CN101931800B - Method and system for encoding region of interest by limited variable bit rate control - Google Patents
Method and system for encoding region of interest by limited variable bit rate control Download PDFInfo
- Publication number
- CN101931800B CN101931800B CN 200910148175 CN200910148175A CN101931800B CN 101931800 B CN101931800 B CN 101931800B CN 200910148175 CN200910148175 CN 200910148175 CN 200910148175 A CN200910148175 A CN 200910148175A CN 101931800 B CN101931800 B CN 101931800B
- Authority
- CN
- China
- Prior art keywords
- macro zone
- quantization parameter
- zone block
- bit rate
- interest
- 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.)
- Active
Links
Images
Abstract
The invention relates to a method and system for encoding a region of interest by limited variable bit rate control, which improves the picture quality of images of a motion region and a region of interest of human eyes at a limited bit rate, and achieves the purpose of saving the encoding bit rate by rejecting high-frequency coefficient information of much background images. Compared with ROI encoding by constant bit rate control (CBR) taking a traditional rate-distortion model as the basis, the limited variable bit rate control algorithm of the invention is matched with a quick block decision mechanism, has lower encoding and decoding complexity, and is more suitable for an instant monitoring coding system.
Description
Technical field
The present invention is relevant for a kind of encoding region of interest method and system, and is particularly to a kind of area-of-interest (Regions Of Interest, ROI) coding method and system that utilizes limited variable bit rate control.
Background technology
Because the demand of video monitoring grows with each passing day, it is more and more important that the field that makes video be coded in Video processing seems.In Video processing, the particular range in the picture is called interesting areas (being designated hereinafter simply as ROI).It is identical that traditional video coding technique is that the priority with all macro zone blocks (Macrob1ocks) in the picture all is considered as, and thus, more important macro zone block possibly can't be known and presents in the picture.
For instance, in video monitoring was handled, ROI was often referred to face area, under situation about not have through special processing, did not have delay in order to make video now being, and the face area in the picture and other moving region will can be not clear especially.If desire to provide high-resolution picture, may produce unnecessary hardware or software cost.
In addition, except face area, the grand zone in the componental movement zone also may present with higher resolution, for example, and the vehicle in people's limb action or the movement.
In the monitoring environment of a fixed background and under the condition of limited bit rate (Bitrate) (transmitting bandwidth), in order to improve the image quality of moving region and ROI, a design method for video coding good and innovation has its necessity.
Summary of the invention
A kind of encoding region of interest method and system of utilizing limited variable bit rate control are provided, can be under limited bit rate, improve the image frame quality of moving region and human eye area-of-interest, and by giving up the high frequency coefficient information of more background image, and save coding bit rate.Compared to the ROI coding of controlling (CBR) based on the fixed bit rate of traditional bit rate-distortion model, limited variable bit rate control algorithm a kind of quick block decision-making mechanism of arranging in pairs or groups, its encoding and decoding complexity is lower, is more suitable for for the immediately monitoring coded system.
A kind of embodiment that utilizes the encoding region of interest method of limited variable bit rate control.The method can be written into computer installation and carry out, and comprise: according to the position of profile in video pictures of area-of-interest, with a plurality of macro zone blocks of this video pictures be divided at least one macro zone block interested, at least one profile extends macro zone block and at least one background macro zone block; And give macro zone block interested, profile extension macro zone block and the different coding priority of background macro zone block, produce a ROI mapping table thus; According to size and the picture update rate of target bit rate, video pictures, determine the macro zone block interested of video pictures, the initial value that profile extends one group of quantization parameter of macro zone block and background macro zone block; According to this group quantization parameter macro zone block interested, profile extension macro zone block and the background macro zone block of video pictures are encoded; And this group quantization parameter of dynamically adjusting macro zone block interested, profile extension macro zone block and the background macro zone block of video pictures.
A kind of embodiment that utilizes the encoding region of interest system of limited variable bit rate control, it comprises the block decision-making module, in order to the position of profile in video pictures according to area-of-interest, a plurality of macro zone blocks of this video pictures are divided at least one macro zone block interested, at least one profile extension macro zone block and at least one background macro zone block, and give macro zone block interested, profile extension macro zone block and the different coding priority of background macro zone block, produce a ROI mapping table thus; The quantization parameter initialization module in order to size and the picture update rate according to target bit rate, video pictures, determines the macro zone block interested of video pictures, the initial value that profile extends one group of quantization parameter of macro zone block and background macro zone block; Module is drilled in limited variable bit rate control, in order to according to this group quantization parameter macro zone block interested, profile extension macro zone block and the background macro zone block of video pictures are encoded; And the parameter update module, in order to the macro zone block interested of dynamically adjusting video pictures, this group quantization parameter that profile extends macro zone block and background macro zone block.
A kind of embodiment of computer recordable media, in order to store computer program, above-mentioned computer program comprises a plurality of procedure code fragments, it is in order to be loaded in the computer system and to make above-mentioned computer system carry out a kind of encoding region of interest method of utilizing limited variable bit rate control, its step comprises the position of profile in video pictures according to area-of-interest, with a plurality of macro zone blocks of this video pictures be divided at least one macro zone block interested, at least one profile extends macro zone block and at least one background macro zone block; And give macro zone block interested, profile extension macro zone block and the different coding priority of background macro zone block, produce a ROI mapping table thus; According to size and the picture update rate of target bit rate, video pictures, determine the macro zone block interested of video pictures, the initial value that profile extends one group of quantization parameter of macro zone block and background macro zone block; According to this group quantization parameter macro zone block interested, profile extension macro zone block and the background macro zone block of this video pictures are encoded; And this group quantization parameter of dynamically adjusting macro zone block interested, profile extension macro zone block and the background macro zone block of video pictures.
Description of drawings
Fig. 1 shows the framework embodiment schematic diagram of the encoding region of interest system that utilizes limited variable bit rate control.
Fig. 2 A shows that the generation of one embodiment of the invention has the flow chart of steps of the ROI mapping table of different coding priority.
Fig. 2 B shows the schematic diagram of the video pictures index value of an embodiment.
Fig. 2 C shows the schematic diagram of the location matrix of an embodiment.
Fig. 3 shows the flow chart of steps of dynamically adjusting the quantification parameter group of an embodiment.
Fig. 4 shows the flow chart of steps of initial value of the decision quantization parameter group of an embodiment.
Fig. 5 is presented at the schematic diagram of the bit rate-distortion model under the different complexities.
Fig. 6 shows the comparison schematic diagram of the experimental data of utilizing different coding method gained.
Fig. 7 A～7E shows that utilization is according to the schematic diagram of the decoded result of the method gained of embodiment technology.
The main element symbol description:
10～block decision-making module
11～ROI mapping table
12～quantization parameter decision-making module
13～quantization parameter class interval
14～quantization parameter initialization module
The initial value of 15～quantization parameter
16～limited variable bit rate control calculation module
17～residual bit rate
18～parameter update module
100～encoding region of interest system
S201..S212～process step
S301..S315～process step
S401..S409～process step
Embodiment
Embodiment cited below particularly describes it in detail, and cooperates appended graphic Fig. 1 to Fig. 7, is described in detail, and provides different embodiment that different execution modes according to the technology of the present invention feature are described.Wherein, the usefulness that is configured to explanation of each element among the embodiment is not in order to limit embodiment.And the part of reference numerals repeats among the embodiment, is for the purpose of simplifying the description, is not the relevance that means between the different embodiment.
Below be that the area-of-interest (ROI) of many motion priority areas (multiple moving priority regions) characteristic describes as implementing example with utilizing H.264.
Disclose an encoding region of interest method and system embodiment who utilizes limited variable bit rate to control, comprise the fixedly block decision-making mechanism of many mobile ROI of ROI/ of an odd number, and a limited variable bit rate control algorithm based on ROI.This sample method detects the moving region automatically with the algorithm that system utilizes the standard object to follow the trail of, perhaps with the self-defining mode of user determine in the monitored picture ROI with improve in the zone image quality (namely, image signal noise is than (Peak Signal to Noise Ratios, PSNR)).Therefore, when safeguarding the monitoring message quality, can take compression usefulness and coding rate into account, more can take into account compatibility, its technology will meet international video standard standard (for example, advanced video encoding (Advanced Video Coding, AVC) etc.) H.264/.
Fig. 1 shows the configuration diagram of the encoding region of interest system embodiment of utilizing limited variable bit rate control.
As shown in Figure 1, encoding region of interest system 100 among this embodiment comprises block decision-making module 10, quantization parameter (Quantification Parameter, QP) (QP-delta) decision-making module 12, quantization parameter initialization (QP Initialization) module 14, limited variable bit rate control are calculated module 16 and parameter update module 18, but are not limited to this.
In this embodiment, can utilize known object cutting/tracking (object segmentation/tracking) or user self-defined (user define) mode to depict zone/profile (blob/contour) that desire is followed the trail of object.For example, desire tracking object can be considered the area-of-interest (ROI) among this embodiment, but is not limited to this.
Block decision-making module 10 is in order to the position of profile in video pictures according to area-of-interest, a plurality of macro zone blocks of video pictures are divided at least one macro zone block interested, at least one profile extension macro zone block and at least one background macro zone block, and give this macro zone block interested, this profile extension macro zone block and the different coding priority (distortion level) of this background macro zone block, produce ROI mapping table (Mapping Table) 11 thus.For example, the macro zone block of video pictures can be distinguished into three major types according to its priority of encoding (distortion level), and the 1st class is ROI itself, and the 2nd class is that partly (ROI-contour extension) extended in the ROI zone, and the 3rd class is background (Background).In other words, macro zone block interested belongs to the 1st class, profile extends macro zone block and belongs to the 2nd class, and the background macro zone block belongs to the 3rd class.The detailed action of block decision-making module 10 will describe in detail in follow-up cooperation Fig. 2 A again.
Quantization parameter decision-making module 12 determines one group of optimal quantization parameter class interval, and { wherein D1 and D2 value represent quantization parameter { QP_ROI, QP_ROI+D1, the gap/class interval between the QP_ROI+D2} of three major types block respectively for D1, D2}13.In this embodiment, QP_ROI is the quantization parameter of macro zone block interested, and QP_ROI+D1 is the quantization parameter that profile extends macro zone block, and QP_ROI+D2 is the quantization parameter of background macro zone block.
Quantization parameter initialization module 14, in order to size and the picture update rate according to target bit rate, video pictures, determine macro zone block interested, profile to extend quantization parameter { QP_ROI, QP_ROI+D1, the initial value 15 of QP_ROI+D2} of macro zone block and background macro zone block.For example, quantization parameter initialization module 14 is in order to size and picture update rate according to target bit rate, video pictures, determine the initial value 15 of the quantization parameter QP_ROI of macro zone block interested, cooperate again and quantize the Parameter Decision Making module 12 quantization parameter class interval { D1 that determine, D2}13 draws quantization parameter { QP_ROI+D1, the initial value of QP_ROI+D2} that profile extends macro zone block and background macro zone block.The detailed action of quantization parameter initialization module 14 will describe in detail in follow-up cooperation Fig. 4 again.
Limited variable bit rate control calculation module 16 is in order to { QP_ROI+D1, the macro zone block interested of the video pictures of QP_ROI+D2}, profile extend macro zone block and the background macro zone block is encoded according to quantization parameter.In order further to control coding bit rate, parameter update module 18 is adjusted at quantization parameter { QP_ROI, QP_ROI+D1 on the picture frame layer in order to according to present residual bit rate (Residual Bits) 17, QP_ROI+D2} avoids encoding buffer overflow or zero load.The detailed action of parameter update module 18 will describe in detail in follow-up cooperation Fig. 3 again.
Odd number fixes/the block decision-making mechanism of a plurality of mobile ROI
Position (PosX at a known tracking object/area-of-interest, PosY), after wide (Wd) and high (Ht), coding priority according to the macro zone block of this object, its macro zone block is distinguished into three major types, comprises that the 1st class mentioned above (ROI itself), the 2nd class (ROI extends partly in the zone) are (background) with the 3rd class.
In the block decision-making mechanism of present embodiment, can further consider following three problems:
(1) position and the size information of profile of given at least one object/area-of-interest, how efficient, determine the coding priority (distortion level) of each macro zone block rapidly;
(2) consider its situation that may overlap to each other of several tracked object/area-of-interests; And
(3) in the algorithm of various object tracking/area-of-interests, when external light source changes, may cause the situation of following the trail of failure.For example, the tracking frame of contour of object temporarily disappears, and may have influence on the result of block decision-making.
Fig. 2 A shows the embodiment flow chart of steps that produces the ROI mapping table with different coding priority.
At first, the coding priority (distortion level) of setting all macro zone blocks is 3 (step S201), and just the quality of macro zone block is the quality of background image, and its distortion level is for the highest.Then judge and in video pictures, whether detect zone/profile (blob) (step S202) that any object/area-of-interest comprises.If when in video pictures, detecting less than any object/area-of-interest, utilize counter to judge and whether in video pictures, do not detect the continuous time of any object/area-of-interest less than a critical value (for example, 10 milliseconds) (step S203).If less than this critical value, represent in this video pictures to be not have object/area-of-interest, and only be maybe not track object/area-of-interest as yet, then use the previous ROI mapping table that produces, this counter (is for example increased by the displacement time, 1 millisecond) (step S204), get back to step S21 then.If not less than this critical value, be illustrated in and do not comprise any object/area-of-interest in this video pictures really, the coding priority (distortion level) of then setting all macro zone blocks in this zone is 3, and this counter is reset to 0 (step S205), gets back to step S201 then.
If when detecting the zone of object/area-of-interest/profile, then in position and the size (wide w and high h) of the zone/profile of given object/area-of-interest, find out that all belong to the macro zone block ((x of this object/area-of-interest in this video pictures, y)～(x+w, y+h) between) index value, and set this zone/profile/in belong to macro zone block interested coding priority (distortion level) be 1 (step S206).Next, (for example use matrix, location matrix (Position Matrix)) confirms the priority (distortion level) (step S207) of other macro zone block (i.e. Zhou Wei macro zone block) in this zone/profile, wherein the central element of this matrix (Cell) representative belongs to the macro zone block of this object/area-of-interest, the profile that 8 elements on every side then belong to this object/area-of-interest extends (Contour Extension) zone, and namely profile extends macro zone block (other macro zone block).
With reference to figure 2B, the macro zone block corresponding index value in its display video picture.With reference to figure 2C, the index value in this 3x3 matrix is corresponding to the concordance list of Fig. 2 B, and wherein the macro zone block at i place is namely represented ROI, and the priority of just encoding (distortion level) is 1 macro zone block, and 8 macro zone blocks around the i are namely represented the profile elongated area of object.
Next, whether the coding priority (distortion level) of the macro zone block in the inspection profile elongated area was denoted as for 1 (representative belongs to the central element of other object area/area-of-interest).(step S208).If the coding priority (distortion level) of macro zone block is not denoted as 1 as yet, that then represents this macro zone block belongs to the profile elongated area really, so this macro zone block is denoted as 2 (step S209).Judge whether the All Ranges of this object/area-of-interest has finished detection (step S210), if still have the zone to detect as yet in this object/area-of-interest, repeating step S206～209 then are up to all zones that check out this object.If the All Ranges of object/area-of-interest has all been finished detection, judge then then whether all objects/the area-of-interest in this video pictures has all finished detection (step S211).If still have in this video pictures object/area-of-interest to need to detect, repeating step S201～210 then, all objects/area-of-interest in checking out this video pictures.If all objects/area-of-interest in this video pictures has all been finished detection, then produce new ROI mapping table (step S212).Be noted that the priority (distortion level) of encoding this moment is set to all macro zone blocks of 3, namely belongs to the background macro zone block.
H.264 limited variable bit rate control algorithm based on ROI
After obtaining the coding priority (distortion level) of each macro zone block, directly give ROI, ROI zone and extend part and these three quantization parameters that the zone is different of background, its set is { Q1, Q2, Q3}, wherein Q1＜Q2＜Q3.This is because under bit rate-distortion-quantification (Bitrate-Distortion-Quantification) model, and generally speaking, distortion level is higher, and employed quantization parameter also can be healed greatly.Though relevant object tracking algorithm is a lot, be without loss of generality, as long as can access regional location and the size of object, just can be suitable for the limited variable bit rate control algorithm of the embodiment of the invention.This limited variable bit rate control algorithm more can provide following operation:
(1) selects quantization parameter { Q1, Q2, Q3}.For example, Q1 is QP_ROI, and Q2 is QP_ROI+D1, and Q3 is QP_ROI+D2.This selection operation comprises following two kinds of possible situations: (i) gap between Q1, Q2 and Q3 is little, though can avoid the PSNR value difference of adjacent macro zone block apart from producing block effect too greatly, but do not meet the expectation of ROI video coding system, and may produce higher coding bit rate.(ii) the gap between Q1, Q2 and Q3 is big, can improve the PSNR value of area-of-interest, and the characteristic of the high frequency coefficient by giving up more background area is saved bit rate, but shortcoming is the puzzlement in the time of may causing moltion estimation (Motion Estimation).So research emphasis of the present invention is to obtain one group of suitable quantization parameter { Q1, Q2, Q3}.
(2) dynamically adjust the quantification parameter group.Fig. 3 shows that the dynamic adjustment of an embodiment quantizes the flow chart of steps of parameter group.In this algorithm, because the quantization parameter group is all fixed at the whole video picture, different when the complexity of different pictures, the bit rate after obtaining compressing also will be different, and this belongs to the research category of " variable bit-rate compression coding " (Variable Bitrate Coding).If want to allow bit rate behind the coding more near a fixing target bit rate (Target Budget Bit, or title budget bit rate), avoids encoding buffer overflow (overflow) or unloaded (underflow), should dynamically adjust quantization parameter { Q1, Q2, the value of Q3}.Below explanation parameter update module 18 is dynamically adjusted the implementing procedure of quantization parameter.
With reference to figure 3, at first judge whether detect any object/area-of-interest (step S301) in the video pictures.If do not detect any object/area-of-interest, then { 40,40,40} (the default parameter group that quantizes) is assigned to the macro zone block (step S302) in the video pictures with one group of uniform quantization parameter.If detect object/area-of-interest, then calculate the utilization rate (Buffer Fullness) (step S303) of buffering area, judge that then whether utilization rate is greater than preset value (for example, 0) (step S304).For example, the utilization rate of the buffering area that calculates is the utilization rate of the buffering area of limited variable bit rate control calculation module 16, but is not limited to this.
If utilization rate is not more than this preset value, then { 35,40,50} is assigned to the macro zone block (step S305) in the video pictures with quantization parameter.If utilization rate is greater than this preset value, then adjust the QP_ROI value according to the coding bit rate (Encoded) of limited variable bit rate control calculation module 16 and the relation between target bit rate (Target), with seasonal D=(Encoded-Target)/Target (step S306).For example, the D value can be considered aforesaid residual bit rate 17, but is not limited to this.
Judge that whether D is less than-0.5 (step S307).If D less than-0.5, then makes the quantization parameter value of block interested subtract 4 (QP
RoI=QP
RoI-4) (step S308).If D is not less than-0.5, judge that then D is whether between-0.5～-0.2 (0.5＜D＜-0.2) (step S309).If D between-0.5～-0.2, then makes the quantization parameter value of block interested subtract 2 (QP
RoI=QP
RoI-2) (step S310).If D is between-0.5～-0.2, judge that then D is whether between 0.2～0.5 (0.2＜D＜0.5) (step S311).
If D between 0.2～0.5, then makes the quantization parameter value of area-of-interest add 2 (QP
RoI=QP
RoI+ 2) (step S312).If D is between 0.2～0.5, judge that then D is whether greater than 0.5 (D＞0.5) (step S313).If D greater than 0.5, then makes the quantization parameter value of area-of-interest add 4 (QP
RoI=QP
RoI+ 4) (step S314).If D is not more than 0.5, then do not change the quantization parameter value (step S315) of block interested.In other words, in the present embodiment, when the D value when negative, 18 of parameter update module can be heightened QP
Ro, and when the D value be timing, 18 of parameter update module can be turned down QP
Ro
(3) initial value of decision quantization parameter.Fig. 4 shows the flow chart of steps of initial value of the decision quantization parameter of the embodiment of the invention.{ QP_ROI, QP_ROI+D1, QP_ROI+D2} can avoid the process of the dynamic adjustment quantization parameter of redundancy, allow the bit rate after encoding can approach target bit rate soon to select one group of suitable quantization parameter at the beginning.Below explanation quantization parameter initialization module 14 determines quantization parameter { QP_ROI, QP_ROI+D1, the implementing procedure of the initial value of QP_ROI+D2}.
With reference to figure 4, definition bpp=T/ (F*h*w), wherein bpp represents the bit number that each pixel is assigned in the video pictures, F represents the picture update rate of video pictures, and the product of h and w represents the size (step S401) of this video pictures, then judges the resolution (step S402) of video pictures.If video pictures utilizes QCIF (Quarter Common Intermediate Format) picture format to handle and can obtain L1=0.1, L2=0.3 and L3=0.6, if video pictures utilizes CIF (Common Intermediate Format) picture format to handle and can obtain L1=0.2, L2=0.6 and L3=1.2, and if video pictures utilizes other picture format to handle and can obtain L1=0.6, L2=1.2 and L3=2.4.Therefore, can obtain one group of resolution parameter { L1, L2, L3}.
Judge that whether bpp is less than L1 (bpp＜L1) (step S403).If bpp less than L1, then can obtain one group of quantization parameter { 35,40,50} (step S404).If bpp is not less than L1, judge that then whether bpp is less than L2 (bpp＜L2) (step S405).If bpp less than L2, then can obtain one group of quantization parameter { 25,30,40} (step S406).If bpp is not less than L2, judge that then whether bpp is less than L3 (bpp＜L2) (step S407).If bpp less than L3, then can obtain one group of quantization parameter { 20,25,35} (step S408).If bpp is not less than L3, then obtain one group of quantization parameter { 10,15,25} (step S409).In this embodiment, initial value and the bpp of the quantization parameter that determines of quantization parameter initialization module 14 have inverse relation.For example, when the bpp value is bigger, quantization parameter value is littler, otherwise quantization parameter value is bigger.
Just select the operation of one group of quantization parameter, the present invention is by the mode of off-line (Off-line) coding, with different quantization parameters different standard picture sequences is compressed, depicted at last in the relation between distortion level and quantization parameter (Distortion-QP) under the different complexities.In the human vision model, it is generally acknowledged image quality greater than more than 30 decibels (dB), be the minimum acceptable picture quality of human eye, greater than namely being expressed as splendid image quality more than the 40dB, non-well-trained eyes are difficult to differentiate the difference with former figure.Based on this idea, the background in the expection monitoring image partly should reach 30dB at least, and area-of-interest (ROI) then should reach more than the 40dB at least, is without loss of generality, and ROI profile elongated area should be between 34-36dB.
Be example with Fig. 5, in the film of low complex degree, this moment, corresponding quantization parameter value combination should be { 25,30,40}.After through experiment repeatedly, under different bit rates, each regional quantization parameter gap should be that { QP_ROI, QP_ROI+D1, QP_ROI+D2}, the most preferred embodiment of wherein 0＜D1＜D2, and D1 and D2 are 5 and 15 (can be 4～6 and 14～16 in the enforcement).So namely meet in encoding and decoding H.264, the quantization parameter difference of maximal contiguous block is the restriction of [25,26].It should be noted that because quantization parameter H.264 is limited between [1,51], so quantization parameter value should drop between [1,36].
Just dynamically adjust the operation that quantizes parameter group, the present invention constantly adjusts quantization parameter value according to the relation between coding bit rate (Encoded) and target bit rate (Target).Make D=(Encoded-Target)/Target, when the D value is negative, represent the coding bit rate of last picture greater than target bit rate, then heighten quantization parameter value, otherwise then reduce quantization parameter value.In addition, must consider following two kinds of special circumstances:
(1) when being detected without any object/area-of-interest, then { 40,40,40} directly is assigned to all macro zone blocks with one group of uniform quantization parameter; And
(2) when the buffering area generation overflow of coding side (for example, limited variable bit rate control calculation module 16), then { 35,40,50} directly is assigned to all macro zone blocks with the quantization parameter of one group of maximum.
Just determine the operation of the initial value of quantization parameter group, the present invention is according to target bit rate, picture size and frame updating speed, calculate the suitable initial quantization parameter of a combination, so can reduce the process of redundant dynamic adjustment quantization parameter, to allow the bit rate after encoding to approach target bit rate soon.Basic conception is the bit number (bpp) of judging that each pixel is assigned to, when the bpp value is bigger, represents exercisable budget bit the more, therefore can reduce quantization parameter value, otherwise then improve quantization parameter value.
Can observe from the experimental data of Fig. 6 and to obtain drawing a conclusion:
(1) gap between resulting each the regional PSNR value of the algorithm of JM13.2, Sun and Sivanantharasa and little, in some cases, the PSNR value in ROI zone even be lower than the PSNR value of non-area-of-interest.And in method proposed by the invention, the quality of area-of-interest is then consistent preferable than the quality of non-area-of-interest;
(2) " utilizing limited variable bit rate control " of proposing can effectively control output bit rate in specific budget bitrate range, error amount is [3.6,3.8] between, and JM ([4.1,3.88]), Sun ([2.3,3.3]) not far from one another with the result of Sivanantharasa ([2.5,3.2]) generation;
(3) computing time on Bit-Rate Control Algorithm relatively, in method proposed by the invention, the required time of individual picture is about 0.4 millisecond (ms), and all required than JM (6-9ms), Sun (9-12ms) and Sivan (6-11ms) time is deficient a lot of; And
(4) with reference to figure 7A～7E, Fig. 7 A is the original image in corridor, Fig. 7 B is the corridor image after encoding through ROI, the zone that its center gets up is non-(Non-ROI) interested zone, Fig. 7 C is foreman's original image, and Fig. 7 D is the foreman's image after encoding through ROI, and the zone that its center gets up is non-(Non-ROI) interested zone, and Fig. 7 E is the foreman's image after encoding through ROI, and the zone that its center gets up is area-of-interest.As from the foregoing, the method for coming proposed by the invention the resulting image quality of area-of-interest (people and foreman (Foreman) in mobile) all more non-area-of-interest be good.
Also disclosed a kind of computer recordable media according to an embodiment, in order to store computer program, above-mentioned computer program comprises a plurality of procedure code fragments, it is in order to be loaded in the computer system and to make above-mentioned computer system carry out a kind of encoding region of interest method of utilizing limited variable bit rate control, comprise the position of profile in video pictures according to area-of-interest, with a plurality of macro zone blocks of this video pictures be divided at least one macro zone block interested, at least one profile extends macro zone block and at least one background macro zone block; And give macro zone block interested, profile extension macro zone block and the different coding priority of background macro zone block, produce a ROI mapping table thus; According to size and the picture update rate of target bit rate, video pictures, determine the macro zone block interested of video pictures, the initial value that profile extends one group of quantization parameter of macro zone block and background macro zone block; According to this group quantization parameter macro zone block interested, profile extension macro zone block and the background macro zone block of video pictures are encoded; And this group quantization parameter of dynamically adjusting macro zone block interested, profile extension macro zone block and the background macro zone block of video pictures.The detailed process of carrying out the step of the method is as described before, is not repeated at this.
Said method can exist by procedure code.When procedure code was written into and carries out by machine, machine became in order to carry out device of the present invention or computer program.
According to the method for the embodiment technology that discloses, or specific kenel or its part, can exist with the kenel of procedure code.Procedure code can be contained in tangible media, get (as embodied on computer readable) storage medium as floppy disk, CD sheet, hard disk or any other machine readable, wherein, when procedure code was written into and carries out by machine such as computer, this machine became the device of implementing in order to participate in.Procedure code also can pass through some transmission mediums, transmit as electric wire or cable, optical fiber or any transmission kenel, wherein, when procedure code by machine such as computer receive, when being written into and carrying out, this machine becomes in order to participate in device of the present invention.When the general service processing unit is implemented, procedure code provides a class of operation to be similar to the unique apparatus of using particular logic circuit in conjunction with processing unit.
Though disclose as above with some embodiment; yet it is not in order to limit embodiment; anyly have the knack of this skill person; in the spirit and scope that do not break away from disclosed embodiment; when being used for a variety of modifications and variations, therefore the protection range of invention is as the criterion when looking the accompanying Claim person of defining.
Claims (18)
1. region of interest ROI coded system of utilizing limited variable bit rate control comprises:
The block decision-making module, in order to the position of profile in video pictures according to area-of-interest, a plurality of macro zone blocks of this video pictures are divided at least one macro zone block interested, at least one profile extension macro zone block and at least one background macro zone block, and give this macro zone block interested, this profile extension macro zone block and the different coding priority of this background macro zone block, produce a ROI mapping table thus;
The quantization parameter initialization module in order to size and the picture update rate according to target bit rate, this video pictures, determines this macro zone block interested of this video pictures, the initial value that this profile extends one group of quantization parameter of macro zone block and this background macro zone block;
Limited variable bit rate control calculation module is in order to encode to this macro zone block interested, this profile extension macro zone block and this background macro zone block of this video pictures according to this group quantization parameter;
The parameter update module is in order to this macro zone block interested of dynamically adjusting this video pictures, this group quantization parameter that this profile extends macro zone block and this background macro zone block; And
The quantization parameter decision-making module, in order to produce first order distance and second level distance, wherein this group quantization parameter comprises first quantization parameter, second quantization parameter and the 3rd quantization parameter that corresponds respectively to this macro zone block interested, this profile extension macro zone block and this background macro zone block, this second quantization parameter be this first quantization parameter and this first order distance and, and the 3rd quantization parameter be this first quantization parameter and this second level distance with;
Wherein this parameter update module is dynamically adjusted this group quantization parameter according to the relation between the coding bit rate of target bit rate and present this video pictures.
2. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 1, wherein this quantization parameter initialization module system is according to the resolution of this video pictures, obtain one group of resolution parameter, and according to size and this picture update rate of this target bit rate, this video pictures, try to achieve a bit number, and determine this initial value of this group quantization parameter by the magnitude relationship of this bit number and this group resolution parameter.
3. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 2, wherein this bit number is the bit number that each pixel is assigned in this video pictures, and this bit number is tried to achieve by equation bpp=T/ (F*h*w), wherein bpp represents that this bit number, T represent this target bit rate, F represents this picture update rate, and the product of h and w represents the size of this video pictures.
4. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 2, wherein this quantization parameter initialization module determines this initial value of this group quantization parameter and this bit number of trying to achieve has inverse relation.
5. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 1, wherein this parameter update module is dynamically adjusted this group quantization parameter by the operation result of equation D=(E-T)/T, E represents this coding bit rate, T represents this target bit rate, and D represents this operation result.
6. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 5, wherein when this operation result is negative value, this parameter update module is heightened this first quantization parameter, and when this operation result be on the occasion of the time, this parameter update module is turned down this first quantization parameter.
7. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 5, wherein when this operation result during greater than critical value, this parameter update module then specifies one group of default quantization parameter to give this macro zone block interested, this profile extension macro zone block and this background macro zone block.
8. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 1, wherein when not detecting this area-of-interest, this parameter update module then specifies one group of default quantization parameter to give this macro zone block interested, this profile extends macro zone block and this background macro zone block, and the default quantization parameter of this group comprises and corresponds respectively to this macro zone block interested, this profile extends first quantization parameter of macro zone block and this background macro zone block, second quantization parameter and the 3rd quantization parameter, this first quantization parameter, this second quantization parameter equates with the 3rd quantization parameter three.
9. the encoding region of interest system that utilizes limited variable bit rate control as claimed in claim 1, wherein to set the coding priority of all macro zone blocks in this video pictures be the lowest priority with identical coding priority to this block decision-making module, produce the 2nd ROI mapping table thus, again according to position and the size of this area-of-interest, find out this macro zone block interested and this profile and extend the manipulative indexing value of macro zone block in the 2nd ROI mapping table of this video pictures, and will this block interested and this profile coding priority of extending macro zone block change respectively and be set at highest priority and medium priority, in order to produce a ROI mapping table.
10. an encoding region of interest method of utilizing limited variable bit rate control comprises the following steps:
According to the position of profile in video pictures of area-of-interest, with a plurality of macro zone blocks of this video pictures be divided at least one macro zone block interested, at least one profile extends macro zone block and at least one background macro zone block;
Give this macro zone block interested, this profile extension macro zone block and the different coding priority of this background macro zone block, produce a ROI mapping table thus;
According to size and the picture update rate of target bit rate, this video pictures, determine this macro zone block interested of this video pictures, the initial value that this profile extends one group of quantization parameter of macro zone block and this background macro zone block;
According to this group quantization parameter this macro zone block interested, this profile extension macro zone block and this background macro zone block of this video pictures are encoded;
Dynamically adjust this macro zone block interested of this video pictures, this group quantization parameter that this profile extends macro zone block and this background macro zone block; And
Produce first order distance and second level distance, wherein this group quantization parameter comprises first quantization parameter, second quantization parameter and the 3rd quantization parameter that corresponds respectively to this macro zone block interested, this profile extension macro zone block and this background macro zone block, this second quantization parameter be this first quantization parameter and this first order distance and, and the 3rd quantization parameter be this first quantization parameter and this second level distance with;
Wherein this group quantization parameter is dynamically adjusted according to the relation between the coding bit rate of target bit rate and present this video pictures.
11. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 10 wherein determines to comprise the step of the initial value of this group quantization parameter:
According to the resolution of this video pictures, obtain one group of resolution parameter;
According to size and this picture update rate of this target bit rate, this video pictures, try to achieve a bit number; And
Determine this initial value of this group quantization parameter by the magnitude relationship of this bit number and this group resolution parameter.
12. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 11, wherein this bit number is the bit number that each pixel is assigned in this video pictures, and this bit number is tried to achieve by equation bpp=T/ (F*h*w), wherein bpp represents that this bit number, T represent this target bit rate, F represents this picture update rate, and the product of h and w represents the size of this video pictures.
13. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 11, this initial value of this group quantization parameter that wherein determines has inverse relation with this bit number of trying to achieve.
14. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 10, wherein this group quantization parameter is dynamically adjusted by the operation result of equation D=(E-T)/T, E represents this coding bit rate, and T represents this target bit rate, and D represents this operation result.
15. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 14 wherein when this operation result is negative value, is heightened this first quantization parameter, and when this operation result be on the occasion of the time, turn down this first quantization parameter.
16. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 14, wherein when this operation result during greater than critical value, specify one group of default quantization parameter to give this macro zone block interested, this profile extension macro zone block and this background macro zone block.
17. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 10, wherein when not detecting this area-of-interest, specify one group of default quantization parameter to give this macro zone block interested, this profile extends macro zone block and this background macro zone block, and the default quantization parameter of this group comprises and corresponds respectively to this macro zone block interested, this profile extends first quantization parameter of macro zone block and this background macro zone block, second quantization parameter and the 3rd quantization parameter, this first quantization parameter, this second quantization parameter equates with the 3rd quantization parameter three.
18. the encoding region of interest method of utilizing limited variable bit rate control as claimed in claim 10, wherein give this macro zone block interested, this profile extension macro zone block and the different coding priority of this background macro zone block, the step that produces a ROI mapping table thus comprises:
The coding priority of setting all macro zone blocks in this video pictures is the lowest priority with identical coding priority, produces the 2nd ROI mapping table thus;
According to position and the size of this area-of-interest, find out this macro zone block interested and this profile and extend the manipulative indexing value of macro zone block in the 2nd ROI mapping table of this video pictures; And
The coding priority that this block interested and this profile are extended macro zone block changes respectively and is set at highest priority and medium priority, in order to produce a ROI mapping table.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910148175 CN101931800B (en) | 2009-06-24 | 2009-06-24 | Method and system for encoding region of interest by limited variable bit rate control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910148175 CN101931800B (en) | 2009-06-24 | 2009-06-24 | Method and system for encoding region of interest by limited variable bit rate control |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101931800A CN101931800A (en) | 2010-12-29 |
| CN101931800B true CN101931800B (en) | 2013-08-14 |
Family
ID=43370684
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200910148175 Active CN101931800B (en) | 2009-06-24 | 2009-06-24 | Method and system for encoding region of interest by limited variable bit rate control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101931800B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103502961B (en) * | 2011-07-20 | 2016-05-11 | 李国君 | The essential data transmission rate quantization method of algorithm |
| GB201312382D0 (en) * | 2013-07-10 | 2013-08-21 | Microsoft Corp | Region-of-interest aware video coding |
| CN104486626B (en) * | 2014-12-30 | 2019-02-15 | 中星技术股份有限公司 | A kind of method for video coding |
| CN111491167B (en) * | 2019-10-28 | 2022-08-26 | 华为技术有限公司 | Image encoding method, transcoding method, device, equipment and storage medium |
| CN113301336A (en) * | 2020-02-21 | 2021-08-24 | 华为技术有限公司 | Video coding method, device, equipment and medium |
| CN114531615B (en) * | 2020-11-03 | 2023-10-27 | 腾讯科技（深圳）有限公司 | Video data processing method, device, computer equipment and storage medium |
| CN114615496A (en) * | 2022-01-28 | 2022-06-10 | 阿里巴巴（中国）有限公司 | Method, device, storage medium and system for obtaining quantization parameter |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1420690A (en) * | 2001-11-17 | 2003-05-28 | Lg电子株式会社 | Method and system for control of bit tate based on object |
| US6850565B2 (en) * | 1998-12-18 | 2005-02-01 | Intel Corporation | Real time bit rate control system |
| CN101188755A (en) * | 2007-12-14 | 2008-05-28 | 宁波中科集成电路设计中心有限公司 | A method for VBR code rate control in AVX decoding of real time video signals |
| CN101309422A (en) * | 2008-06-23 | 2008-11-19 | 北京工业大学 | Macroblock level quantized parameter process method and apparatus |
-
2009
- 2009-06-24 CN CN 200910148175 patent/CN101931800B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6850565B2 (en) * | 1998-12-18 | 2005-02-01 | Intel Corporation | Real time bit rate control system |
| CN1420690A (en) * | 2001-11-17 | 2003-05-28 | Lg电子株式会社 | Method and system for control of bit tate based on object |
| CN101188755A (en) * | 2007-12-14 | 2008-05-28 | 宁波中科集成电路设计中心有限公司 | A method for VBR code rate control in AVX decoding of real time video signals |
| CN101309422A (en) * | 2008-06-23 | 2008-11-19 | 北京工业大学 | Macroblock level quantized parameter process method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101931800A (en) | 2010-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101931800B (en) | Method and system for encoding region of interest by limited variable bit rate control | |
| US5473377A (en) | Method for quantizing intra-block DC transform coefficients using the human visual characteristics | |
| WO2021244341A1 (en) | Picture coding method and apparatus, electronic device and computer readable storage medium | |
| CN114071189B (en) | Video processing device and video streaming processing method | |
| US7162096B1 (en) | System and method for dynamic perceptual coding of macroblocks in a video frame | |
| KR100307618B1 (en) | Device and method for encoding image | |
| EP2809073A1 (en) | Bit-Rate control for video coding using object-of-interest data | |
| JPH10257489A (en) | Device and method for adjusting amount of bits to be generated for encoding image | |
| JPH09200760A (en) | Method and circuit for deciding quantizing interval for video encoder | |
| JP2017532858A (en) | Perceptual optimization for model-based video coding | |
| JPH09214963A (en) | Method for coding image signal and encoder | |
| CN106664409B (en) | Method, system, device and medium for golden frame selection in video coding | |
| US20200267396A1 (en) | Human visual system adaptive video coding | |
| US11197021B2 (en) | Coding resolution control method and terminal | |
| CN107211145A (en) | The almost video recompression of virtually lossless | |
| KR20150054554A (en) | Apparatus and method for processing image | |
| CN114466189A (en) | Code rate control method, electronic device and storage medium | |
| EP4333433A1 (en) | Video coding method and apparatus, and electronic device | |
| EP1227684A2 (en) | Encoding of video signals | |
| US10805519B2 (en) | Perception-based image processing apparatus and associated method | |
| JP2018113671A (en) | Video encoding method and encoder | |
| US20230108722A1 (en) | Allocating bit rate between video streams using machine learning | |
| CN112584143B (en) | Video coding method, device and system and computer readable storage medium | |
| CN101977317A (en) | Intra-frame prediction method and device | |
| CN111757112A (en) | HEVC (high efficiency video coding) perception code rate control method based on just noticeable distortion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |