WO2014203351A1

WO2014203351A1 - Image processing device and image processing method

Info

Publication number: WO2014203351A1
Application number: PCT/JP2013/066856
Authority: WO
Inventors: 小林　晃
Original assignee: 株式会社東芝
Priority date: 2013-06-19
Filing date: 2013-06-19
Publication date: 2014-12-24

Abstract

According to an embodiment, in the event of an increase in bandwidth during pixel data processing, the number of processing bits of the pixel data is dynamically changed in order to prevent a data processing failure. A motion-compensation image data generation unit decodes an I-picture by means of intra prediction utilizing reference image data within a frame, a P-picture by means of forward prediction utilizing reference image data within a past frame, and a B-picture by means of bi-directional prediction utilizing both past and future reference image data. An image data storage control unit (140) stores the decoded image data into a frame memory (45). A reference image data acquisition unit (120) reads from the frame memory (45) reference image data for the aforementioned decoding and inputs the reference image data to the motion-compensation image data generation unit. Here, when the B-picture is to be decoded, the reference image data acquisition unit inputs a smaller number of bits of pixel data to the motion-compensation image data generation unit in comparison to when the other pictures are to be decoded.

Description

Image processing apparatus and image processing method

Embodiments described herein relate generally to an image processing apparatus, an image processing method, and a computer-readable storage medium for an image processing program.

In recent image processing technologies, image quality has been improved, and apparatuses corresponding to so-called 2K1K image data display having 1920 pixels in the horizontal direction and 1080 pixels in the vertical direction have become mainstream.

On the other hand, recently, image data called 4K2K with 3840 pixels in the horizontal direction and 2160 pixels in the vertical direction, and image data called 8K4K with 7680 pixels in the horizontal direction and 4320 pixels in the vertical direction, Ultra high-definition image data has also been realized. Also, UHDTV (Ultra High Definition TV) capable of displaying ultra-high definition image data has been developed.

Generally, in order to transmit image data, it is necessary to reduce the transmission capacity in order to efficiently use the bandwidth of the transmission path. For this purpose, the transmission side encodes image data and transmits the encoded image data. On the receiving side, the received encoded image data is decoded. In a decoding apparatus, when a compressed bit stream is decoded, an I picture (image data encoded by intra frame prediction), a B picture (image data encoded by bidirectional prediction), and a P picture (forward prediction) The decoding process using the image data encoded by (1) is executed.

The number of digital bits of the image data pixel referred to as 2K1K described above is normally 8 bits.

JP 2007-306152 A JP 2010-052837 A JP 2007-533212 A

However, in the case of pixels of high-definition image data such as 4K2K or 8K4K described above, the number of digital bits may be 10 bits or more. For example, there is a standard called HEVC (High Efficiency Video Coding) as 4K2K encoding, but a Main 10 profile is also defined. In this profile, 10 bits are adopted as the number of digital bits of the pixel.

As a result, a high-speed clock is used in a high-resolution image data encoding device and decoding device. When this type of decoding device is designed, there is a tendency that the access frequency to the memory increases, the power consumption thereby increases, and the configuration scale of the LSI (Large Scale Integrated) circuit increases.

In the decoding process, when the number of bits to be processed per unit time increases according to the prediction method and prediction vector, the bandwidth being processed changes greatly. When processing 10-bit pixels, the bandwidth is increased by about 25% compared to the bandwidth when processing 8-bit pixels in the decoding apparatus.

However, manufacturing a decoding device that realizes data processing that can always satisfy the maximum bandwidth has an increase in LSI circuit scale and power consumption, and when the number of pixel bits increases in the future, Bandwidth cannot be guaranteed. In a small mobile device, an increase in LSI circuit scale and power consumption are undesirable.

Therefore, in this embodiment, when the pixel data is processed, if the bandwidth increases, the number of processing bits of the pixel data can be dynamically changed, and the failure of data processing can be prevented. An object is to provide a computer-readable storage medium for an image processing apparatus, an image processing method, and an image processing program.

According to the embodiment, the motion compensated image data generation unit decodes the first encoded picture by intra prediction using the reference image data in the frame, and references the second encoded picture to a past frame. Decoding is performed by forward prediction using image data, and the second encoded picture is decoded by bidirectional prediction using both past and future reference image data. The image data accumulation control unit stores the decoded image data in the frame memory 45. The reference image data acquisition unit reads the reference image data for the decoding from the frame memory and inputs the reference image data to the motion compensation image data generation unit. Here, when the third encoded picture is decoded, the reference image data acquisition unit supplies pixel data having a smaller number of bits to the motion compensation image data generation unit than when other encoded pictures are decoded. input.

FIG. 1 is a configuration example of an image data encoding device, and is a diagram shown as a premise for explaining the present invention. FIG. 2 is a configuration example of the image data decoding apparatus, and is a diagram shown as a premise for explaining the present invention. FIG. 3 is a diagram illustrating a block configuration of the image processing apparatus according to the embodiment. FIG. 4A is a diagram illustrating an example of a format of image data stored in the frame memory according to the embodiment. FIG. 4B is a diagram illustrating another example of a format of image data stored in the frame memory according to the embodiment.

Hereinafter, embodiments will be described with reference to the drawings. FIG. 1 is a configuration example of an image data encoding device, and is a diagram shown as a premise for explaining the present invention. Input image data from the input terminal 11 is input to the subtractor 12 and to the motion detector 23.

The subtracter 12 outputs difference image data that is a difference between the input image data and the predicted image data. The transform and quantization unit 13 transforms the difference image data in the spatial domain into, for example, frequency domain image data (digital cosine transform (DCT)), and converts the transformed data into an inverse quantization and inverse transform unit 14 and an entropy code. To the conversion unit 26.

The inverse quantization and inverse transform unit 14 restores the encoded difference image data and inputs it to the adder 15. The adder 15 combines the difference image data and the predicted image data from the selection unit 17 to obtain restored image data corresponding to the input image data. The restored image data is input to the intra prediction unit 16 and the deblocking filter 21.

The intra prediction unit 16 generates predicted pixel data using a plurality of pixel data in one frame. Therefore, the predicted image data is almost the same as the input image data. The restored image data output from the adder 15 is input to the deblocking filter 21 and is a filter for reducing distortion between blocks that are quantization units.

The output of the deblocking filter 22 is input to the frame memory 22.

In the description of the embodiment, the image data will be described as a collection of a plurality of pixel data.

The restored image data stored in the frame memory 22 is input to the motion compensation prediction unit 24 and also input to the motion detection unit 23. The motion detection unit 23 compares the restored image data with the input image data, and detects a motion prediction vector representing the motion of the image.

The motion compensation prediction unit 24 corrects the restored image data from the frame memory 22 based on the motion prediction vector, and converts it into motion image predicted image data.

The predicted image data is input to the subtracter 12. The subtracter 12 outputs difference image data that is a difference between the input image data and the predicted image data.

The difference image data is encoded by the transform and quantization unit 13 and further subjected to entry-pe encoding by the entry-pe encoding unit 26. The entropy encoding unit 26 outputs a bit stream including encoded data and a motion prediction vector. The bit stream also includes information indicating the picture type (I picture or P or B picture) along with the encoded image data.

An I picture is data encoded by being closed within a picture and does not refer to other pictures (intra-coded picture).

The P picture is a picture (predictive-coded picture) that is encoded using only forward prediction for prediction, that is, predicting forward and predicting the current frame from past frames.

The B picture is a picture (Bidirectionally predictive-coded picture) that can use forward prediction, backward prediction, and bidirectional prediction. Here, in addition to forward prediction, a future frame in time is encoded first, and then prediction (reverse prediction backwardbackprediction) is performed, or bidirectional prediction using both past and future ( Interpolation (prediction) can improve the prediction efficiency.

FIG. 2 is a configuration example of an image data decoding device, and is a diagram shown as a premise for explaining the present invention.

The input terminal 30 receives a bit stream output from the entropy encoding unit 26 in FIG. The data that has been entry-py decoded by the entropy decoding unit 31 is input to the inverse quantization and inverse transform unit 32.

The inverse quantization and inverse transform unit 32 performs inverse quantization and inverse transform on the encoded image data, and the restored differential image data is input to the adder 33. The adder 33 adds the restored image data and the difference image data from the selection unit 34.

The adder 33 adds the difference image data and the restored image data, and outputs decoded image data. The decoded image data that is the output of the adder 33 is input to the deblocking filter 40 and subjected to deblocking filtering to become output image data.

The decoded image data that is the output of the adder 33 is input to the intra prediction unit 41. The intra prediction unit 41 decodes the I picture using the intra-frame image.

The output of the deblocking filter 40 is input to the frame memory 45. The output of the frame memory 45 generates motion prediction image data by using the motion prediction vector that has been sent. When the picture type is P or B, the selection unit 34 selects the motion image from the motion compensation prediction unit 46 and supplies it to the adder 33.

FIG. 3 is a diagram illustrating a block configuration of the image processing apparatus according to the embodiment. 3 corresponds to the motion compensation prediction unit 46, the selection unit 34, the adder 33, the intra prediction unit 41, and the periphery of the frame memory 45 when corresponding to the block of FIG.

A block surrounded by a broken line is an image data processing unit 500. The image data processing unit 500 includes a motion compensation image data generation unit 101, a reference image data acquisition unit 120, and a current restoration image data accumulation control unit 140.

The motion compensated image data generation unit 101 restores an I picture, a B picture, and a P picture. For example, the following rules are used. For example, when the B picture F2 shown on the left side of FIG. 3 is restored, the I picture F1 and the P picture F4 are referred to and restored. When the B picture F3 is restored, the I picture F1 and the P picture F4 are referenced and restored. Here, the I picture F1 and the P picture F4 as the reference image data are read from the frame memory 45 by the reference image data acquisition unit 120.

Furthermore, when the P picture F3 is restored, the I picture F1 is referenced and restored.

The reference image data acquisition unit 120 includes a reference image data buffer 121, a reference image data area selection control unit 122, a reference image vector generation unit 130, a bandwidth measurement unit 123, and a current restoration image data accumulation control unit 140. .

The current restored image data accumulation control unit 140 performs an operation of writing the restored image data generated by the motion compensated image data generation unit 101 into the frame memory 45. At this time, the current restoration image data accumulation control unit 140 designates the write address of the frame memory 45 via the bus 141 and transfers the write pixel data.

In this case, the current restoration image data accumulation control unit 140 sets the pixel data to the upper 8-bit area data (simply referred to as upper bit image data) and the lower 2-bit area data (simply lower And may be referred to as “side bit image data”). The data format at this time is, for example, a format shown in FIGS. 4A and 4B described later. Note that the 10-bit data may be referred to as all-bit image data.

In the frame memory 45, a read address is designated by the reference image data area selection control unit 122. As a result, the reference image data is read from the frame memory 45 and temporarily stored in the reference image data buffer 121.

The reference image data acquisition unit 120 acquires data for acquiring reference image data necessary for the motion compensated image data generation unit 101.

The reference image data buffer 121 stores reference image data read from the frame memory 45 and functions as a buffer memory that can be accessed at high speed.

The reference image data area selection control unit 122 transfers the reference image data read by accessing the frame memory 45 to the reference image data buffer 121.

The reference image data area selection control unit 122 determines a read address for accessing the frame memory 45 using the reference image vector corresponding to the motion prediction vector from the reference image vector generation unit 130.

The reference image vector generation unit 130 generates the reference image vector by using the motion prediction vector sent from the encoding device in association with the B picture and the P picture.

In the region where the image motion is frequent, the change frequency of the motion prediction vector is high. When the image motion is frequent and the image frequency is high (when the luminance change frequency is high or there are many fine images), the change frequency of the motion prediction vector is high.

The reference image data area selection control unit 122 temporarily stores the reference image data (pixel data is, for example, 10 bits or 8 bits) read from the frame memory 45 in the reference image data buffer 121. The image data stored in the reference image data buffer 121 is transmitted to the motion compensated image data generation unit 101.

In the above apparatus, whether the upper 8-bit area data of the pixel data is read from the frame memory 45 and processed by the control signal of the bandwidth measuring unit 123, or the upper 8-bit area data and It is determined whether 10-bit pixel data including the data of the lower 2-bit area is read from the frame memory 45 and processed.

Thereby, the reference image data acquisition unit 120 and the motion compensated image data generation unit 101 are switched between the 8-bit pixel data processing mode and the 10-bit pixel data processing mode.

Here, an example of the format of the image data stored in the frame memory 45 will be shown with reference to FIGS. 4A and 4B.

The example of FIG. 4A is an example in which one pixel is handled with 10 bits. A first luminance data area (Y1-1) in which upper 8 bits of luminance pixel data are stored, and a first color data area (C1-1) in which upper 8 bits of color difference pixel data Cb and Cr are stored ), A second luminance data area (Y1-2) in which the lower 2 bits of the luminance pixel data are stored, and a second color data area (Y1-2) in which the lower 2 bits of the color difference pixel data Cb and Cr are stored. C1-2) is an example of a format that is one set.

The example of FIG. 4B is an example in which one pixel is handled with 12 bits. A first luminance data area (Y2-1) in which upper 8 bits of luminance pixel data are stored, and a first color data area (C2-1) in which upper 8 bits of color difference pixel data Cb and Cr are stored ), A second luminance data area (Y2-2) in which the lower 4 bits of the luminance pixel data are stored, and a second color data area (Y2-2) in which the lower 4 bits of the color difference pixel data Cb and Cr are stored. C2-2) is an example of a format that is one set.

Note that the number of bits and the division method shown in FIGS. 4A and 4B are not limited, and a division method that facilitates byte alignment is preferable.

Referring back to FIG. As described above, the reference image data acquisition unit 120 and the motion-compensated image data generation unit 101 perform the 8-bit pixel data processing mode, the 10-bit pixel data processing mode, and the like according to the switching control signal of the bandwidth measurement unit 123. Switch to one of the following.

As specific examples, the following embodiments and / or combinations thereof are employed.

(1) The switching control signal of the bandwidth measurement unit 123 controls the reference image data area selection unit control unit 122, for example, when a B picture restoration process is executed. As a result, the entire reference image data acquisition unit 120 reads 8-bit pixel data from the frame memory 45 and supplies it to the motion compensated image data generation unit 101.

Therefore, the processing load on the reference image data acquisition unit 120 and the motion compensation image data generation unit 101 is reduced as compared with the case of processing 10-bit pixel data.

When 8-bit pixel data is employed, the lower-order 2-bit pixel data is input as a fixed value to the motion compensated image data generating unit 101, and the operation of the motion compensated image data generating unit 101 is detailed. There is no need to switch.

In addition, when a gradation image with upper 8 bits is displayed on the liquid crystal panel, the lower 2 bits are an error in restoration. However, in the liquid crystal panel, it is difficult to distinguish the gradation based on the lower bits of the pixel data unless approaching the vicinity of the screen. In addition, some LCD panels, which are called 10-bit compatible, are shown as 10 bits in a pseudo manner by an 8-bit dithering method in the circuit in the liquid crystal module, and the original 10-bit gradation can be expressed. Some are not.

Therefore, in the embodiment, it is preferable to use 8-bit pixel data for a B picture that is relatively difficult to identify a restoration error.

(2) The reference image data acquisition unit 120 and the motion compensation image data generation unit 101 are switched to the 8-bit image data processing mode. For example, when the number of bits to be processed per unit time exceeds the first threshold value (when the bandwidth becomes large), 8-bit pixel data is read from the frame memory 45 and supplied to the motion compensation image data generation unit 101. To do. Further, when the number of bits to be processed per unit time becomes smaller than the second threshold (when the bandwidth becomes small), 10-bit pixel data is read from the frame memory 45 and the motion compensation image data generation unit 101 is read. To supply.

The number of bits to be processed per unit time can be estimated, for example, by monitoring a motion prediction vector. For example, when the value of the motion prediction vector is small and the number of motion prediction vectors for reproducing one picture is large, it can be determined that the bandwidth becomes large.

(3) When restoring the previous frame (previous frame), if the bandwidth exceeds or is very close to the preset bandwidth, that is, it is processed in unit time. When it is measured that the number of bits of power pixel data has exceeded a predetermined threshold, when restoring the next frame, the mode may be shifted to the 8-bit pixel data processing mode.

This idea is to predict that the bandwidth of the previous frame will be large, so the next frame will also have a large bandwidth. In order to shift to the 10-bit pixel data processing mode, the shift is made when the number of bits to be processed per unit time is smaller than a preset second threshold value (when the bandwidth is reduced).

The above description refers to the bandwidth when the immediately preceding frame restoration process is performed, but depending on the result of statistics of a plurality of frames, the 10-bit pixel data processing mode or the 8-bit pixel data processing You may enter mode.

(4) Furthermore, it may be determined whether the bandwidth increases or decreases depending on whether the motion prediction vector includes a decimal point. The motion prediction vector including a decimal point means that it is necessary to generate a complementary pixel between the original pixel and the original pixel, not the motion prediction in the original pixel array unit. In such a case, it is necessary to use many peripheral pixels in order to increase the accuracy of the interpolation pixel. As a result, the bandwidth related to image data processing increases.

Based on the above idea, the bandwidth measuring unit 123 of the present apparatus counts motion prediction vectors including a decimal point among a plurality of motion prediction vectors used to restore one frame. Then, the bandwidth measuring unit 123 detects a percentage or several of the plurality of motion prediction vectors including a decimal point, and according to the detection result, the 10-bit pixel data processing mode Alternatively, the transition to the 8-bit pixel data processing mode may be controlled.

(5) The reference image data acquisition unit 120 is configured to read 8-bit pixel data from the frame memory 45 and supply the 8-bit pixel data to the motion compensated image data generation unit 101 according to a device in which the image data processing device 500 is incorporated. May be. This is particularly effective when the device to be incorporated is a mobile device that requires power saving.

(6) The transition to the 10-bit pixel data processing mode and the 8-bit pixel data processing mode may be controlled dynamically by selectively combining the above (1) to (4).

In FIG. 3, the image data processing apparatus 500 is shown as a block configuration, but the operation of this block may be configured by a program (software). Further, it may be configured by a plurality of data processors or DSPs (digital signal processors), and the function of the block may be realized by instructions (instructions) stored in a computer-readable recording medium.

As described above, the motion compensated image data generation unit 101 decodes the first encoded picture (I picture) by intra prediction using the reference image data in the frame, and outputs the second encoded picture (P Picture) is decoded by forward prediction using reference image data of a past frame, and the second encoded picture (B picture) is decoded by bidirectional prediction using both past and future reference image data. The image data accumulation control unit 140 stores the decoded image data in the frame memory 45. The reference image data acquisition unit 120 reads reference image data for the decoding from the frame memory 45 and inputs it to the motion compensation image data generation unit. Here, when the second encoded picture is decoded, the reference image data acquisition unit 120 receives the pixel data having a smaller number of bits when the second encoded picture is decoded than when the other encoded picture is decoded. To enter.

In the above embodiment, it has been described that the image data accumulation control unit 140 stores the restored image data in the frame memory 45 in the format of FIG. 4A or 4B.

However, the present invention is not limited to this method. For example, only the B picture may be divided, that is, divided into the upper first bit number (8) and the lower second bit number (2).

In the above embodiment, the pixel data stored in the frame memory 45 is described as being stored in the upper 8 bits and the lower 2 bits. However, the present invention is not limited to this, all the pixel data is stored in the frame memory 45 in units of 10 bits, and when the reference image data is taken into the reference image data buffer 121, only the upper 8 bits or 10 bits are selected. And may be operated so as to be captured.

In the above description, the term “part” may be replaced with “device”, “vessel”, “block”, and “module”, and is of course included in this embodiment. Furthermore, in the constituent elements of the claims, even when the constituent elements are expressed in a divided manner, when they are expressed in combination, or when they are expressed in combination, they are included in this embodiment. Even if the claims are expressed as a method, the apparatus of this embodiment is applied.

Although several embodiments have been described, these embodiments are presented as examples, and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

The basic components in the above embodiment will be added below.

(1) A first encoded picture is decoded by intra prediction using reference image data in a frame, and a second encoded picture is decoded by forward prediction using reference image data of a past frame; A motion compensated image data generation unit 101 that decodes the third encoded picture by bidirectional prediction using both past and future reference image data;
An image data accumulation control unit 140 for storing the decoded image data in the frame memory 45;
A reference image data acquisition unit 120 that reads reference image data for the decoding from the frame memory 45 and inputs the reference image data to the motion compensation image data generation unit,
When the third encoded picture is decoded, the reference image data acquisition unit 120 reads out pixel data having a bit number smaller than the number of bits when another encoded picture is decoded, and the motion compensated image An image processing apparatus that inputs data to a data generation unit.

(2) The image data accumulation control unit 140 stores the pixel data in the frame memory 45 by dividing the pixel data into an upper first bit number and a lower second bit number (1). Image processing apparatus.

(3) The reference image data acquisition unit 120 has a bandwidth measurement unit 123, and this bandwidth measurement unit measured that the number of bits of pixel data to be processed per unit time exceeded the first threshold value. In this case, the image processing apparatus according to (1), wherein the reference image data acquisition unit and the motion compensation image data generation unit are switched to a higher-order bit image data processing mode.

(4) The reference image data acquisition unit 120 includes a bandwidth measurement unit 123. This bandwidth measurement unit determines the number of bits of pixel data to be processed per unit time when a preceding frame is restored. The image processing apparatus according to (1), wherein when it is measured that a predetermined threshold value is exceeded, the reference image data acquisition unit and the motion compensation image data generation unit are switched to an upper bit image data processing mode.

(5) The reference image data acquisition unit 120 includes a bandwidth measurement unit 123, and the bandwidth measurement unit includes motion prediction including a decimal point among a plurality of motion prediction vectors used to restore one frame. The image processing apparatus according to (1), wherein when there are a plurality of vectors, the reference image data acquisition unit and the motion compensated image data generation unit are switched to an upper bit image data processing mode.

(6) The motion compensated image data generating unit 101 decodes the first encoded picture by intra prediction using the reference image data in the frame, and the second encoded picture is the reference image data of the past frame. The third encoded picture is decoded by bidirectional prediction using both past and future reference image data, and the image data accumulation control unit 140 converts the decoded image data into a frame. In the image processing method of storing in the memory 45, the reference image data acquisition unit 120 reads the reference image data for the decoding from the frame memory 45, and inputs the reference image data to the motion compensation image data generation unit.
When the reference image data acquisition unit 120 decodes the third encoded picture, the motion compensated image data is read by reading out pixel data having a bit number smaller than the number of bits when another encoded picture is decoded. An image processing method to be input to the generation unit.

(7) The image data accumulation control unit 140 stores the pixel data in the frame memory 45 by dividing the pixel data into the upper first bit number and the lower second bit number. Image processing method.

(8) When the bandwidth measurement unit of the reference image data acquisition unit 120 measures that the number of bits of pixel data to be processed per unit time exceeds a first threshold, the reference image data acquisition unit and the reference image data acquisition unit The image processing method according to (6), wherein the motion compensated image data generation unit is switched to a higher-order bit image data processing mode.

(9) The bandwidth measurement unit 123 of the reference image data acquisition unit 120 includes a plurality of motion prediction vectors including a decimal point among a plurality of motion prediction vectors used to restore one frame. The image processing method according to (6), wherein the reference image data acquisition unit and the motion compensation image data generation unit are switched to a higher-order bit image data processing mode.

(10) The bandwidth measurement unit 123 of the reference image data acquisition unit 120 measures that the number of bits of pixel data to be processed per unit time exceeds a predetermined threshold when the preceding frame is restored. In such a case, the image processing method according to (6), wherein the reference image data acquisition unit and the motion compensation image data generation unit are switched to a higher-order bit image data processing mode.

(11) An information recording medium readable by a computer, and instructions stored in the information recording medium are
The motion compensated image data generating unit 101 decodes the first encoded picture by intra prediction using the reference image data in the frame, and converts the second encoded picture into the reference image data of the past frame. An instruction to decode by the forward prediction used, and to decode the second encoded picture by bidirectional prediction using both past and future reference image data;
A command for the image data storage control unit 140 to store the decoded image data in the frame memory 45;
A command for causing the reference image data acquisition unit 120 to read reference image data for the decoding from the frame memory 45;
Here, when the third encoded picture is decoded, the reference image data acquisition unit 120 reads out pixel data having a bit number smaller than the number of bits when another encoded picture is decoded. A computer-readable information recording medium having instructions to be input to the motion compensation image data generation unit.

(12) A command for the image data accumulation control unit 140 to store the pixel data in the frame memory 45 by dividing the pixel data into the upper first bit number and the lower second bit number. The computer-readable information recording medium according to (11).

(13) The reference image data acquisition unit 120 includes a bandwidth measurement unit 123. The reference image data acquisition unit 120 indicates that the number of bits of pixel data to be processed per unit time exceeds the first threshold. The computer-readable information recording medium according to (11), further comprising an instruction to switch the reference image data acquisition unit and the motion compensation image data generation unit to a higher-order bit image data processing mode when measured.

(14) The reference image data acquisition unit 120 includes a bandwidth measurement unit 123, and when the preceding frame is subjected to restoration processing, the bit of pixel data to be processed in a unit time. (11) The computer according to (11), further comprising: an instruction to switch the reference image data acquisition unit and the motion compensation image data generation unit to an upper bit image data processing mode when it is measured that the number exceeds a predetermined threshold. A readable information recording medium.

(15) The reference image data acquisition unit 120 includes a bandwidth measurement unit 123. The reference image data acquisition unit 120 includes a decimal point among a plurality of motion prediction vectors used to restore one frame with respect to the bandwidth measurement unit. The computer-readable information according to (11), further comprising an instruction to switch the reference image data acquisition unit and the motion compensation image data generation unit to an upper bit image data processing mode when a plurality of motion prediction vectors exist. recoding media.

(16) It is composed of a plurality of data processors, and the plurality of data processors are
The first encoded picture is decoded by intra prediction using the reference image data in the frame, the second encoded picture is decoded by forward prediction using the reference image data of the past frame, and the second A processor that realizes a motion compensated image data generation unit 101 that decodes an encoded picture by bidirectional prediction using both past and future reference image data;
A processor for realizing the image data accumulation control unit 140 for storing the decoded image data in the frame memory 45;
A processor that realizes a reference image data acquisition unit 120 that reads reference image data for decoding from the frame memory 45;
Here, when the third encoded picture is decoded, the reference image data acquisition unit 120 reads out pixel data having a bit number smaller than the number of bits when another encoded picture is decoded. An apparatus and / or method having a processor for inputting to the motion compensation image data generation unit.

(17) When the processor realizing the image data accumulation control unit 140 stores the pixel data in the frame memory 45, the pixel data is stored separately in the upper first bit number and the lower second bit number. (16) The apparatus and / or method according to the above.

(18) The reference image data acquisition unit 120 includes a processor that implements the bandwidth measurement unit 123, and the number of bits of pixel data to be processed per unit time has a first threshold value with respect to the bandwidth measurement unit. The apparatus and / or method according to (16), wherein, when exceeding is measured, the reference image data acquisition unit and the motion compensation image data generation unit are switched to an upper bit image data processing mode.

(19) The reference image data acquisition unit 120 includes a processor that implements the bandwidth measurement unit 123. When the preceding frame is restored, the reference image data acquisition unit 120 should process the unit of time. An instruction to switch the reference image data acquisition unit and the motion compensation image data generation unit to the upper bit image data processing mode when it is measured that the number of bits of the pixel data exceeds a predetermined threshold value (16 ) Described apparatus and / or method.

(20) The reference image data acquisition unit 120 includes a processor that implements a bandwidth measurement unit 123. Among the plurality of motion prediction vectors used to restore one frame for the bandwidth measurement unit. The apparatus according to (16), further comprising: an instruction to switch the reference image data acquisition unit and the motion compensation image data generation unit to an upper bit image data processing mode when a plurality of motion prediction vectors including a decimal point exist. And / or method.

Furthermore, it is needless to say that the concepts described in the above (1) to (20) can be applied to devices such as a television receiver, a portable terminal, a personal computer, and a player. The above concept can also be used when power saving is performed for power saving in a portable terminal or the like.

DESCRIPTION OF SYMBOLS 12 ... Adder, 13 ... Transformation and quantization part, 14 ... Inverse quantization and inverse transformation part, 15 ... Adder, 16 ... Intra prediction part, 17 ... Selection part , 21 ... Deblocking filter, 22 ... Frame memory, 23 ... Motion detection unit, 24 ... Motion compensation prediction unit, 31 ... Entropy decoding unit, 32 ... Inverse quantization and Inverse transform unit, 33 ... adder, 40 ... deblocking filter, 41 ... intra prediction unit, 45 ... frame memory, 46 ... motion compensation prediction unit, 500 ... image data processing , 101... Compensation image data generation unit, 120... Reference image data acquisition unit, 140... Current restoration image data accumulation control unit, 121... Reference image data buffer, 122. Area selection control unit 130 ... reference image vector generation unit, 123 ... bandwidth measuring section, 140 ... current restored image data storage control unit.

Claims

The first encoded picture is decoded by intra prediction using the reference image data in the frame, the second encoded picture is decoded by forward prediction using the reference image data of the past frame, and the third A motion compensated image data generation unit that decodes an encoded picture by bidirectional prediction using both past and future reference image data;
An image data accumulation control unit for storing the decoded image data in a frame memory;
A reference image data acquisition unit that reads reference image data for the decoding from the frame memory and inputs the reference image data to the motion compensation image data generation unit,
The reference image data acquisition unit
Image processing in which when the third encoded picture is decoded, pixel data having a smaller number of bits than the number of bits when another encoded picture is decoded is read and input to the motion compensation image data generation unit apparatus.
The image data accumulation control unit, for the frame memory,
The image processing apparatus according to claim 1, wherein the pixel data is stored by being divided into an upper first bit number and a lower second bit number.
The reference image data acquisition unit has a bandwidth measurement unit, and when the bandwidth measurement unit measures that the number of bits of pixel data to be processed per unit time exceeds a first threshold, the reference image data The image processing apparatus according to claim 1, wherein the acquisition unit and the motion compensation image data generation unit are switched to a higher-order bit image data processing mode.
The reference image data acquisition unit has a bandwidth measurement unit, and the bandwidth measurement unit has a number of bits of pixel data to be processed per unit time exceeding a predetermined threshold when a preceding frame is restored. The image processing apparatus according to claim 1, wherein when the measurement is measured, the reference image data acquisition unit and the motion compensation image data generation unit are switched to a higher-order bit image data processing mode.
The reference image data acquisition unit includes a bandwidth measurement unit, and the bandwidth measurement unit includes a plurality of motion prediction vectors including a decimal point among a plurality of motion prediction vectors used to restore one frame. The image processing apparatus according to claim 1, wherein the reference image data acquisition unit and the motion compensation image data generation unit are switched to a higher-order bit image data processing mode.
The motion compensated image data generation unit decodes the first encoded picture by intra prediction using the reference image data in the frame, and the second encoded picture in the order using the reference image data of the past frame. Decoding by direction prediction, decoding the third encoded picture by bidirectional prediction using both past and future reference image data,
An image data accumulation control unit stores the decoded image data in a frame memory,
In an image processing method in which a reference image data acquisition unit reads reference image data for the decoding from the frame memory and inputs the reference image data to the motion compensation image data generation unit,
When the reference image data acquisition unit decodes the third encoded picture, the motion compensation image data generation is performed by reading out pixel data having a bit number smaller than the number of bits when another encoded picture is decoded. Enter in the department,
Image processing method.
The image processing method according to claim 6, wherein the image data accumulation control unit stores the pixel data in the frame memory by dividing the pixel data into a first bit number on the upper side and a second bit number on the lower side.
When the bandwidth measurement unit of the reference image data acquisition unit measures that the number of bits of pixel data to be processed per unit time has exceeded a first threshold, the reference image data acquisition unit and the motion compensation image data The image processing method according to claim 6, wherein the generation unit is switched to a higher-order bit image data processing mode.
When there are a plurality of motion prediction vectors including a decimal point among a plurality of motion prediction vectors used by the bandwidth measurement unit of the reference image data acquisition unit to restore one frame, the reference image data acquisition unit The image processing method according to claim 6, wherein the motion compensation image data generation unit is switched to an upper bit image data processing mode.
When the bandwidth measurement unit of the reference image data acquisition unit measures that the number of bits of pixel data to be processed per unit time exceeds a predetermined threshold when the preceding frame is restored, the reference The image processing method according to claim 6, wherein the image data acquisition unit and the motion compensation image data generation unit are switched to a higher-order bit image data processing mode.