摘要:
An image processing apparatus for decoding an image coded in coding processes including (i) prediction on at least one prediction unit (PU) and (ii) frequency transform on transform units (TUs) included in a region including the at least one PU includes a processing unit configured to perform image processing for a predictive image of the at least one PU, depending on a transform order predetermined for the TUs.
摘要:
An image processing apparatus for decoding an image coded in coding processes including (i) prediction on at least one prediction unit (PU) and (ii) frequency transform on transform units (TUs) included in a region including the at least one PU includes a processing unit configured to perform image processing for a predictive image of the at least one PU, depending on a transform order predetermined for the TUs.
摘要:
An image decoding device is provided that decodes, without an increase in the circuit size, a coded video sequence including a large-size block. The image decoding apparatus includes: a block division unit which divides a current block that is to be decoded and is included in a picture of the coded video sequence, into sub-blocks; an image obtainment unit which obtains, for each of the sub-blocks, image data corresponding to the sub-block from a recording medium; a prediction unit which generates, for each of the sub-blocks, a predicted image of the sub-block, based on the image data obtained by the image obtainment unit; and a reconstruction unit which reconstructs each of the sub-blocks, by adding the predicted image generated for the sub-block to the sub-block.
摘要:
A motion compensation apparatus performs motion compensation on a current block having a non-rectangular shape, using a reference image stored in a frame memory and includes: a reference block memory for storing part of the reference image; a frame-memory-transfer control unit which identifies, in the reference image, a pixel block having a rectangular shape and including a reference block having a non-rectangular shape and used for motion compensation, and transfers pixel data of the identified pixel block from the frame memory to the reference block memory; and a motion compensation processing unit which generates a prediction block of the current block, using the reference block included in the pixel block stored in the reference block memory.
摘要:
When compression encoding processing of an image is performed in units of macroblocks using pipeline structure, application of the skip mode or the like according to MPEG4AVC to compression encode an encoding target block requires motion vectors and the like of adjacent blocks of the encoding target block. However, depending on the structure of the pipeline stages, the motion vectors and the like may not be determined. In such cases, the skip mode cannot be applied to compression encode the encoding target block. The present invention aims to solve this problem and (i) calculates all motion information candidates, of the encoding target block, corresponding to all motion information selectable by a previous block of the encoding target block, and (ii) selects, as the motion information of the encoding target block in the skip mode, the motion information corresponding to the motion information determined for the previous block.
摘要:
With use of a simplified program or calculating device for motion compensation, a video decoding device decodes video data compressed by motion detection operations on macroblock units, as in the MPEG-4AVC standard. The video decoding device splits compressed data blocks of the prescribed size, 16×16 pixels for instance, to generate sub-blocks, which are smaller than the blocks and on which the video decoding device is able to execute motion compensation operations. The video decoding device duplicates a motion vector assigned to a given block to generate as many motion vectors as there are sub-blocks in the given block, and executes motion compensation on each sub-block using the corresponding duplicate motion vector. Data resulting from the motion compensation operation on each sub-block is combined to obtain a target block corresponding to the given block.
摘要:
An image decoding device is provided that decodes, without an increase in the circuit size, a coded video sequence including a large-size block. The image decoding apparatus includes: a block division unit which divides a current block that is to be decoded and is included in a picture of the coded video sequence, into sub-blocks; an image obtainment unit which obtains, for each of the sub-blocks, image data corresponding to the sub-block from a recording medium; a prediction unit which generates, for each of the sub-blocks, a predicted image of the sub-block, based on the image data obtained by the image obtainment unit; and a reconstruction unit which reconstructs each of the sub-blocks, by adding the predicted image generated for the sub-block to the sub-block.
摘要:
A moving image decoding apparatus which enables reduction in the memory bandwidth and the memory access latency for the motion compensation filter coefficients for use in inter-picture prediction involving motion compensation using variable coefficients includes: a decoding unit (101) which decodes, from a coded stream, a plurality of motion compensation filter coefficients; a memory (109) for holding the motion compensation filter coefficients included in the coded stream; a filter coefficient storage unit (103) for holding at least one of the motion compensation filter coefficients which is required for the motion compensation; a motion compensation unit (107) which performs motion compensation using the required motion compensation filter coefficient held in the filter coefficient storage unit; and a filter coefficient transfer control unit (102) which writes, in the memory, the motion compensation filter coefficients decoded by the decoding unit, and transfers the required motion compensation filter coefficient from the memory to the filter coefficient storage unit, only when the required coefficient is not yet stored therein.
摘要:
With use of a simplified program or calculating device for motion compensation, a video decoding device decodes video data compressed by motion detection operations on macroblock units, as in the MPEG-4AVC standard. The video decoding device splits compressed data blocks of the prescribed size, 16×16 pixels for instance, to generate sub-blocks, which are smaller than the blocks and on which the video decoding device is able to execute motion compensation operations. The video decoding device duplicates a motion vector assigned to a given block to generate as many motion vectors as there are sub-blocks in the given block, and executes motion compensation on each sub-block using the corresponding duplicate motion vector. Data resulting from the motion compensation operation on each sub-block is combined to obtain a target block corresponding to the given block.
摘要:
A decoding apparatus (100) according to the present invention includes: a decoding unit (101) which decodes identification information identifying an orthogonal transform basis for inverse orthogonal transform; an orthogonal transform basis accumulation unit (110) accumulating orthogonal transform bases for inverse orthogonal transform; an orthogonal transform basis storage unit (103) storing an orthogonal transform basis for inverse transform, from among the stored orthogonal transform bases; an inverse orthogonal transform unit (112) which performs inverse orthogonal transform using the identified orthogonal transform basis; and an orthogonal transform basis transfer control unit (102) which transfers the identified orthogonal transfer basis from the orthogonal transform basis accumulation unit (110) to the orthogonal transfer basis storage unit (103) only when the identified orthogonal transform basis is not yet stored therein. With this structure, it is possible to reduce the memory bandwidth for the memory storing the orthogonal transform basis and the memory access latency.