receives the image data and the prediction direction set by the prediction direction setting means 101A as inputs, calculates and outputs a prediction evaluation value in the prediction direction (step ST302). The prediction direction extraction unit 103 receives the prediction evaluation value output from the first prediction evaluation value calculation unit 102 as an input, and performs prediction based on the prediction evaluation value (step ST303 and step ST308) to calculate another prediction evaluation value. The direction is extracted (step ST304, step ST309 and step ST311). The second prediction evaluation value calculation means 104 receives the image data and the prediction direction extracted by the prediction direction extraction means 103, and calculates and outputs a prediction evaluation value in the prediction direction (step ST305, step ST310 and step ST312). ). The prediction direction determination means 105 receives the prediction evaluation values output from the first prediction evaluation value calculation means 102 and the second prediction evaluation value calculation means 104 as input, and determines an optimal prediction direction (step ST306). The prediction mode conversion means 106 receives the prediction direction output from the prediction direction determination means 105, and outputs a prediction mode corresponding to the prediction direction (step ST307).

  A prediction mode selection processing procedure in the second embodiment of the present invention shown in the flowchart of FIG. 3 will be described. The prediction evaluation value in the prediction direction indicated by the number n shown in FIG. The determination thresholds Th0 and Th1 will be described as values satisfying Th0> Th1 ≧ 1.

  First, similarly to the first embodiment of the present invention, in step ST301, two prediction directions 1 and 5, that is, prediction evaluation values in the horizontal direction and the vertical direction are set to be calculated. In step ST302, prediction evaluation values E1 and E5 in the two set prediction directions are calculated.

  Next, in determination 1 in step ST303 and determination 2 in step ST308, according to the ratio E5 / E1 of the predicted evaluation values E1 and E5, is there a sufficient difference between E1 and E5 at the determination threshold Th0, or at the determination threshold Th1? Determine if there is a relative difference greater than a certain ratio. From the two determinations, E5 / E1> Th0, Th0 ≧ E5 / E1> Th1, and E5 / E1 ≦ Th1 are classified. In E5 / E1> Th0, the relationship is such that Th0> 1 to E1 is sufficiently smaller than E5. In the case of Th0 ≧ E5 / E1> Th1, the relationship is such that Th1 ≧ 1 and E1 is slightly smaller than E5. In E5 / E1 ≦ Th1, the relationship is such that Th1 ≧ 1 and E1 is almost equal to E5 (when Th1> 1) or E1 is greater than or equal to E5.

  In the branch of “Yes” where E5 / E1> Th0 and E1 is determined to be sufficiently smaller than E5 by comparison with the determination threshold Th0> 1 of determination 1, in step ST304, the prediction direction 1 is set as the dominant prediction direction, Prediction directions 0 and 2 near the prediction direction 1 are extracted. In step ST305, the prediction evaluation values E0 and E2 in the extracted prediction direction are calculated. In step ST306, the prediction direction that takes the minimum value of the calculated prediction evaluation values E0, E1, and E2 is determined as the optimal prediction direction. In step ST307, the determined optimal prediction direction is converted into a corresponding prediction mode.

  In the branch of “Yes” in which E1 is determined to be slightly smaller than E5 by Th0 ≧ E5 / E1> Th1 by comparison with the determination threshold Th1 ≧ 1 in determination 2 after “No” is determined in determination 1. In step ST309, the prediction direction 1 is set as the dominant prediction direction, and prediction directions 3 and 4 sandwiched between the prediction directions 1 and 5 are extracted. In step ST310, the prediction evaluation values E3 and E4 in the extracted prediction direction are calculated. In step ST306, the prediction direction that takes the minimum value of the calculated prediction evaluation values E1, E3, and E4 is determined as the optimal prediction direction. In step ST307, the determined optimal prediction direction is converted into a corresponding prediction mode.

  After judging “No” in judgment 1, by comparison with judgment threshold Th1 ≧ 1 in judgment 2, E1 is almost equal to E5 when E5 / E1 ≦ Th1 (when Th1> 1) or E1 is equal to or greater than E5. In the “No” branch to be determined, in step ST311, the prediction direction 5 is set as the dominant prediction direction, and the prediction directions 6 and 7 are extracted. In step ST312, the prediction evaluation values E6 and E7 in the extracted prediction direction are calculated. In step ST306, the prediction direction that takes the minimum value of the calculated prediction evaluation values E5, E6, and E7 is determined as the optimal prediction direction. In step ST307, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Here, in the “No” branch of determination 2 in step ST308, if the determination threshold Th1 = 1, E1 = E5 may be obtained, but E1 <E5 is not satisfied. Since both the branch of “Yes” of decision 1 in step ST303 and the branch of “Yes” in decision 2 take the prediction direction 1, the branch of “No” in decision 2 gives priority to the prediction direction 5 and adds the prediction evaluation value E1. The predicted evaluation values E5, E6, and E7 are set to three instead of four. If the determination threshold value Th1> 1, then in the range of Th1 ≧ E5 / E1> 1, E1 <E5, so the optimum prediction direction is determined from the three calculated prediction evaluation values E1, E6, and E7. It doesn't matter. However, if the determination threshold Th1 is a value close to 1, the difference between E1 and E5 is small, so that it may be determined to take E5 exceptionally.

  In the second embodiment of the present invention, the concept of extracting a prediction direction according to a ratio E1 / E5 of prediction evaluation values, calculating a prediction evaluation value of the extracted prediction direction, and determining an optimal prediction direction is as follows. This is basically the same as Embodiment 1 of the present invention.

  In the second embodiment of the present invention, the relationship between the determination thresholds Th0 and Th1 has been described as Th0> Th1 ≧ 1, but by setting Th0> 1> Th1> 0, the ratio of E5 / E1 in FIG. When the prediction direction 1 that is sufficiently larger than 1 is dominant in the “Yes” branch of the determination 1, after the “No” branch of the determination 1, and in the vicinity of 1 in the “Yes” branch of the determination 2, The branch of “No” may be divided into three when the prediction direction 5 sufficiently smaller than 1 is dominant. Further, it can be determined symmetrically with respect to E1 and E5 as three sections where the ratio of E5 / E1 is Th1 = 1 / Th0.

  In the second embodiment of the present invention, similarly to the first embodiment of the present invention, the prediction direction to be initially set and the number and direction of the prediction directions extracted thereafter may be arbitrarily changed. Here, although the prediction direction set first is the horizontal direction and the vertical direction, a prediction direction that is orthogonal or a prediction direction that is not orthogonal may be set by a combination other than the prediction directions 1 and 5.

  As described above, according to the second embodiment of the present invention, instead of calculating prediction evaluation values for all prediction directions, an optimal prediction direction can be obtained by calculating prediction evaluation values for fewer prediction directions. And the amount of calculation can be reduced.

Embodiment 3 FIG.
In the third embodiment of the present invention, another prediction direction is designated based on the ratio of the prediction evaluation values of at least two prediction directions set at first, and the prediction evaluation value of the designated prediction direction is added. An image coding apparatus that extracts a prediction direction estimated from a combination ratio, calculates a prediction evaluation value, and selects a prediction mode that provides an optimal prediction direction from the calculated prediction evaluation values of the prediction direction will be described. In Embodiment 3 of the present invention, the prediction direction is additionally specified once or plural times until the prediction direction in which the prediction evaluation value is to be calculated is finally extracted, and the prediction direction specified at one time is also 1 One or more.

  FIG. 4 is a block diagram showing an image coding apparatus according to Embodiment 3 of the present invention. FIG. 5 is a flowchart showing an example of a prediction mode selection processing procedure based on the prediction evaluation value in the prediction direction of intra prediction encoding according to the image encoding device in Embodiment 3 of the present invention. The block configuration of FIG. 4 will be described in association with the processing steps of FIG.

  In FIG. 4, the prediction direction setting means 201A sets at least two prediction directions for calculating a prediction evaluation value (step ST401). First prediction evaluation value calculation means 202 receives the prediction direction set by image data and prediction direction setting means 201A as input, and calculates and outputs a prediction evaluation value in that prediction direction (step ST402). The prediction direction designation unit 203 receives the prediction evaluation value output from the first prediction evaluation value calculation unit 202 as an input, and specifies a prediction direction for calculating another prediction evaluation value based on the prediction evaluation value (step ST403). (Step ST404 and Step ST413). The second prediction evaluation value calculating means 204 receives the image data and the prediction direction extracted by the prediction direction designating means 203, calculates and outputs a prediction evaluation value in the prediction direction (step ST405 and step ST414). The prediction direction extraction unit 205 receives the prediction evaluation values output from the first prediction evaluation value calculation unit 202 and the second prediction evaluation value calculation unit 204 as input, and based on the prediction evaluation values (step ST406 and step ST415). Then, a prediction direction for calculating another prediction evaluation value is extracted (step ST407, step ST411, step ST416, and step ST418). The third prediction evaluation value calculation means 206 receives the image data and the prediction direction extracted by the prediction direction extraction means 205, and calculates and outputs a prediction evaluation value in the prediction direction (step ST408, step ST412, step ST417). And step ST419). The prediction direction determination means 207 receives the prediction evaluation values output from the first prediction evaluation value calculation means 202, the second prediction evaluation value calculation means 204, and the third prediction evaluation value calculation means 206 as an input, and the optimal prediction direction. Is determined (step ST409). The prediction mode conversion means 208 receives the prediction direction output from the prediction direction determination means 207 and outputs a prediction mode corresponding to the prediction direction (step ST410).

  Note that the first prediction evaluation value calculation unit 202, the second prediction evaluation value calculation unit 204, and the third prediction evaluation value calculation unit 206 output the prediction evaluation value and the prediction direction in association with each other. In addition, the first predicted evaluation value calculating means 202, the second predicted evaluation value calculating means 204, and the third predicted evaluation value calculating means 206 are independent means for explanation, but the desired prediction direction is You may comprise as one prediction evaluation value calculating part 251 which outputs a prediction evaluation value.

  Moreover, although the prediction direction determination means 207 and the prediction mode conversion means 208 are independent means for explanation, prediction is performed by adding a prediction evaluation value using a direction other than the prediction direction such as DC prediction as in Non-Patent Document 1. In the case of determining the mode, etc., the correlation between the prediction direction shown in FIG. 11 and the prediction mode shown in the non-patent document is clearly specified, and the prediction mode is determined only by the prediction direction. Moreover, as long as it can determine without rearranging, it is good also as the prediction mode determination part 252A comprised integrally.

  Further, the prediction direction designated by the prediction direction designation unit 203 and the prediction directions extracted by the prediction direction extraction unit 205 are not limited to one and may be plural.

  A prediction mode selection processing procedure in the third embodiment of the present invention shown in the flowchart of FIG. 5 will be described. The prediction evaluation value in the prediction direction indicated by the number n shown in FIG. The determination threshold values Th0, Th1, and Th2 are set to values of 1 or more, but here, description will be made assuming that all are 1.

  First, the prediction direction initially set is different from that of the first embodiment of the present invention, and in step ST401, the prediction evaluation values in the two prediction directions 3 and 4 (see FIG. 11) are set to be calculated. In step ST402, prediction evaluation values E3 and E4 in the two set prediction directions are calculated.

  Next, in determination 1 of step ST403, a preferential prediction direction is estimated according to the ratio E4 / E3 of the prediction evaluation values E3 and E4, and further, with respect to the prediction direction 3, the prediction direction 1 and the prediction direction 4 By additionally specifying the prediction direction 6, a more preferential prediction direction is estimated according to the ratio of the prediction evaluation values, and other prediction directions are extracted.

  In the branch of “Yes” where E4 / E3> Th0 and E3 <E4 is determined by comparison with the determination threshold Th0 = 1 of determination 1, in step ST404, the prediction direction 3 is set as the dominant prediction direction. For example, the prediction direction Specify one additional prediction direction 1 from 3. In step ST405, the prediction evaluation value E1 in the designated prediction direction 1 is calculated.

  Next, in decision 2 of step ST406, a more dominant prediction direction is estimated according to the prediction direction 3 and the ratio E4 / E3 of the prediction evaluation values E3 and E1 of the additionally specified prediction direction 1. In the branch of “Yes” where E3 / E1> Th1 and E1 <E3 are determined by comparison with the determination threshold Th1 = 1 in determination 2, in step ST407, for example, the vicinity of the dominant prediction direction 1 at this time In step ST408, a prediction evaluation value E0 in the prediction direction 0 is calculated. In step ST409, the prediction direction having a smaller value from the prediction evaluation value E1 of the prediction direction 1 having the smallest prediction evaluation value and the prediction evaluation value E0 of the prediction direction 0 extracted last is determined as the optimum prediction direction. Determine as. In step ST410, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Further, in the “No” branch in which E3 / E1 ≦ Th1 and E1 ≧ E3 is determined by comparison with the determination threshold Th1 = 1 in determination 2, in step ST411, for example, with respect to the dominant prediction direction 3 at this time A prediction direction 2 in the vicinity thereof is extracted, and a prediction evaluation value E2 in the prediction direction 2 is calculated in step ST412. In step ST409, the prediction direction having a smaller value from the prediction evaluation value E3 in the prediction direction 3 having the smallest prediction evaluation value and the prediction evaluation value E2 in the prediction direction 2 extracted last is determined as the optimum prediction direction. Determine as. In step ST410, the determined optimal prediction direction is converted into a corresponding prediction mode. Note that although E3 = E1 may be obtained when the determination threshold Th1 = 1, the prediction direction 1 is used in the “Yes” branch of the determination 2 in which it is determined that E1 <E3, and therefore the prediction direction 3 has priority. did.

  In the branch of “No” where E4 / E3 ≦ Th0 and E3 ≧ E4 is determined by comparison with the determination threshold Th0 = 1 of determination 1, in step ST413, the prediction direction 4 is set as the dominant prediction direction. 4. Specify one additional prediction direction 6 from 4. In step ST414, the prediction evaluation value E6 in the designated prediction direction 6 is calculated. Note that E3 = E4 may be obtained when the determination threshold Th0 = 1, but since the prediction direction 3 is used in the “Yes” branch of determination 1 in which it is determined that E3 <E4, the prediction direction 4 is given priority. did.

  Next, in decision 3 of step ST415, a more preferential prediction direction is estimated according to the prediction evaluation values E4 and E6 ratio E4 / E6 of the prediction direction 6 additionally specified. In the branch of “Yes” where E4 / E6> Th2 and E6 <E4 is determined by comparison with the determination threshold Th2 = 1 in determination 3, in step ST416, for example, in the vicinity of the dominant prediction direction 6 at this time In step ST417, a prediction evaluation value E7 in the prediction direction 7 is calculated. In step ST409, the prediction direction having a smaller value from the prediction evaluation value E6 in the prediction direction 6 having the smallest prediction evaluation value and the prediction evaluation value E7 in the prediction direction 7 extracted last is determined as the optimum prediction direction. Determine as. In step ST410, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Further, in the branch of “No” where E4 / E6 ≦ Th2 and E6 ≧ E4 is determined by comparison with the determination threshold Th2 = 1 in determination 3, in step ST418, for example, for the prevailing prediction direction 4, A prediction direction 5 in the vicinity thereof is extracted, and a prediction evaluation value E5 in the prediction direction 5 is calculated in step ST419. In step ST409, the prediction direction that has a smaller value from the prediction evaluation value E4 of the prediction direction 4 having the smallest prediction evaluation value and the prediction evaluation value E5 of the prediction direction 5 extracted last is determined as the optimum prediction direction. Determine as. In step ST410, the determined optimal prediction direction is converted into a corresponding prediction mode. Note that E4 = E6 may be obtained when the determination threshold Th2 = 1, but since the prediction direction 6 is used in the “Yes” branch of the determination 3 in which E6 <E4 is determined, the prediction direction 4 is given priority. did.

  Here, when the determination threshold values Th0 to Th2 of the determinations 1 to 3 are larger than 1, for example, as the determination 1 of FIG. Preferential treatment or exceptional treatment is performed in the prediction direction of a certain prediction evaluation value.

  It should be noted that after the decision 1 branch, there may be no other prediction direction in which the prediction evaluation value is to be calculated. Further, after the calculation of the prediction evaluation value in the prediction direction, the process is re-branched by the determination 2 or the determination 3. However, after this determination, the prediction evaluation value in another prediction direction may be calculated again and the determination may be repeated. Furthermore, as the prediction evaluation value in the prediction direction used in the subsequent determination, a part of the prediction evaluation value in the prediction direction used in the upper determination may not be used in common. Further, different judgment thresholds may be applied to use a prediction evaluation value in a common prediction direction.

  In the third embodiment of the present invention, as in the first embodiment of the present invention, the prediction direction initially set and the number and direction of the prediction directions extracted thereafter may be arbitrarily changed. Here, the prediction direction initially set is not the horizontal direction or the vertical direction, but is not limited to the combination of the horizontal direction and the vertical direction, and an orthogonal prediction direction may be set.

  As described above, according to the third embodiment of the present invention, instead of calculating prediction evaluation values for all prediction directions, an optimal prediction direction can be obtained only by calculating prediction evaluation values for fewer prediction directions. And the amount of calculation can be reduced.

Embodiment 4 FIG.
In Embodiment 4 of the present invention, the prediction evaluation value is calculated by extracting another prediction direction estimated based on the ratio of the prediction evaluation values of at least two prediction directions set at first, and the calculated prediction direction An image encoding apparatus that selects a prediction mode that provides an optimal prediction direction from among the prediction evaluation values will be described. In the first embodiment of the present invention, it is estimated that the prediction direction includes the vicinity of the dominant prediction direction in at least two prediction directions, and in the second embodiment of the present invention, the prediction direction that is dominant in at least two prediction directions is included. In the fourth embodiment of the present invention, the prediction direction is not limited to the vicinity based on the ratio of the prediction evaluation values of at least two prediction directions. It shall be extracted by narrowing down to a smaller number.

  A block diagram showing an image coding apparatus according to Embodiment 4 of the present invention will be described with the same configuration as the block diagram of FIG. 1 shown in Embodiment 1 of the present invention. FIG. 6 is a flowchart showing an example of a prediction mode selection processing procedure based on the prediction evaluation value in the prediction direction of intra prediction encoding according to the image encoding device in Embodiment 4 of the present invention. The block configuration of FIG. 1 will be described in association with the processing steps of FIG.

  In FIG. 1, the prediction direction setting means 101A sets at least two prediction directions for calculating a prediction evaluation value (step ST501). First prediction evaluation value calculation means 102 receives the image data and the prediction direction set by prediction direction setting means 101A as inputs, and calculates and outputs a prediction evaluation value in the prediction direction (step ST502). The prediction direction extraction unit 103 receives the prediction evaluation value output from the first prediction evaluation value calculation unit 102, and based on the prediction evaluation value (step ST503, step ST504, step ST509, step ST514 and step ST517), Prediction directions for calculating other prediction evaluation values are extracted (step ST505, step ST510, step ST512, step ST515, step ST518 and step ST520). The second prediction evaluation value calculation means 104 receives the image data and the prediction direction extracted by the prediction direction extraction means 103, and calculates and outputs a prediction evaluation value in the prediction direction (step ST506, step ST511, step ST513). Step ST516, Step ST519 and Step ST521). The prediction direction determination means 105 receives the prediction evaluation values output from the first prediction evaluation value calculation means 102 and the second prediction evaluation value calculation means 104 as input, and determines an optimal prediction direction (step ST507). The prediction mode conversion means 106 receives the prediction direction output from the prediction direction determination means 105, and outputs a prediction mode corresponding to the prediction direction (step ST508).

  A prediction mode selection processing procedure according to the fourth embodiment of the present invention shown in the flowchart of FIG. 6 will be described. The prediction evaluation value in the prediction direction indicated by the number n shown in FIG. The determination threshold values Th0, Th1, Th2, Th3, and Th4 are values that satisfy Th0> Th1> Th2> = 1> Th3> Th4> 0.

  First, similarly to Embodiment 1 of the present invention, in step ST501, two prediction directions 1 and 5, that is, prediction evaluation values in the horizontal direction and the vertical direction are set to be calculated. In step ST502, prediction evaluation values E1 and E5 in the two set prediction directions are calculated.

  Next, in determinations 1 to 5, the ratio E5 / E1 of E1 and E5 is divided into six based on the five determination thresholds Th0 to Th4, and the prediction direction for calculating the prediction evaluation value is determined. That is, the conditions from the case where E1 is much smaller than E5 to the case where E1 is much larger than E5 are stepwise classified according to the determination threshold, and E0, E2, E3, E4, E6, E7 according to each condition. The prediction evaluation value in any one of the prediction directions is calculated. Here, the determination thresholds Th0 to Th4 may be compared so that E1 is very small from E5 in order, but the number of determinations until E1 is very large than E5 increases. 1 shows a flowchart in which the magnitude relation with respect to the determination threshold Th2 is first determined at 1, and the maximum number of determinations is reduced until the prediction direction is extracted by the comparison procedure from the determination threshold Th0 to Th1 and from Th3 to Th4 after branching. Yes.

  Processing after the branch of “Yes” in which E5 / E1> Th2 is determined by comparison with the determination threshold Th2 ≧ 1 of determination 1 will be described. In the branch of “Yes” determined as E5 / E1> Th0 in the determination 2 of step ST504, in step ST505, for example, the prediction direction 0 is extracted with respect to the dominant prediction direction 1, and in step ST506, the prediction direction 0 is extracted. A prediction evaluation value E0 in the prediction direction 0 is calculated. In step ST507, a prediction direction having a smaller value is determined as the optimal prediction direction from the prediction evaluation value E1 in the prediction direction 1 having a small prediction evaluation value and the extracted prediction evaluation value E0 in the prediction direction 0 in the determination so far. . In step ST508, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Further, the processing after the branch of “No” determined as Th0 ≧ E5 / E1> Th2 in the determination 2 of step ST504 will be described. In the branch of “Yes” where Th0 ≧ E5 / E1> Th1 is determined in the determination 3 of step ST509, in step ST510, for example, the prediction direction 2 is extracted with respect to the prediction direction 1 that is dominant at this time, and in step ST511 Then, the prediction evaluation value E2 in the prediction direction 2 is calculated. In step ST507, a prediction direction having a smaller value is determined as an optimal prediction direction from the prediction evaluation value E1 in the prediction direction 1 having a small prediction evaluation value and the extracted prediction evaluation value E2 in the prediction direction 2 in the determination so far. . In step ST508, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Further, in the “No” branch in which it is determined that Th1 ≧ E5 / E1> Th2 in the determination 3 in step ST509, in step ST512, for example, the prediction direction 3 is extracted with respect to the dominant prediction direction 1, and the step In ST513, a prediction evaluation value E3 in the prediction direction 3 is calculated. In step ST507, a prediction direction having a smaller value is determined as an optimal prediction direction from the prediction evaluation value E1 in the prediction direction 1 having a small prediction evaluation value and the extracted prediction evaluation value E3 in the prediction direction 3 in the determination so far. . In step ST508, the determined optimal prediction direction is converted into a corresponding prediction mode.

  On the other hand, the processing after the branch of “No” determined as E5 / E1 ≦ Th2 by comparison with the determination threshold Th2 ≧ 1 of determination 1 will be described. In the branch of “Yes” in which it is determined that Th2 ≧ E5 / E1> Th3 in the determination 4 in step ST514, in step ST515, for example, the prediction direction 4 is extracted with respect to the dominant prediction direction 5, and in step ST516. The prediction evaluation value E4 in the prediction direction 4 is calculated. In step ST507, a prediction direction having a smaller value is determined as an optimal prediction direction from the prediction evaluation value E5 in the prediction direction 5 having a small prediction evaluation value and the extracted prediction evaluation value E4 in the prediction direction 4 in the determination so far. . In step ST508, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Here, in the “Yes” branch of the determination 4 in step ST514, if the determination threshold Th2 = 1, E1 = E5 may be obtained, but E1 <E5 is not satisfied. Since the “Yes” branch side of determination 1 in step ST503 takes the prediction direction 1, the prediction direction 5 is given priority on the “No” branch side of determination 1, and the prediction evaluation value E1 is not added to the three predictions. Two evaluation values E4 and E5 are provided. Note that if the determination threshold Th2> 1, then in the range of Th2 ≧ E5 / E1> 1, E1 <E5, so the optimum prediction direction is determined from the two calculated prediction evaluation values E1 and E4. It doesn't matter. However, if the determination threshold Th2 is a value close to 1, the difference between E1 and E5 is small, so that it may be determined to take E5 exceptionally.

  In addition, the process after the branch of “No” determined as E5 / E1 ≦ Th3 in the determination 4 of step ST514 will be described. In the branch of “Yes” determined as Th3 ≧ E5 / E1> Th4 in the determination 5 of step ST517, in step ST518, for example, the prediction direction 6 is extracted with respect to the dominant prediction direction 5, and in step ST519 Then, the prediction evaluation value E6 in the prediction direction 6 is calculated. In step ST507, a prediction direction having a smaller value is determined as the optimum prediction direction from the prediction evaluation value E5 in the prediction direction 5 having a small prediction evaluation value and the extracted prediction evaluation value E6 in the prediction direction 6 in the determination so far. . In step ST508, the determined optimal prediction direction is converted into a corresponding prediction mode.

  Further, in the “No” branch in which it is determined that Th4 ≧ E5 / E1 in the determination 5 of step ST517, in step ST520, for example, the prediction direction 7 is extracted with respect to the dominant prediction direction 5, and in step ST521 The prediction evaluation value E7 in the prediction direction 7 is calculated. In step ST507, a prediction direction having a smaller value is determined as the optimum prediction direction from the prediction evaluation value E5 in the prediction direction 5 having a small prediction evaluation value and the extracted prediction evaluation value E7 in the prediction direction 7 in the determination so far. . In step ST508, the determined optimal prediction direction is converted into a corresponding prediction mode.

  In the fourth embodiment of the present invention, the relationship between the determination threshold values Th0, Th1, Th2, Th3, and Th4 has been described as Th0> Th1> Th2> = 1> Th3> Th4> 0, but the ratio of E5 / E1 is It is also possible to determine symmetrically with respect to E1 and E5 as six sections with Th2 = 1, Th3 = 1 / Th1, and Th4 = 1 / Th0.

  In the fourth embodiment of the present invention, as in the first embodiment of the present invention, the prediction direction initially set and the number and direction of the prediction directions extracted thereafter may be arbitrarily changed. Here, although the prediction direction set first is the horizontal direction and the vertical direction, a prediction direction that is orthogonal or a prediction direction that is not orthogonal may be set by a combination other than the prediction directions 1 and 5.

  As described above, according to the fourth embodiment of the present invention, instead of calculating prediction evaluation values in all prediction directions, it is optimal only to calculate prediction evaluation values in three prediction directions selected as candidates. A correct prediction direction can be obtained, and the amount of calculation can be reduced.

Embodiment 5 FIG.
In the fifth embodiment of the present invention, at least two prediction directions to be initially set are not fixed in the horizontal direction and the vertical direction with respect to the first embodiment, the second embodiment, and the fourth embodiment of the present invention. An image coding apparatus capable of specifying rotational symmetry while maintaining the crossing positional relationship will be described. Since the prediction direction applied to predictively encode the adjacent encoding target block is expected to have a strong correlation, the prediction direction applied in the encoded block is stored, and the stored prediction is stored. A prediction direction to be applied to an encoding target block to be encoded based on the direction is designated.

  FIG. 7 is a block diagram showing an image coding apparatus according to Embodiment 5 of the present invention. The prediction direction storage means 107 is added to the image coding apparatus shown in FIG. 1 according to Embodiment 1 of the present invention.

  In FIG. 7, the prediction direction storage unit 107 stores the optimum prediction direction determined by the prediction direction determination unit 105 for each encoding target block. The prediction direction setting unit 101B estimates the reference prediction direction with reference to the prediction direction applied to the encoded block stored in the prediction direction storage unit 107, and sets at least two prediction directions for the encoding target block. . Based on the encoding direction set in this way, the prediction direction determining means 105 determines the optimal prediction direction through the operations described in the first embodiment, the second embodiment, and the third embodiment of the present invention. To do. The determined optimal prediction direction is stored in the prediction direction storage unit 107, and the prediction mode determination unit 106 converts the prediction direction into the prediction mode.

  Since the first prediction evaluation value calculation means 102 and the second prediction evaluation value calculation means 104 output the prediction evaluation value and the prediction direction in association with each other as described in the first embodiment of the present invention, The prediction direction extraction unit 103 and the prediction direction determination unit 105 that receive the output can perform processing while maintaining a rotationally symmetrical prediction direction having a relative positional relationship with respect to at least two prediction directions set by the prediction direction setting unit 101B. become. Note that the prediction pixel is calculated from an arithmetic expression defined corresponding to the prediction direction in FIG.

  For example, the prediction direction which is a reference estimated by the prediction direction setting unit 101B from the prediction direction referred from the prediction direction storage unit 107 may be used as a reference by referring to the prediction direction applied to the block adjacent to the encoding target block. Alternatively, in the case of a moving image, the prediction direction applied to the block at the same position one frame before may be used. For the same image frame, the prediction direction is applied to the immediately adjacent block and the immediately preceding (left) block, or is applied to the immediately adjacent block and the immediately preceding (left) block. The prediction direction may be an intermediate prediction direction.

  The operation when the prediction direction serving as a reference is estimated from the prediction direction referred from the prediction direction storage unit 107 by the prediction direction setting unit 101B will be described. Here, as an example, the procedure for determining the prediction direction and the prediction mode in Embodiment 1 of the present invention is considered to be rotationally symmetric. First, prediction directions 0 to 7 in FIG. 11 are numbered in eight directions clockwise. The horizontal direction and the vertical direction correspond to the prediction directions 1 and 5, and the numerical values of the orthogonal prediction directions are separated by 4.

  In the flowchart of FIG. 2, the estimated reference direction n is assumed to be one of the initially set prediction directions, and any of the prediction directions 1 and 5 may be used. explain. Hereinafter, the prediction direction m processed in each step is relatively treated as a prediction direction (n + d) reflecting a numerical difference d = m−1 from the prediction direction 1 replaced with the prediction direction n. Here, when (n + d) exceeds 7, it is sufficient to subtract 8 and normalize from 0 to 7. When (n + d) becomes a negative number, the total number 8 in the prediction direction may be added and normalized from 0 to 7.

  When the prediction direction numbering in FIG. 11 is not performed clockwise or counterclockwise and is not performed regularly, for example, the prediction direction numbering in FIG. What is necessary is just to prepare and apply the correspondence table | surface converted into the numbering of the prediction direction in 1 or prediction mode.

  In step ST201, prediction directions n and (n + 4) are set, and in step ST202, prediction evaluation values En and E (n + 4) are calculated. In the branch of “Yes” where E (n + 4) / En> Th0 is satisfied in the determination 1 of step ST203, the prediction directions (n−2), (n−1), (n + 1), and (n + 2) are extracted in step ST204. In step ST205, the predicted evaluation values E (n-2), E (n-1), E (n + 1), and E (n + 2) are calculated. In step ST206, the optimal prediction direction is determined based on the prediction evaluation values of the prediction directions (n-2), (n-1), n, (n + 1), and (n + 2), and in step ST207, the corresponding prediction is performed. Convert to mode. Here, although not shown in the flowchart of FIG. 2, the optimal prediction direction determined in step ST206 is stored in the storage means.

  Further, in the branch of “No” in which E (n + 4) / En ≦ Th0 is satisfied in the determination 1 of step ST203, the prediction directions (n + 2), (n + 3), (n + 5), and (n + 6) are extracted in step ST208. In step ST209, the predicted evaluation values E (n + 2), E (n + 3), E (n + 5), and E (n + 6) are calculated. In step ST206, the optimum prediction direction is determined based on the prediction evaluation values of prediction directions (n + 2), (n + 3), (n + 4), (n + 5), and (n + 6), and in step ST207, the corresponding prediction mode is set. Convert.

  In the first embodiment of the present invention, four prediction directions are extracted. However, in the fifth embodiment of the present invention, in step ST204, the prediction directions (n−1), (n + 1), and step ST208 are extracted. , (N + 3) and (n + 5) are limited to only two directions, the prediction directions are extracted in a rotationally symmetrical manner, so that there is a possibility that all eight prediction directions are extracted depending on the reference prediction direction.

  Here, the estimated reference direction n as an estimated reference is added to one of the prediction directions initially set. For example, the prediction direction set in step ST201 in the flowchart of FIG. May be used as orthogonal prediction directions (n−2) and (n + 2).

  Similarly, each flowchart in FIG. 3 according to the second embodiment of the present invention and FIG. 6 according to the fourth embodiment of the present invention similarly performs processing corresponding to the prediction direction that is rotationally symmetric with respect to the prediction direction n. Can be explained.

  In the fifth embodiment of the present invention, it is possible to specify a prediction direction as a reference with reference to a prediction direction applied in a block having a high correlation with the encoding target block, and processing based on the specified prediction direction of the reference The prediction direction can be changed to rotational symmetry. Thereby, a more preferable prediction direction is extracted, and a more optimal prediction direction can be determined based on each prediction evaluation value in the prediction direction.

  As described above, according to the fifth embodiment of the present invention, instead of calculating prediction evaluation values for all prediction directions, an optimum prediction direction can be obtained by calculating prediction evaluation values for fewer prediction directions. And the amount of calculation can be reduced.

Embodiment 6 FIG.
In the sixth embodiment of the present invention, compared to the third embodiment of the present invention, the prediction direction to be initially set is not fixed, and the image coding that can be specified rotationally symmetric while maintaining the crossing positional relationship is maintained. The apparatus will be described. Since the prediction direction applied to predictively encode the adjacent encoding target block is expected to have a strong correlation, the prediction direction applied to the encoded block is the same as in the fifth embodiment of the present invention. , And the prediction direction to be applied to the encoding target block to be encoded based on the stored prediction direction is designated.

  FIG. 8 is a block diagram showing an image coding apparatus according to Embodiment 6 of the present invention. A prediction direction storage unit 209 is added to the image coding apparatus shown in FIG. 4 according to Embodiment 3 of the present invention.

  In FIG. 8, the prediction direction storage unit 209 stores the optimal prediction direction determined by the prediction direction determination unit 207 for each encoding target block. The prediction direction setting unit 201B estimates a reference prediction direction with reference to the prediction direction applied to the encoded block stored in the prediction direction storage unit 209, and sets at least two prediction directions for the encoding target block. . Based on the encoding direction set in this way, the prediction direction determining means 207 determines the optimal prediction direction through the operations described in the third embodiment of the present invention. The prediction direction storage unit 209 stores the determined optimum prediction direction, and the prediction mode determination unit 208 converts the prediction direction into a prediction mode.

  As described in the third embodiment of the present invention, the first predicted evaluation value calculating unit 202, the second predicted evaluation value calculating unit 204, and the third predicted evaluation value calculating unit 206 set the predicted evaluation value and the prediction direction. Since these are output in association with each other, the prediction direction specifying unit 203, the prediction direction extracting unit 205, and the prediction direction determining unit 207 that receive the output are relative to at least two prediction directions set by the prediction direction setting unit 201B. Therefore, it is possible to perform processing while maintaining the predicted directions having a proper positional relationship in rotational symmetry. Note that the prediction pixel is calculated from an arithmetic expression defined corresponding to the prediction direction in FIG.

  The prediction direction serving as a reference estimated by the prediction direction setting unit 201B from the prediction direction referenced from the prediction direction storage unit 209 may be determined in the same manner as in the fifth embodiment of the present invention.

  The operation when the prediction direction serving as the reference is estimated from the prediction direction referenced from the prediction direction storage unit 209 by the prediction direction setting unit 201B is the same as that described in the fifth embodiment of the present invention, as shown in the flowchart of FIG. This can be explained by performing processing corresponding to the prediction direction that is rotationally symmetric with respect to the prediction direction n in the figure.

  The designated reference prediction direction n is, for example, one of the prediction directions for calculating the prediction evaluation value in step ST401 and step ST402 in the flowchart of FIG. 5, and the other adjacent prediction direction is the prediction direction (n -1) or prediction direction (n + 1).

  In Embodiment 6 of the present invention, it is possible to specify a reference prediction direction by referring to a prediction direction applied in a block having a high correlation with the encoding target block in the group determination, and the prediction of the specified reference The prediction direction for calculating the prediction evaluation value based on the direction can be changed to be rotationally symmetric, and the prediction direction candidate for calculating the prediction evaluation value of the group to be discriminated can be changed accordingly. Thereby, more preferable prediction direction candidates can be extracted, and a more optimal prediction direction can be determined based on each prediction evaluation value in the prediction direction.

  As described above, according to the sixth embodiment of the present invention, instead of calculating prediction evaluation values for all prediction directions, an optimum prediction direction can be obtained only by calculating prediction evaluation values for fewer prediction directions. And the amount of calculation can be reduced.

Embodiment 7 FIG.
In the seventh embodiment of the present invention, the prediction direction in the vicinity of the superior prediction direction estimated based on the ratio of the prediction evaluation values of the at least two prediction directions set at the beginning set in the first embodiment is extracted. In addition to the function of calculating the prediction evaluation value and selecting the prediction mode that provides the optimal prediction direction from the prediction evaluation values of the calculated prediction direction, the prediction evaluation values of at least two prediction directions that are initially set In some cases, an image coding apparatus having a function of selecting a prediction mode that provides an optimum prediction direction from prediction evaluation values of at least two prediction directions set at the beginning will be described.

  FIG. 9 is a block diagram showing an image coding apparatus according to the seventh embodiment of the present invention. The block diagram (FIG. 1) showing the image coding apparatus according to the first embodiment of the present invention is provided with the operation stop determination means 108. This is an added configuration. FIG. 10 is a flowchart showing an example of a prediction mode selection processing procedure based on the prediction evaluation value in the prediction direction of the intra prediction encoding according to the image encoding device in Embodiment 7 of the present invention. FIG. 2 is a flowchart showing an example of a prediction mode selection processing procedure based on a prediction evaluation value in the prediction direction of intra prediction coding according to the image coding apparatus in Embodiment 1 (step ST210 and step ST211). Yes. The block configuration of FIG. 9 will be described in association with the processing steps of FIG.

  In FIG. 9, the prediction direction setting means 101C sets at least two prediction directions for calculating a prediction evaluation value (step ST201). The first prediction evaluation value calculation means 102 receives the image data and the prediction direction set by the prediction direction setting means 101C as inputs, calculates and outputs a prediction evaluation value in the prediction direction (step ST202). The calculation stop determination unit 108 sets the calculation stop determination threshold value from the prediction direction setting unit 101C, and determines whether at least one of the evaluation values calculated by the first prediction evaluation value calculation unit is equal to or less than the set threshold value. Determination is made (step ST210 and step ST211). The prediction direction extraction unit 103 receives the prediction evaluation value output from the first prediction evaluation value calculation unit 102 and extracts a prediction direction for calculating another prediction evaluation value based on the prediction evaluation value (step ST203). (Step ST204 and Step ST208). The second prediction evaluation value calculation means 104 receives the image data and the prediction direction extracted by the prediction direction extraction means 103, and calculates and outputs a prediction evaluation value in the prediction direction (step ST205 and step ST209). The prediction direction determination unit 105B receives the prediction evaluation value output from the first prediction evaluation value calculation unit 102 and the second prediction evaluation value calculation unit 104 and the determination result of the calculation stop determination unit 108 as an input, and determines an optimal prediction direction. Determine (step ST206). The prediction mode conversion means 106 receives the prediction direction output by the prediction direction determination means 105B and outputs a prediction mode corresponding to the prediction direction (step ST207).

  The calculation stop determination threshold value set in the calculation stop determination unit 108 is a threshold value corresponding to the prediction evaluation value in the prediction direction set in the first prediction evaluation value calculation unit 102 by the prediction direction setting unit 101C.

  Further, as shown in FIG. 9, the result determined by the calculation stop determination means 108 based on the calculation stop determination threshold is notified to the prediction direction determination means 105B, and the evaluation calculated by the first prediction evaluation value calculation means. If at least one of the values is less than or equal to the set threshold value, the prediction direction determination unit 105B ignores the prediction evaluation value from the second prediction evaluation value calculation unit 104 and outputs the first evaluation value. An optimal prediction direction may be determined based only on the prediction evaluation value from the prediction evaluation value calculation unit 102. Further, as indicated by a broken line, the determination result by the calculation stop determination unit 108 is directly notified to the second prediction evaluation value calculation unit 104 or indirectly notified via the prediction direction extraction unit 103, thereby If at least one of the evaluation values calculated by one prediction evaluation value calculation means is equal to or less than the set threshold value, it is notified that the calculation of the prediction evaluation value in the second prediction evaluation value calculation means 104 is unnecessary. It is also possible to determine the optimum prediction direction by inputting only the prediction evaluation value from the first prediction evaluation value calculation means 102 to the prediction direction determination means 105B.

  Further, although the prediction direction determination unit 105B and the prediction mode conversion unit 106 are independent units for explanation, prediction is performed by adding a prediction evaluation value using a direction other than the prediction direction such as DC prediction as in Non-Patent Document 1. In the case of determining the mode, etc., the correlation between the prediction direction shown in FIG. 11 and the prediction mode shown in the non-patent document is clearly specified, and the prediction mode is determined only by the prediction direction. Moreover, as long as it can determine without rearranging, it is good also as the prediction mode determination part 152C comprised integrally.

  A prediction mode selection processing procedure according to the seventh embodiment of the present invention shown in the flowchart of FIG. 10 will be described. The prediction evaluation value in the prediction direction indicated by the number n shown in FIG. The determination threshold Th0 is set to a value of 1 or more, but will be described as being 1 here.

  First, in step ST201, two prediction directions 1 and 5, that is, prediction evaluation values in the horizontal direction and the vertical direction are set to be calculated. In step ST202, prediction evaluation values E1 and E5 in the two set prediction directions are calculated.

  Next, in the determination 1 of step ST210, the predicted evaluation value E1 is compared with the calculation stop determination threshold value α set from the outside. When the predicted evaluation value E1 is less than or equal to α, the branch of “Yes” causes a decision 2 in step ST211, a decision 3 in step ST203, a prediction direction extraction in step ST204, an evaluation value calculation in step ST205, and a prediction direction extraction in step ST208. The evaluation value calculation in step ST209 is not performed, and in step ST206, the prediction direction that takes the minimum value of the calculated prediction evaluation values E1 and E5 is determined as the optimal prediction direction. In step ST207, the determined optimal prediction direction is converted into a corresponding prediction mode.

  When the predicted evaluation value E1 is larger than α, the “No” branch is used to compare the predicted evaluation value E5 with the operation stop determination threshold value β set from the outside in the determination 2 of step ST211. When the predicted evaluation value E5 is less than or equal to β, the branch of “Yes” causes the decision 3 in step ST203, extraction of the prediction direction in step ST204, calculation of the evaluation value in step ST205, extraction of the prediction direction in step ST208, and calculation of the evaluation value in step ST209. In step ST206, the prediction direction that takes the minimum value of the calculated prediction evaluation values E1 and E5 is determined as the optimal prediction direction. In step ST207, the determined optimal prediction direction is converted into a corresponding prediction mode.

  When the predicted evaluation value E5 is larger than β, a superior prediction direction is estimated according to the ratio E5 / E1 of the predicted evaluation values E1 and E5 in the determination 3 of step ST203 by branching “No”. Hereinafter, including the determination 3 in step ST203, after extracting from the prediction direction in step ST204, calculating the evaluation value in step ST205, extracting the prediction direction in step ST208, calculating the evaluation value in step ST209, and after joining again The processing of each step of the prediction direction determination of step ST206 and the prediction mode conversion of step ST207 is the same as step ST209 from step ST203 (determination 1) described in the flowchart of FIG. 2 in the first embodiment of the present invention. Each corresponding process shall be performed.

  Note that the determinations of step ST210 and step ST211 may be performed in the order of step ST211 and step ST210, or in parallel, and it may be determined whether or not to proceed to the determination of step ST203.

  As described above, according to the seventh embodiment of the present invention, when at least one of the predicted evaluation values calculated by the first predicted evaluation value calculating means is equal to or less than a preset calculation stop determination threshold value, Since the optimum prediction direction is selected from the prediction evaluation values in the first prediction evaluation value calculation means, the prediction direction determination means receives the notification from the calculation stop determination means, so that the second prediction evaluation value calculation is performed. It is not necessary to wait until the prediction evaluation values output by the means are complete, and the calculation processing time can be shortened as compared with the image coding apparatus according to the first embodiment.

  Further, according to the seventh embodiment of the present invention, similarly, since the optimum prediction direction is selected from the prediction evaluation values in the first prediction evaluation value calculation means, the second prediction evaluation value calculation means. Since it is possible to prevent the calculation of the prediction evaluation value from being executed by receiving the notification directly from the calculation stop determination unit or through the prediction direction extraction unit, the image encoding apparatus according to the first embodiment is more suitable. , Low power consumption can be achieved.

  The calculation stop determination threshold set in the calculation stop determination means may be a fixed value or a variable value. By appropriately changing the calculation stop determination threshold, even if the same predicted evaluation value is calculated in the first predicted evaluation value calculating means, the necessity of calculating the predicted evaluation value in the second predicted evaluation value calculating means is detailed. Since the determination can be made, it is possible to arbitrarily control the calculation processing time and the power consumption.

  Further, in the seventh embodiment of the present invention, depending on the prediction evaluation values of at least two prediction directions initially set in the image coding apparatus described in the first embodiment, The configuration and operation have been described in which the function of selecting the prediction mode that provides the optimal prediction direction from the prediction evaluation values has been described, but the present invention can also be applied to the second to sixth embodiments described so far. It is.

  When applied to the third and sixth embodiments, the determination result of the calculation stop determination unit is directly notified to the second prediction evaluation value calculation unit and the third prediction evaluation value calculation unit, The calculation of the predicted evaluation value may not be executed. Alternatively, the determination result of the calculation stop determination unit is indirectly notified to the second prediction evaluation value calculation unit and the third prediction evaluation value calculation unit via the prediction direction designation unit and the prediction direction extraction unit, and the second You may control calculation of a prediction evaluation value with respect to a prediction evaluation value calculating means and a 3rd prediction evaluation value calculating means.

  In the embodiments of the present invention described so far, the prediction coding applied in the H.264 intra coding, which is an example of the image coding device for moving images, has been described. It is obvious that the present invention can be applied to an image coding apparatus for still images as well, since it is performed by using only an image frame to be converted.

  In the embodiment of the present invention, more prediction directions are defined in a block of 8 pixels × 8 lines than in a block of 4 pixels × 4 lines, and an optimal prediction is made based on a determination threshold value among these prediction evaluation values. The direction can be selected. By increasing the size of the block, the total number of prediction directions that can be defined increases, and the number of constituent pixels also increases. Therefore, the amount of calculation can be significantly reduced by efficiently limiting the prediction directions for calculating the prediction evaluation value.

  In the flowcharts of FIGS. 2, 3, 5, 6, and 10 shown in the first to fourth embodiments and the seventh embodiment of the present invention, prediction evaluation values in two prediction directions are used. In the determination, the ratio is used for the explanation. However, when the ratio of the prediction evaluation values in the two prediction directions is compared with the determination threshold, the section defined with the determination threshold as a boundary, etc. The number is included on the upper bound side, but may be included on the lower bound side.

  Further, in the flowchart of FIG. 10 shown in the seventh embodiment of the present invention, when the prediction evaluation values in the two prediction directions are compared with the calculation stop determination threshold, the interval defined as the determination threshold is used as a section. On the other hand, the equal sign is included on the lower bound side, but it may be included on the upper bound side.

  Further, in the flowcharts of FIGS. 2, 3, 5, 6, and 10 shown in the first to fourth embodiments and the seventh embodiment of the present invention, the prediction direction finally extracted The number may be increased or decreased from the number of candidates shown in each flowchart diagram as an example. Further, the extracted prediction direction is not limited to that shown in each flowchart.

  Further, in the flowcharts of FIGS. 2, 3, 5, 6, and 10 shown in the first to fourth embodiments and the seventh embodiment of the present invention, an example of the extracted prediction direction is shown. Although shown, these may be defined arbitrarily. The extracted prediction direction is assumed to be a prediction direction related to the dominant prediction direction obtained by the determination before extraction. For example, a prediction direction superior to the prediction direction shown in FIG. 11 is obtained, a prediction direction in the vicinity of an appropriate range or a prediction direction with an appropriate interval is defined, and a prediction corresponding to the definition for each extraction You may pick up the direction. Further, based on the definition, a list of prediction directions extracted for the dominant prediction directions may be created in advance and picked up. Note that if the extracted prediction evaluation value in the prediction direction has been calculated at the time of determination before extraction, there is no need to calculate again, and the prediction evaluation value at the time of determination may be used.

  In the block configuration diagrams of FIGS. 1, 7, and 9 shown in the first embodiment, the second embodiment, the fourth embodiment, the fifth embodiment, and the seventh embodiment of the present invention, the first prediction The prediction evaluation value calculated by the evaluation value calculation means is directly input to the prediction direction determination means, but is input to the prediction direction determination means via the second prediction evaluation value calculation means via the prediction direction extraction means. It doesn't matter. At this time, in the seventh embodiment, even when the second prediction evaluation value calculation means does not calculate the prediction evaluation value, the prediction evaluation value calculated by the first prediction evaluation value calculation means is used as the second prediction evaluation value. It is assumed that the calculation direction is input to the prediction direction determination unit.

  Further, in the block configuration diagrams of FIGS. 4 and 8 shown in the third and sixth embodiments of the present invention, the predicted evaluation value calculated by the first predicted evaluation value calculating means is directly input to the prediction direction determining means. Although it is input, it may be input to the prediction direction determining means via the second prediction evaluation value calculating means via the prediction direction specifying means, or may be further input to the third prediction evaluation via the prediction direction extracting means. You may make it input into a prediction direction determination means through a value calculation means. Similarly, the prediction evaluation value calculated by the second prediction evaluation value calculation means is directly input to the prediction direction determination means, but is predicted via the third prediction evaluation value calculation means via the prediction direction extraction means. You may make it input into a direction determination means.

  As described above, according to the embodiment of the present invention, the prediction direction used for determination and the prediction direction extracted by the determination are calculated without calculating all prediction evaluation values for the plurality of defined prediction directions. Since the optimal prediction direction can be determined only by calculating the prediction evaluation values of some prediction directions, the amount of calculation can be efficiently reduced as the total number of prediction directions increases.

  Further, according to the embodiment of the present invention, when the image encoding device is configured by hardware, effects such as reduction in circuit scale and power consumption can be obtained. Further, when the image encoding device is configured by software, an effect such as shortening the processing time can be obtained. Further, when processing is performed by a general-purpose computer or the like, even if it is configured and operated by an arithmetic processing unit having a low processing capability, it can be executed more comfortably.

It is a block block diagram which concerns on the image coding apparatus in Embodiment 1, Embodiment 2 and Embodiment 4 of this invention. It is a flowchart figure which shows an example of the prediction mode selection processing procedure based on the prediction evaluation value of the prediction direction of the intra prediction encoding which concerns on the image coding apparatus in Embodiment 1 of this invention. It is a flowchart figure which shows an example of the prediction mode selection processing procedure based on the prediction evaluation value of the prediction direction of the intra prediction encoding which concerns on the image coding apparatus in Embodiment 2 of this invention. It is a block block diagram which concerns on the image coding apparatus in Embodiment 3 of this invention. It is a flowchart figure which shows an example of the prediction mode selection process sequence based on the prediction evaluation value of the prediction direction of the intra prediction encoding which concerns on the image coding apparatus in Embodiment 3 of this invention. It is a flowchart figure which shows an example of the prediction mode selection processing procedure based on the prediction evaluation value of the prediction direction of the intra prediction encoding which concerns on the image coding apparatus in Embodiment 4 of this invention. It is a block block diagram which concerns on the image coding apparatus in Embodiment 5 of this invention. It is a block block diagram which concerns on the image coding apparatus in Embodiment 6 of this invention. It is a block block diagram which concerns on the image coding apparatus in Embodiment 7 of this invention. It is a flowchart figure which shows an example of the prediction mode selection processing procedure based on the prediction evaluation value of the prediction direction of the intra prediction encoding which concerns on the image coding apparatus in Embodiment 7 of this invention. It is explanatory drawing which shows the prediction mode of the intra prediction in a prior art. It is explanatory drawing which shows the macroblock and block of intra prediction in a prior art. It is explanatory drawing which shows the encoding object pixel and prediction value of the prediction direction in a prior art. It is a flowchart figure which shows the prediction mode selection processing procedure based on the prediction evaluation value of the prediction direction of the intra prediction encoding which concerns on the image coding apparatus in a prior art.

Explanation of symbols

101A, 101B Prediction direction setting means 102 First prediction evaluation value calculation means 103 Prediction direction extraction means 104 Second prediction evaluation value calculation means 105, 105B Prediction direction determination means 106 Prediction mode conversion means 107 Prediction mode storage means 108 Computation stop Determination unit 151 Prediction evaluation value calculation unit 152A, 152B, 152C Prediction mode determination unit 201A, 201B Prediction direction setting unit 202 First prediction evaluation value calculation unit 203 Prediction direction extraction unit 204 Second prediction evaluation value calculation unit 205 Prediction direction Designating means 206 Third predictive evaluation value calculating means 207 Predicting direction determining means 208 Predictive mode converting means 209 Predictive mode storage means 251 Predictive evaluation value calculating parts 252A, 252B Predictive mode determining part

Claims (11)

An image encoding device that performs predictive encoding,
Prediction direction setting means for setting at least two prediction directions for the encoding target block;
First prediction evaluation value calculating means for calculating prediction evaluation values in at least two prediction directions set by the prediction direction setting means;
Prediction direction extraction for determining a superior prediction direction based on the prediction evaluation value of each prediction direction calculated by the first prediction evaluation value calculation means, and extracting one or a plurality of prediction directions related to the dominant direction. Means,
Second prediction evaluation value calculating means for calculating a prediction evaluation value in each prediction direction extracted by the prediction direction extracting means;
A prediction direction determination unit that determines an optimal prediction direction based on the prediction evaluation value of each prediction direction calculated by the second prediction evaluation value calculation unit;
An image coding apparatus comprising: a prediction mode conversion unit that converts an optimal prediction direction determined by the prediction direction determination unit into a prediction mode corresponding to the prediction direction. An image encoding device that performs predictive encoding,
Prediction direction setting means for setting at least two prediction directions for the encoding target block;
First prediction evaluation value calculating means for calculating prediction evaluation values in at least two prediction directions set by the prediction direction setting means;
A prediction direction designating unit for designating one or more prediction directions in which a prediction evaluation value is to be calculated based on the prediction evaluation value of each prediction direction calculated by the first prediction evaluation value calculating unit;
Second prediction evaluation value calculating means for calculating a prediction evaluation value in each prediction direction designated by the prediction direction designating means;
Prediction direction extraction for determining a superior prediction direction based on the prediction evaluation value of each prediction direction calculated by the second prediction evaluation value calculation means, and extracting one or a plurality of prediction directions related to the dominant direction Means,
Third prediction evaluation value calculating means for calculating a prediction evaluation value in each prediction direction extracted by the prediction direction extracting means;
A prediction direction determination unit that determines an optimal prediction direction based on the prediction evaluation value of each prediction direction calculated by the third prediction evaluation value calculation unit;
An image coding apparatus comprising: a prediction mode conversion unit that converts an optimal prediction direction determined by the prediction direction determination unit into a prediction mode corresponding to the prediction direction. The image encoding apparatus according to claim 1, wherein the prediction directions set by the prediction direction setting unit are prediction directions orthogonal to each other. A prediction direction storage unit that stores the optimum prediction direction determined by the prediction direction determination unit for each encoding target block;
The image according to claim 1, wherein the prediction direction setting unit sets the at least two prediction directions based on one or more prediction directions stored by the prediction direction storage unit. Encoding device. The prediction direction setting means refers to the prediction direction applied to the encoded block located directly above and / or to the left of the encoding target block from the prediction direction storage means, and estimates based on the prediction direction 5. The image encoding apparatus according to claim 4, wherein the at least two prediction directions are set. The prediction direction setting means refers to the prediction direction applied by the encoded block at the same position of the image one frame before the encoding target block from the prediction direction storage means, and estimates based on the prediction direction 5. The image encoding apparatus according to claim 4, wherein the at least two prediction directions are set. An image encoding method for performing predictive encoding,
A prediction direction setting step of setting at least two prediction directions for the encoding target block;
A first prediction evaluation value calculation step for calculating prediction evaluation values in at least two prediction directions set by the prediction direction setting step;
Prediction direction extraction that determines a superior prediction direction based on the prediction evaluation value of each prediction direction calculated by the first prediction evaluation value calculation step, and extracts one or more prediction directions related to the superior direction. Steps,
A second prediction evaluation value calculation step for calculating a prediction evaluation value in each prediction direction extracted by the prediction direction extraction step;
A prediction direction determination step of determining an optimal prediction direction based on the prediction evaluation value of each prediction direction calculated by the second prediction evaluation value calculation step;
An image encoding method comprising: a prediction mode conversion step of converting the optimal prediction direction determined in the prediction direction determination step into a prediction mode corresponding to the prediction direction. An image encoding method for performing predictive encoding,
A prediction direction setting step of setting at least two prediction directions for the encoding target block;
A first prediction evaluation value calculation step for calculating prediction evaluation values in at least two prediction directions set by the prediction direction setting step;
A prediction direction designating step for designating one or a plurality of prediction directions in which a prediction evaluation value is to be calculated based on the prediction evaluation value of each prediction direction calculated by the first prediction evaluation value calculating step;
A second prediction evaluation value calculating step for calculating a prediction evaluation value in each prediction direction specified by the prediction direction specifying step;
Prediction direction extraction that determines a superior prediction direction based on the prediction evaluation value of each prediction direction calculated in the second prediction evaluation value calculation step, and extracts one or more prediction directions related to the superior direction. Steps,
A third prediction evaluation value calculation step for calculating a prediction evaluation value in each prediction direction extracted by the prediction direction extraction step;
A prediction direction determination step of determining an optimal prediction direction based on the prediction evaluation value of each prediction direction calculated by the third prediction evaluation value calculation step;
An image encoding method comprising: a prediction mode conversion step of converting the optimal prediction direction determined in the prediction direction determination step into a prediction mode corresponding to the prediction direction. A prediction direction storage step for storing the optimum prediction direction determined by the prediction direction determination step for each encoding target block;
The image according to claim 7 or 8, wherein the prediction direction setting step sets the at least two prediction directions based on one or more prediction directions stored in the prediction direction storage step. Encoding method. Other predictions except the prediction direction corresponding to the prediction evaluation value calculated by the first prediction evaluation value calculating means by comparing the prediction evaluation value calculated by the first prediction evaluation value calculating means with the calculation stop determination threshold value Computation stop determination means for determining whether or not to execute the prediction evaluation value calculation for the direction,
The prediction direction determination means predicts each prediction direction calculated by the first prediction evaluation value calculation means when the calculation stop determination means determines that the prediction evaluation value is not calculated for the other prediction directions. The image encoding apparatus according to claim 1, wherein an optimal prediction direction is determined based only on the evaluation value. Other predictions except the prediction direction corresponding to the prediction evaluation value calculated by the first prediction evaluation value calculation step by comparing the prediction evaluation value calculated by the first prediction evaluation value calculation step and the calculation stop determination threshold value A calculation stop determination step for determining whether or not to execute a calculation of a predicted evaluation value for a direction,
In the prediction direction determination step, when the calculation stop determination step determines that the prediction evaluation value is not calculated for the other prediction directions, the prediction of each prediction direction calculated by the first prediction evaluation value calculation step is performed. 9. The image encoding method according to claim 7, wherein an optimal prediction direction is determined based only on the evaluation value.

Images