CN109640079A - The adaptive template prediction technique of bandwidth reduction - Google Patents

The adaptive template prediction technique of bandwidth reduction Download PDF

Info

Publication number
CN109640079A
CN109640079A CN201811261722.7A CN201811261722A CN109640079A CN 109640079 A CN109640079 A CN 109640079A CN 201811261722 A CN201811261722 A CN 201811261722A CN 109640079 A CN109640079 A CN 109640079A
Authority
CN
China
Prior art keywords
adaptive template
epitope
value
reconstructed value
current
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN201811261722.7A
Other languages
Chinese (zh)
Inventor
罗瑜
张莹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Cresun Innovation Technology Co Ltd
Original Assignee
Xian Cresun Innovation Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian Cresun Innovation Technology Co Ltd filed Critical Xian Cresun Innovation Technology Co Ltd
Priority to CN201811261722.7A priority Critical patent/CN109640079A/en
Publication of CN109640079A publication Critical patent/CN109640079A/en
Withdrawn legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/103—Selection of coding mode or of prediction mode
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
    • H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
    • H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/59—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Analysis (AREA)

Abstract

The present invention relates to a kind of adaptive template prediction techniques of bandwidth reduction, comprising: step 1 updates the corresponding adaptive template of current MB；Step 2, the prediction residual that the current MB is obtained according to the updated adaptive template；Step 3 judges whether to obtain and completes all MB prediction residuals, if so, prediction terminates；Otherwise, step 1 is jumped to.The present invention is by defining adaptive template, calculate the predicted pixel values of current predicted macrablock, compared with the conventional method, when the texture of image to be compressed is complex, it defines different adaptive templates and corresponds to different texture regions, it is easier to which the probability for improving the pixel matching selected in pixel and the adaptive template in current predicted macrablock can be improved the precision for seeking complex texture region prediction residual value, theoretical limit entropy is further decreased, bandwidth reduction rate is improved.

Description

The adaptive template prediction technique of bandwidth reduction
Technical field
The present invention relates to a kind of compression technique area, in particular to the adaptive template prediction technique of a kind of bandwidth reduction.
Background technique
Continuous improvement with the public to video quality requirement, the image resolution ratio of video also therewith at the increase of multiple, Thus make the data volume of video image very huge, need to occupy more memory space and transmission bandwidth.In this case, The memory space and transmission bandwidth that image is improved using the BCT Bandwidth Compression Technique in chip just seem particularly necessary.
BCT Bandwidth Compression Technique is mainly made of four parts, includes: prediction module, quantization modules, code control module and entropy are compiled Code module.The wherein prediction module module important as one is using existing spatial redundancies between adjacent pixel, according to neighbour Nearly Pixel Information predicts that current pixel value, the standard deviation of prediction difference is much smaller than the standard deviation of raw image data, because This encodes prediction difference, and being more advantageous to makes the theoretical entropy of image data reach minimum, reaches the mesh for improving compression efficiency 's.The algorithm of prediction module is broadly divided into two classes, texture correlation predictive and pixel value correlation predictive at present.
However, when the texture of image to be compressed is complicated and changeable, when predicting the complex texture region of image to be compressed often Because the correlation between image texture is poor, predictive coding cannot be referred to accurately, and theoretical limit entropy is caused to cannot get maximum The reduction of change influences the quality of prediction module.Therefore, when the texture of image to be compressed is complicated and changeable, prediction module is improved Quality becomes the problem of urgent need to resolve.
Summary of the invention
It therefore, is to solve technological deficiency of the existing technology and deficiency, the present invention proposes a kind of the adaptive of bandwidth reduction Answer template prediction method.
Specifically, a kind of adaptive template prediction technique for bandwidth reduction that one embodiment of the invention proposes, comprising:
Step 1 updates current macro (Macro Block, abbreviation MB) corresponding adaptive template；
Step 2, the prediction residual that the current MB is obtained according to the adaptive template of update；
Step 3 judges whether to obtain and completes all MB prediction residuals, if so, prediction terminates；Otherwise, step is jumped to Rapid 1.
In one embodiment of the invention, before the step 1, further includes:
Step X1, the adaptive template list epitope number and epitope serial number are determined；
Step X2, the adaptive template is filled in initialization.
In one embodiment of the invention, the step 1 includes:
Step 11, the current MB of detection neighboring reference direction MB reconstructed value；
Step 12 has been filled with reconstructed value according in the reconstructed value of the MB in the neighboring reference direction and the adaptive template Whether have consistency, determines the update mode of the adaptive template.
In one embodiment of the invention, the step 12 includes:
If step 121, the neighboring reference direction MB reconstructed value and the adaptive template in filling reconstructed value It is inconsistent, then the reconstructed value of the MB in the neighboring reference direction is updated into the setting epitope serial number to the adaptive template, and By the epitope serial number of the adaptive template from sequential shifts after the setting epitope serial number；
If step 122, the neighboring reference direction MB reconstructed value and the adaptive template in have been filled with reconstruction Value has consistency, then will have been filled with reconstructed value and the setting epitope ordinal position in the consistent adaptive template Reconstructed value replacement.
In one embodiment of the invention, the neighboring reference direction includes upper reference direction, left reference direction, upper left Reference direction or upper right reference direction.
In one embodiment of the invention, the step 12 judge the reconstructed value of the neighboring reference direction MB with it is described The consistent formula of filling reconstructed value in adaptive template are as follows:
Wherein, Cur is the original pixel value of current MB, and CurRec is the reconstructed value of current MB, and ABS is to seek absolute value, Pred is the reconstructed value filled in template, and MBnum is pixel quantity in current MB, and a1 and a2 are weight coefficient, and Thr0 is threshold value.
In one embodiment of the invention, the step 2 includes:
Step 21 chooses M candidate reconstructed value in the updated adaptive template according to the first preset formula；
Step 22, the predicted value for determining current MB by the M candidate reconstructed values using the second preset formula；
Step 23 passes through the point-to-point prediction residual for asking difference to obtain current MB
In one embodiment of the invention, first preset formula in the step 21 are as follows:
Wherein, Cur is the original pixel value of current MB, and Pred is the reconstructed value filled in template；MBnum is in current MB Pixel quantity, c1 and c2 are weight coefficient.
In one embodiment of the invention, second preset formula in the step 22 are as follows:
predwi=(w1*Predi-1+w2*Predi+w3*Predi+1+w4)/4
Wherein, W1, W2, W3, W4 are one group of Prediction Parameters.
In one embodiment of the invention, the step 2 includes:
All reconstructed values in the pixel value of current MB and the updated adaptive template are subjected to adaptive texture Prediction, obtains the prediction residual.
Based on this, the present invention has following advantage:
Beneficial effects of the present invention are mainly manifested in: the present invention is by defining the quantity of adaptive template epitope and the picture of MB The mode of element, calculates the prediction residual of current MB, compared with the existing methods, when the texture of image to be compressed is complex, It defines different adaptive templates and corresponds to different texture regions, it is easier to improve in the pixel and adaptive template in current MB The probability of selected pixel matching can be improved the precision for seeking complex texture region prediction residual value, further decrease theory Limit entropy increases bandwidth reduction rate.
Through the following detailed description with reference to the accompanying drawings, other aspects of the invention and feature become obvious.But it should know Road, which is only the purpose design explained, not as the restriction of the scope of the present invention, this is because it should refer to Appended claims.It should also be noted that unless otherwise noted, it is not necessary to which scale attached drawing, they only try hard to concept Ground illustrates structure and process described herein.
Detailed description of the invention
Below in conjunction with attached drawing, specific embodiments of the present invention will be described in detail.
Fig. 1 is a kind of adaptive template prediction technique flow chart provided in an embodiment of the present invention；
Fig. 2 is a kind of schematic diagram of adaptive template provided in an embodiment of the present invention；
Fig. 3 is the schematic diagram of another adaptive template provided in an embodiment of the present invention；
Fig. 4 is another adaptive template prediction technique flow chart provided in an embodiment of the present invention；
Fig. 5 is another adaptive template prediction technique flow chart provided in an embodiment of the present invention；
Fig. 6 is a kind of schematic diagram of the neighboring reference pixel of adaptive texture prediction provided in an embodiment of the present invention.
Specific embodiment
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing to the present invention Specific embodiment be described in detail.
Embodiment one
Referring to Figure 1, Fig. 1 is a kind of adaptive template prediction technique flow chart provided in an embodiment of the present invention, this method Include the following steps:
Step 1 updates the corresponding adaptive template of current MB；
Step 2, the prediction residual that the current MB is obtained according to the updated adaptive template；
Step 3 judges whether to obtain and completes all MB prediction residuals, if so, prediction terminates；Otherwise, step is jumped to Rapid 1.
Wherein, before step 1, can also include:
Step X1, the adaptive template list epitope number and epitope serial number are determined；
Step X2, the adaptive template is filled in initialization.
Wherein, step 1 may include:
Step 11, the current MB of detection neighboring reference direction MB reconstructed value；
Step 12, judge the neighboring reference direction MB reconstructed value and the adaptive template in filling reconstructed value It is whether consistent, with the update mode of the determination adaptive template.
Wherein, step 12 may include:
If step 121, the neighboring reference direction MB reconstructed value and the adaptive template in filling reconstructed value It is inconsistent, then the reconstructed value of the MB in the neighboring reference direction is updated into the setting epitope serial number to the adaptive template, and By the epitope serial number of the adaptive template from sequential shifts after the setting epitope serial number；
If step 122, the neighboring reference direction MB reconstructed value and the adaptive template in filling reconstructed value Has consistency, then by the reconstruction of filling reconstructed value and the setting epitope ordinal position in the consistent adaptive template Value replacement.
Further, the neighboring reference direction in step 121 and step 122 include upper reference direction, left reference direction, Upper left reference direction or upper right reference direction.
Further, for the consistency judgment formula in step 121 and step 122 are as follows:
Wherein, Cur is the original pixel value of current MB, and CurRec is the reconstructed value of current MB, and ABS is to seek absolute value, Pred is the reconstructed value filled in template, and MBnum is pixel quantity in current MB, and a1 and a2 are weight coefficient, and Thr0 is threshold value.
Wherein, step 2 obtains the prediction residual of the current MB according to the updated adaptive template, including following Step:
Step 21 chooses M candidate reconstructed value in the updated adaptive template according to the first preset formula；
Step 22, the predicted value for determining current MB by the M candidate reconstructed values using the second preset formula；
Step 23 passes through the point-to-point prediction residual for asking difference to obtain current MB.
Wherein, the first preset formula in step 21 are as follows:
Wherein, Cur is the original pixel value of current MB, and Pred is the reconstructed value filled in template；MBnum is in current MB Pixel quantity, c1 and c2 are weight coefficient.
Wherein, the second preset formula in step 22 are as follows:
predwi=(w1*Predi-1+w2*Predi+w3*Predi+1+w4)/4
Wherein, W1, W2, W3, W4 are one group of Prediction Parameters.
Wherein, step 2 obtains the prediction residual of the current MB according to the updated template, can also include:
All reconstructed values in the pixel value of current MB and the updated adaptive template are subjected to adaptive texture Prediction, obtains the prediction residual.
Embodiment two
Fig. 2 is referred to, Fig. 2 is a kind of schematic diagram of adaptive template provided in an embodiment of the present invention.The present embodiment is upper It states and adaptive template is described in detail on the basis of embodiment, the foundation of the adaptive template includes the following steps:
Step 1, the epitope number and epitope serial number for defining adaptive template
Preferably, the epitope number that can define adaptive template is 4,8,16 or 32；The present embodiment is with table Bit quantity illustrates that the epitope of other quantity is similarly for being 16.The epitope number of adaptive template is 16, and epitope serial number is from 0 It is arranged successively to 15, serial number is smaller, and priority is higher, and each epitope records one group of reconstructed value of a MB.MB size can be set, The present embodiment is by taking 16*1 as an example, i.e., the size of each MB is 16*1 pixel, i.e., each MB has 16 reconstructed values.
The initialization filling of step 2, adaptive template
The original state of adaptive template is sky, and the reconstructed value of a certain MB is filled into the epitope of serial number 0；It continues to fill up Adaptive template, by the epitope of the reconstructed value sequential shifts in the epitope of serial number 0 to serial number 1, by the weight of next MB Built-in value is filled into the epitope of serial number 0；And so on, it is when filling adaptive template every time, the N number of epitope position having been filled with is suitable Sequence rearward displacement moves an epitope position, the reconstructed value of MB to be filled is filled into the epitope of serial number 0, until adaptive 16 epitopes filling in template finishes.Specifically: before the reconstructed value filling for carrying out MB every time, from small to large by serial number, detection The consistency of all reconstructed values having been filled in the reconstructed value and list of current MB；If not having consistency, then list is from sequence Number 0 arrives serial number N-1, and N number of active position sequence rearward displacement, the reconstructed value of current MB are placed on 0 position of list altogether.If having consistent Property, template epitope position is constant in list, may be selected to have filled out in adaptive template with the current consistent epitope of MB reconstructed value Reconstructed value is filled to be updated to the reconstructed value of current MB or do not update.The formula for detecting consistency is as follows:
Wherein, Cur is the original pixel value of current MB, and CurRec is the reconstructed value of current MB, and ABS is to seek absolute value, Pred is the reconstructed value filled in template, and MBnum is pixel quantity in current MB, and a1 and a2 are weight coefficient, and Thr0 is threshold value, The value of Thr0 is determined according to user demand.The value of a1 and a2 can be preset fixed value, further, a1+a2 =1, it is preferable that a1 can be chosen for 0.5, a2 and can be chosen for 0.5, a1 and a2 size can also be adjusted flexibly.WhenValue be less than Thr0 When, the value of k is that 1,1 representative has consistency, then may determine that have consistency；Conversely, working asValue be greater than Thr0 when, The value of k is 0, then may determine that not have consistency.
Step 3, adaptive template initialize filled update
After adaptive template initialization filling, remaining MB in detection image updates adaptive template, and update method is such as Under:
If current MB refers to MB there are upper, detection is just upper with reference to the consistent of the reconstructed value having been filled in MB and adaptive template Property, if not having consistency, first by all epitope serial number sequential shifts since 0 of adaptive template, epitope serial number the last one List is removed, then is updated with reference to the reconstructed value of MB to the position of adaptive template epitope serial number 0 by；If having consistency, The reconstructed value that the position of reconstructed value and epitope serial number 0 will be had been filled in consistent adaptive template is exchanged, and can update mould All reconstructed values in plate.
Similarly, if current MB refers to MB there are left, the reconstructed value having been filled in left side reference MB and adaptive template is detected Consistency, if not having consistency, the epitope numeric order displacement that will be first started in adaptive template with epitope serial number 1, The last one removes list to epitope serial number, then the left reconstructed value with reference to MB is updated to the position of adaptive template epitope serial number 1 It sets；If having consistency, the reconstructed value pair of the position of reconstructed value and epitope serial number 1 will be had been filled in consistent adaptive template It changes, and can all reconstructed values in more new template.
If there are upper lefts to refer to MB by current MB, detection upper left refers to the one of the reconstructed value having been filled in MB and adaptive template Cause property will first be shifted if not having consistency in adaptive template with the epitope numeric order that epitope serial number 2 starts, epitope The last one removes list to serial number, then updates with reference to the reconstructed value of MB to the position of adaptive template epitope serial number 2 by；If Having consistency, the reconstructed value that the position of reconstructed value and epitope serial number 2 will be had been filled in consistent adaptive template is exchanged, and All reconstructed values in adaptive template can be updated.
If there are upper right references by current MB, upper right is detected with reference to the consistent of the reconstructed value having been filled in MB and adaptive template Property, if not having consistency, it will first be shifted in adaptive template with the epitope numeric order that epitope serial number 3 starts, epitope sequence Number the last one remove list, then the right reconstructed value with reference to MB is updated to the position of adaptive template epitope serial number 3；If tool Standby consistency, the reconstructed value that the position of reconstructed value and epitope serial number 3 will be had been filled in consistent adaptive template are exchanged, and can To update all reconstructed values in adaptive template.
Wherein, consistency detecting method is with reference to the formula for detecting consistency in step 2.
Embodiment three
Fig. 3 is referred to, Fig. 3 is the schematic diagram of another adaptive template provided in an embodiment of the present invention.The present embodiment exists Another adaptive template proposed by the present invention is described in detail on the basis of above-described embodiment, the adaptive template Foundation includes the following steps:
Step 1 defines adaptive template epitope number and epitope serial number
Preferably, can define adaptive template epitope number is 4,8,16 or 32；The present embodiment is with adaptive Answer template epitope number be 8 for illustrate, the adaptive template epitope of other quantity is similarly.The adaptive template that quantity is 8 Epitope, epitope serial number are arranged successively from 0 to 8, and serial number is smaller, and priority is higher, and each epitope records one group of reconstruction of a MB Value.MB size can be set, and the present embodiment is by taking 8*2 size as an example, i.e., the size of each MB is 8*2 pixel, i.e., each MB has 8*2 A reconstructed value.
Step 2, Adaptive template-updating
The position of adaptive template epitope serial number 4~7 stores preset 4 groups of reconstructed values；Detect phase on current MB The consistency for the reconstructed value being had been filled in the reconstructed value and adaptive template of adjacent MB, if not having consistency, by upper adjacent MB's Reconstructed value is filled into the position of adaptive template epitope serial number 0；If having consistency, will be filled out in consistent adaptive template The reconstructed value for filling the position of reconstructed value and epitope serial number 0 is exchanged, and can all reconstructed values in more new template.
The consistency for detecting the reconstructed value having been filled in the reconstructed value and adaptive template of the left adjacent MB of current MB, if not having The reconstructed value of left adjacent MB is filled into the position of adaptive template epitope serial number 1 by standby consistency；It, will if having consistency The reconstructed value that the position of reconstructed value and epitope serial number 1 is had been filled in consistent adaptive template is exchanged, and can more new template In all reconstructed values.
The consistency for detecting the reconstructed value having been filled in the reconstructed value and adaptive template of the adjacent MB in the current upper left MB, if not Have consistency, the reconstructed value of the adjacent MB in upper left is filled into the position of adaptive template epitope serial number 2；If having consistent Property, the reconstructed value that the position of reconstructed value and epitope serial number 2 will be had been filled in consistent adaptive template is exchanged, and can be updated All reconstructed values in template.
The consistency for detecting the reconstructed value having been filled in the reconstructed value and adaptive template of the adjacent MB of current MB upper right, if not Have consistency, the reconstructed value of the adjacent MB of upper right is filled into the position of adaptive template epitope serial number 3；If having consistent Property, the reconstructed value that the position of reconstructed value and epitope serial number 3 will be had been filled in consistent adaptive template is exchanged, and can be updated All reconstructed values in template.
Wherein, the formula of consistency is detected in consistency detecting method reference implementation example two in step 2.
Example IV
Fig. 4 is referred to, Fig. 4 is the prediction technique flow chart of another adaptive template provided in an embodiment of the present invention.This Embodiment is on the basis of the above embodiments discussed in detail a kind of prediction technique of adaptive template proposed by the present invention, this is pre- Survey method includes the following steps:
The update of step 1, adaptive template
Default adaptive template has carried out initialization filling.If current MB refers to MB there are upper, detection it is just upper with reference to MB with The consistency for the reconstructed value being had been filled in adaptive template, if not having consistency, first by all epitope serial numbers of adaptive template The sequential shifts since 0, the last one removes list to epitope serial number, then updates with reference to the reconstructed value of MB to adaptive template by The position of epitope serial number 0；If having consistency, reconstructed value and epitope serial number 0 will be had been filled in consistent adaptive template Position reconstructed value exchange, and can all reconstructed values in more new template.
Similarly, if current MB refers to MB there are left, the reconstructed value having been filled in left side reference MB and adaptive template is detected Consistency, if not having consistency, the epitope numeric order displacement that will be first started in adaptive template with epitope serial number 1, The last one removes list to epitope serial number, then the left reconstructed value with reference to MB is updated to the position of adaptive template epitope serial number 1 It sets；If having consistency, the reconstructed value pair of the position of reconstructed value and epitope serial number 1 will be had been filled in consistent adaptive template It changes, and can all reconstructed values in more new template.
If there are upper lefts to refer to MB by current MB, detection upper left refers to the one of the reconstructed value having been filled in MB and adaptive template Cause property will first be shifted if not having consistency in adaptive template with the epitope numeric order that epitope serial number 2 starts, epitope The last one removes list to serial number, then updates with reference to the reconstructed value of MB to the position of adaptive template epitope serial number 2 by；If Having consistency, the reconstructed value that the position of reconstructed value and epitope serial number 2 will be had been filled in consistent adaptive template is exchanged, and All reconstructed values in adaptive template can be updated.
If there are upper right references by current MB, upper right is detected with reference to the consistent of the reconstructed value having been filled in MB and adaptive template Property, if not having consistency, it will first be shifted in adaptive template with the epitope numeric order that epitope serial number 3 starts, epitope sequence Number the last one remove list, then the right reconstructed value with reference to MB is updated to the position of adaptive template epitope serial number 3；If tool Standby consistency, the reconstructed value that the position of reconstructed value and epitope serial number 3 will be had been filled in consistent adaptive template are exchanged, and can To update all reconstructed values in adaptive template.
Wherein, the formula of consistency is detected in consistency detecting method reference implementation example two in step 2.
Step 2 obtains the optimal reconstructed value of adaptive template
After Adaptive template-updating is completed, current MB is matched with the reconstructed value of epitope each in list, according to Formula chooses M optimal epitope.Formula is as follows:
Wherein, Cur is the original pixels of current MB, and Pred is the reconstructed value of each epitope filling in adaptive template； MBnum is pixel quantity in current MB, and c1 and c2 are weight coefficient, and final rdo is smaller, then the weight in the adaptive template epitope Built-in value is more excellent.
In one embodiment, the value of c1 and c2 can be preset fixed value, further, for the ease of It calculates, c1 directly can be set as 1, c2 is set as 0.
Step 3 determines weight estimation pixel value
Respectively reconstructed value weighting any in the reconstructed value in M epitope is handled to obtain predicted pixel values.Following formula:
predwi=(w1*Predi-1+w2*Predi+w3*Predi+1+w4)/4
Wherein, W1, W2, W3, W4 are one group of Prediction Parameters, and predw is predicted pixel values, and Pred is M epitope in template In the filling of any epitope reconstructed value, i is sequence of the Pred in epitope.
The various combination of default T kind W1, W2, W3, W4, can be generated T kind predicted pixel values, M table for an epitope There are M*T kind predicted pixel values in position, finally in M*T kind possibility, according to rdo formula, select optimal epitope and corresponding W1, W2,W3,W4.The predicted pixel values that the reconstructed value in the optimal epitope is calculated according to W1, W2, W3, W4, the reference as current MB Value.
In one embodiment, the value of W1, W2, W3, W4 can be preset fixed value, further, W1+ W2+W3=3, it is preferable that W1, W2, W3 are chosen for 1 respectively, and W1, W2, W3 are chosen for 0.5,2,0.5, W1, W2, W3 respectively can be with According to the actual situation, size is adjusted flexibly.Further, W4 can be chosen for the flat of the reconstructed value of all fillings in current epitope Mean value can also be chosen for reconstructed value corresponding with current predictive rank-ordered pixels in epitope.
Further, optimal epitope is the corresponding epitope of minimum value in rdo.
Step 4 seeks residual error
It may be selected point-to-point poor mode or adaptive prediction mode to be asked to seek residual error.Finally by residual values, list epitope serial number Decoding end is sent to the value of W1, W2, W3, W4, wherein point-to-point to seek poor mode to correspond to each pixel value in current MB Subtract the corresponding predicted pixel values of each reconstructed value in optimal epitope.
Further, it rebuilds pixel and refers to that having compressed image MB decompression rebuilds obtained pixel, the pixel value for rebuilding pixel is logical Frequently referred to reconstructed value.According to the available reconstructed value of prediction residual, i.e., reference value is added into the available reconstructed value of prediction residual.
Step 5 judges whether MB is disposed
After current MB completes point-to-point prediction, continue to determine whether that all MB in image complete predicted operation, if so, Then prediction terminates, and otherwise, jumps to step 1, continues the predicted operation of subsequent MB.
Embodiment five
Fig. 5 is referred to, Fig. 5 is the prediction technique flow chart of another adaptive template provided in an embodiment of the present invention.It should Prediction technique includes the following steps:
Current MB is corresponded to Adaptive template-updating by step 1
Defining adaptive template epitope number is 4,8,16 or 32；The present embodiment is with adaptive template epitope number Amount illustrates that the adaptive template epitope of other quantity is similarly for being 8.The adaptive template that quantity is 8, epitope serial number is from 0 It is arranged successively to 8, serial number is smaller, and priority is higher, and each epitope records the reconstructed value of one group of MB.MB size can be set, this reality Example is applied by taking 8*2 size as an example, i.e., the size of each MB is 8*2 pixel component, i.e., each MB has 8*2 reconstructed value.
The position of adaptive template epitope serial number 4~7 stores preset 4 groups of reconstructed values；Detect phase on current MB The consistency for the reconstructed value being had been filled in the reconstructed value and adaptive template of adjacent MB, if not having consistency, by upper adjacent MB's Reconstructed value is filled into the position of adaptive template epitope serial number 0；If having consistency, will be filled out in consistent adaptive template The reconstructed value for filling the position of reconstructed value and epitope serial number 0 is exchanged, and can all reconstructed values in more new template.
The consistency for detecting the reconstructed value having been filled in the reconstructed value and adaptive template of the left adjacent MB of current MB, if not having The reconstructed value of left adjacent MB is filled into the position of adaptive template epitope serial number 1 by standby consistency；It, will if having consistency The reconstructed value that the position of reconstructed value and epitope serial number 1 is had been filled in consistent adaptive template is exchanged, and can more new template In all reconstructed values.
The consistency for detecting the reconstructed value having been filled in the reconstructed value and adaptive template of the adjacent MB in the current upper left MB, if not Have consistency, the reconstructed value of the adjacent MB in upper left is filled into the position of adaptive template epitope serial number 2；If having consistent Property, the reconstructed value that the position of reconstructed value and epitope serial number 2 will be had been filled in consistent adaptive template is exchanged, and can be updated All reconstructed values in template.
The consistency for detecting the reconstructed value having been filled in the reconstructed value and adaptive template of the adjacent MB of current MB upper right, if not Have consistency, the reconstructed value of the adjacent MB of upper right is filled into the position of adaptive template epitope serial number 3；If having consistent Property, the reconstructed value that the position of reconstructed value and epitope serial number 3 will be had been filled in consistent adaptive template is exchanged, and can be updated All reconstructed values in template.
Wherein, the formula of consistency is detected in consistency detecting method reference implementation example two in step 2.
Step 2, adaptive prediction
After Adaptive template-updating is completed, existing MB all in the pixel value and adaptive template of current MB are rebuild Value carries out adaptive texture prediction, solves prediction residual.
Wherein, Fig. 6 is referred to, Fig. 6 is a kind of neighboring reference picture of adaptive texture prediction provided in an embodiment of the present invention The schematic diagram of element.Reference pixel in adaptive texture prediction is selected, A, B, C, E are picture around current pixel is adjacent Reconstructed value corresponding with current pixel in any epitope of element, i.e. template, D are the weight of the left adjacent pixel component of current pixel component Built-in value, wherein pixel A is upper left neighboring reference pixel, pixel B is upper neighboring reference pixel, pixel C is upper right neighboring reference picture Element, the left neighboring reference pixel that pixel D is left neighboring reference pixel, pixel E is pixel A:
If a. ABS (D-E) is minimum, i.e. 135 degree of textures, then reference pixel is pixel A；
If b. ABS (D-A) is minimum, i.e. vertical texture, then reference pixel is pixel B；
If c. ABS (D-B) is minimum, i.e. 45 degree of textures, then reference pixel is pixel C；
If d. ABS (B-A) is minimum, i.e. horizontal texture, then reference pixel is pixel D；
According to aforesaid way, choose reference pixel, traverse all epitopes in template, by the reference pixel of selection most Small value asks difference to obtain the prediction of the mode as final reference pixel, by the pixel value of final reference pixel value and current MB Residual error.
Step 3 judges whether MB is disposed
After current MB completes adaptive prediction, continue to determine whether that all MB in image complete predicted operation, if so, Then prediction terminates, and otherwise, jumps to step 1, continues the predicted operation of subsequent MB.
In conclusion specific case used herein is to the present invention is based on the adaptive template prediction techniques of bandwidth reduction It is expounded, the above description of the embodiment is only used to help understand the method for the present invention and its core ideas；Meanwhile for Those of ordinary skill in the art have change according to the thought of the present invention in specific embodiments and applications Place, in conclusion the contents of this specification are not to be construed as limiting the invention, protection scope of the present invention should be with appended power Subject to benefit requires.

Claims (10)

1. a kind of adaptive template prediction technique based on bandwidth reduction characterized by comprising
Step 1 updates the corresponding adaptive template of current MB；
Step 2, the prediction residual that the current MB is obtained according to the updated adaptive template；
Step 3 judges whether to obtain and completes all MB prediction residuals, if so, prediction terminates；Otherwise, step 1 is jumped to.
2. the method according to claim 1, wherein before the step 1, further includes:
Step X1, the epitope number and epitope serial number of the adaptive template are determined；
Step X2, the adaptive template is filled in initialization.
3. according to the method described in claim 2, it is characterized in that, the step 1 includes:
The neighboring reference direction MB reconstructed value of step 11, the current MB of detection；
Filling reconstructed value in step 12, the reconstructed value for judging the neighboring reference direction MB and the adaptive template whether one It causes, with the update mode of the determination adaptive template.
4. according to the method described in claim 3, it is characterized in that, the step 12 includes:
If step 121, the reconstructed value of the neighboring reference direction MB and the filling reconstructed value in the adaptive template are inconsistent, The MB reconstructed value in the neighboring reference direction is then updated into the setting epitope serial number to the adaptive template, and will be described adaptive Answer the epitope serial number of template from sequential shifts after the setting epitope serial number；
If step 122, the reconstructed value of the neighboring reference direction MB are consistent with the filling reconstructed value in the adaptive template, The reconstructed value of filling reconstructed value and the setting epitope ordinal position in the consistent adaptive template is replaced.
5. according to the method described in claim 4, it is characterized in that, the neighboring reference direction includes upper reference direction, Zuo Can Examine direction, upper left reference direction or upper right reference direction.
6. according to the method described in claim 4, it is characterized in that, the step 12 judges the weight of the neighboring reference direction MB The consistent formula of filling reconstructed value in built-in value and the adaptive template are as follows:
Wherein, Cur is the original pixel value of current MB, and CurRec is the reconstructed value of current MB, and ABS is to seek absolute value, and Pred is The reconstructed value filled in template, MBnum are pixel quantity in current MB, and a1 and a2 are weight coefficient, and Thr0 is threshold value.
7. the method according to claim 1, wherein the step 2 includes:
Step 21 chooses M epitope in the adaptive template updated according to the first preset formula；
Step 22, the predicted pixel values for determining current pixel by the reconstructed value in the M epitope using the second preset formula；
Step 23 passes through the point-to-point prediction residual for asking difference to obtain current MB all pixels.
8. the method according to the description of claim 7 is characterized in that first preset formula in the step 21 are as follows:
Wherein, Cur is the original pixel value of current MB, and Pred is the reconstructed value of each epitope filling in adaptive template；MBnum For pixel quantity in current MB, c1 and c2 are weight coefficient.
9. the method according to the description of claim 7 is characterized in that second preset formula in the step 22 are as follows:
predwi=(w1*Predi-1+w2*Predi+w3*Predi+1+w4)/4
Wherein, W1, W2, W3, W4 are one group of Prediction Parameters, and predw is predicted pixel values, and Pred is to appoint in M epitope in template The reconstructed value of one epitope filling, i are sequence of the Pred in epitope.
10. the method according to claim 1, wherein the step 2 includes:
All reconstructed values in the pixel value of current MB and the updated adaptive template are subjected to adaptive texture prediction, Obtain the prediction residual.
CN201811261722.7A 2018-10-26 2018-10-26 The adaptive template prediction technique of bandwidth reduction Withdrawn CN109640079A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811261722.7A CN109640079A (en) 2018-10-26 2018-10-26 The adaptive template prediction technique of bandwidth reduction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811261722.7A CN109640079A (en) 2018-10-26 2018-10-26 The adaptive template prediction technique of bandwidth reduction

Publications (1)

Publication Number Publication Date
CN109640079A true CN109640079A (en) 2019-04-16

Family

ID=66066808

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811261722.7A Withdrawn CN109640079A (en) 2018-10-26 2018-10-26 The adaptive template prediction technique of bandwidth reduction

Country Status (1)

Country Link
CN (1) CN109640079A (en)

Similar Documents

Publication Publication Date Title
US9451266B2 (en) Optimal intra prediction in block-based video coding to calculate minimal activity direction based on texture gradient distribution
CN104581164B (en) The coding unit of image boundary is carried out method and the equipment of Code And Decode
CN103004201A (en) Video encoding device, video decoding device, video encoding method, and video decoding method
CN107810632B (en) Intra prediction processor with reduced cost block segmentation and refined intra mode selection
CN112055203B (en) Inter-frame prediction method, video coding method and related devices
CN107277511A (en) Picture decoding method and picture decoding apparatus
US20180192070A1 (en) Method of image encoding and facility for the implementation of the method
CN104754338A (en) Selection method and device for intra-frame predication mode
CN113347416B (en) Chroma intra prediction method and device, and computer storage medium
MXPA05007303A (en) Method and device for choosing a mode of coding.
CN109640079A (en) The adaptive template prediction technique of bandwidth reduction
CN111200730A (en) Multi-mode two-stage selection prediction method for complex texture in bandwidth compression
CN114143536B (en) Video coding method of SHVC (scalable video coding) spatial scalable frame
CN109413421A (en) Method for video coding, equipment and video encoding/decoding method, equipment
CN111107344A (en) Video image coding method and device
CN109600607A (en) Rear selection prediction technique in bandwidth reduction
CN109413420B (en) Dual-mode selection prediction method for complex texture in bandwidth compression
CN109510996A (en) Rear selection prediction technique in bandwidth reduction
CN109640092A (en) Rear selection prediction technique in bandwidth reduction
CN109561302A (en) Adaptive forecasting method based on video compress
CN109391820A (en) Prediction technique based on video compress
CN111200731A (en) Multi-mode two-stage selection prediction method for complex texture in bandwidth compression
CN109842798A (en) The double mode of complex texture selects prediction technique in bandwidth reduction
CN109561314B (en) Self-adaptive template prediction method for bandwidth compression
CN109302615B (en) Self-adaptive template prediction method for video compression

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20190416