CN1139257C - Initializing method according to feature block of detected shape in block matching method to which genetic algorithm is applied - Google Patents
Initializing method according to feature block of detected shape in block matching method to which genetic algorithm is applied Download PDFInfo
- Publication number
- CN1139257C CN1139257C CNB981184901A CN98118490A CN1139257C CN 1139257 C CN1139257 C CN 1139257C CN B981184901 A CNB981184901 A CN B981184901A CN 98118490 A CN98118490 A CN 98118490A CN 1139257 C CN1139257 C CN 1139257C
- Authority
- CN
- China
- Prior art keywords
- piece
- shape
- motion vector
- genetic algorithm
- image
- 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.)
- Expired - Fee Related
Links
Images
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
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
-
- 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
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/136—Incoming video signal characteristics or properties
- H04N19/137—Motion inside a coding unit, e.g. average field, frame or block difference
-
- 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
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/136—Incoming video signal characteristics or properties
- H04N19/14—Coding unit complexity, e.g. amount of activity or edge presence estimation
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Analysis (AREA)
Abstract
A method for initiation according to characteristic blocks of an object extracting in a gene algorithm is provided to calculate an optimum motion value considering the characteristics of image blocks over a gene algorithm, thereby reducing a calculation amount and obtaining a precise motion value. The method includes the steps of: (a) dividing an image into predetermined blocks after extracting a predetermined object from a reference frame of the image, (b) dividing per characteristics by relating the divided image blocks with the object by using the gene algorithm, and (c) initiating a motion value by applying a motion value proper to the blocks divided per characteristics by the gene algorithm.
Description
Technical field
The present invention relates to a kind of method for compressing image, more particularly, relate to the initial method in the block matching method that adopts genetic algorithm, be used for basis and divide characteristic block, and carry out different initialization computings according to each characteristic block that is divided from the relation of the detected reservation shape of image reference frame.
Background technology
In Image Compression, there are two types.The one, intraframe coding promptly utilizes the spatial redundancies compressed information that exists in image; The 2nd, interframe encode is promptly utilized the temporal redundancy compressed information between the time sequencing epigraph.
Particularly, when transmission during one width of cloth moving image, the motion compensation encoding of using interframe encode typically is used for reducing volume of transmitted data.
Motion compensation encoding relate to a moving target of transmission expression shift value motion vector and according to the predicated error of the motion compensation of adjacent image frame in time.
Block matching algorithm (BMA) is used in the real time kinematics image delivering system that relates to hardware, as the momental method of measurement target.
All direction search method, three step searching methods are used the searching method one by one of high speed algorithm and the BMA method of layering and are used in the block matching algorithm.In all direction search method, data flow is regular, like this, compares with other method, is easy on the hardware realize, and has good prediction characteristic.But all direction search method has a shortcoming, and it needs a large amount of calculating.
Genetic algorithm is based on the searching algorithm of natural selection and genetic structure, wherein by manual system, has realized natural self adaptation phenomenon.
In order to realize genetic algorithm, optimised theme must pass through one group of parameter expression,, contains effective hyte of hereditary information that is.This group parameter is important in genetic algorithm.Hyte must be confirmed as with natural target similar, is enough to express the information of hope like this.
When the hereditary information of determining was transferred to the next generation, hereditary feature had been changed.According to suitability (fitness), promptly be used to change the transfer function of hereditary feature, determine operator to be selected.Genetic algorithm is the method that a kind of optimization contains the colony of different hereditary information, and it utilizes inherent hereditary information rather than the direct phenomenon of being used by other searching method.
Genetic algorithm is a kind of stochastic search methods.Yet, do not resemble other stochastic search methods, in genetic algorithm,, also there is natural hereditary feature except simple stochastic behaviour.Genetic algorithm is expressed as the set of an expression hereditary information, for example evolution and the change of hereditary feature (DNA or chromosome) and hereditary feature.Therefore, the genetic search algorithm not by random function determine and be meant the natural phenomena of set of expression hereditary information.For this point, the function of hereditary regeneration (reproduction), intersection (crossover), conversion (mutation) has been determined and biological feature has been employed, because hereditary feature is evolved from a generation to another generation, change causes entropy to reduce.As the above, genetic algorithm is a random function, and the influence of local characteristics is reduced, and the whole characteristic as natural phenomena that is determined by essential behavior is had efficient search.
In genetic algorithm, according to suitability, the hereditary feature of colony has changed from a generation to another generation.According to suitability, genetic algorithm is only selected a suitable hereditary feature.For such process, the genetic operator that changes hereditary feature is necessary.Constitute by regeneration, intersection, three operators of conversion at do very well result's simple generic algorithm of actual application problem.
Regeneration is to duplicate hereditary feature to follow-on operation according to the target function that their quilt is called the suitability function.Target function can be an anticipation function table (barometer), and it shows optimum state according to applied problem.Duplicate hereditary feature according to suitability, mean that the hereditary feature that has high suitability probably remains to the next generation.
After the regeneration, intersect with the execution of two steps.In the new hereditary feature that produces of coupling at random with after selecting the optional position of any hereditary feature, hereditary feature exchanges to last position with corresponding hereditary feature.
Then, the conversion that produces by given probability is arranged, this is from 0 to 1 or from 1 to the 0 change hereditary feature operator of position arbitrarily.In genetic algorithm, conversion helps out, only by the regeneration and the result that just can obtain optimum without conversion of intersecting.
Summary of the invention
The initial method that the purpose of this invention is to provide a kind of foundation characteristic block of detected shape in the block matching method of having used genetic algorithm.In this way by according to dividing macro block, and carry out suitable initialization, can reduce the amount of calculation of motion vector according to the piece that is divided with the relation property that in an image reference frame, detects shape.
Therefore,, provide a kind of initial method in the block matching method of using genetic algorithm, having comprised: (a) from the reference frame of piece image, detected a reservation shape, and this image segmentation is become predetermined piece in order to reach first target; (b), divide by each piece of segmented image for the piece irrelevant, overlapping piece and the piece in detecting shape are arranged with the outline line that detects shape with detecting shape according to the relationship characteristic of each piece and this shape; (c) in each piece of dividing according to described feature, come the initialization motion vector by using suitable motion vector.Wherein, at step (c), the piece irrelevant with detecting shape adopted the dynamic population control method, the piece of the outline line of the overlapping shape that detects is adopted the motion vector of the adjacent block with weighted value, the piece in detecting shape is adopted the motion vector of adjacent block.
Description of drawings
By being described in detail with reference to the attached drawings a preferred embodiment, above-mentioned target of the present invention and advantage will be more clear, wherein:
Fig. 1 has described a block matching method of using genetic algorithm;
Fig. 2 has described according to initial method of the present invention;
Fig. 3 has described the initial method of having used genetic algorithm according to all kinds of Fig. 2; With
Fig. 4 has represented a search block.
Embodiment
Fig. 1 has described the block matching method of using genetic algorithm, and it is embodiment of the correlation application genetic algorithm of a motion vector that utilizes adjacent block.
At first, determine the motion vector (S10) of reference block.When not considering characteristics of image, the motion vector that produces at random is defined as initial value.Yet, having in the moving image of correlation with adjacent block, the application correlation is defined as initial value with the mean value of the motion vector of adjacent block.When mean value is confirmed as initial value,, can reduce amount of calculation by iterations that reduces genetic algorithm and the piece number that will search for.
Adopt mean square deviation (MSE) to come based on initialized motion vector computation suitability (S12).That is, along with error diminishes the suitability change greatly.
A pair of colony with the Minimum Mean Square Error MSE that is used for calculating suitability, i.e. optimum population selected (S14).
By the gene of combination or the selected colony of exchange step S14, carry out intersection process (S16).Here, cross processing is to operate by a mask that is called even intersection (UX) to carry out.That is, produce the mask (mask) of a chromosome equal length of organizing with the father.These value is determined at random.When one value was 1, the value of filial generation equaled the value of one of chromosome of two colonies.When one value was 0, the value of filial generation equaled another chromosomal value.The value of other filial generation is determined by the counter-rotating mask.
Because lose owing to the reason of adding up has a lot of feature uncles' groups, so, carry out conversion process, the chromosomal value of reversing (step S18) in order to compensate this point.That is, can find out the total optimization value, and be not only a local optimum.
At last, repeating step S10 is to S18, and up to obtaining a predetermined optimal value, producing the strongest chromosome and this chromosomal value is exactly total optimization value (S20 step).
Fig. 2 has described according to initial method of the present invention, shows the portrait of a simplification, i.e. detected target from the reference frame of image.
Here, from reference frame, be divided into following: do not belong to the piece of detected shape, resemble piece A1-B1 (category-A piece) by the macro block of segmentation; The piece that comprises detected shaped wheel profile resembles piece A4-B2 (category-B piece); Be comprised in the piece in the detected shape fully, resemble piece A5-B3 (C class piece).
In the category-A piece, its motion vector is to be difficult to prediction, determines initial value with reference to the motion vector of adjacent piece.Because under many circumstances, a category-B piece has a motion vector at least, and it is not effective utilizing adjacent piece predicted motion.C class piece has identical motion vector to have a high probability.In some part a small different motion is arranged.Particularly, people's eyes and face are good examples.The C type blocks is suitable for using genetic algorithm.
Fig. 3 has described according to the type application of Fig. 2 the initial method of genetic algorithm.Two father's chromosomes of formation and two chromosomal four colonies of son are arranged.
At first, the characteristic of the macro block relevant with detected shape is determined (S30).
According in S30 step result of determination, dynamic population control (DPC) method is applied to and detects the irrelevant piece of piece, promptly in the category-A piece.That is, in this method, be defined as father's chromosome with motion vector of the piece of former frame same position and the motion vector of each piece that is positioned at the upper and lower, left and right of this piece.From the piece of the field of search shown in Figure 4, select two motion vectors arbitrarily, and had two chromosomes replacements of smallest adaptation in six kinds of selections.DPC of per three substitute performances.Each arbitrary value is selected from the first and the 3rd field of search of the third generation.Each arbitrary value is selected from the second and the 4th field of search in the 6th generation.In the 9th generation and the 12 generation, repeat to select the region of search with above-mentioned order.The eyes of considering the people are more responsive in vertical direction, just determine order like this.
Whether overlapping outline line is determined (step 34) to the piece relevant with the shape that detects among the step S30.If overlapping outline line, the adjacent motion vectors with weighted value is used to (S36) in the initialization.That is, both made some part of shape be arranged in piece, the motion of shape also can influence whole.Yet, exist the background motion can not uncared-for situation, with the attached both of these case of giving of different weighted values.In category-A piece from adjacent block and the C class piece, select chromosomal each motion vector as the father.Carry out and intersect, so that more effectively follow the chromosomal feature of the father who belongs to the C class.That is, not by carrying out any mask but produce filial generation, can more effectively follow C class piece like this by giving weighted value.
The result who determines according to step S34, not under the situation of overlapping outline line, adjacent motion vectors is used to (S38) in the initialization at piece.In this case, preferably can in by the piece in the segmented shape, search for small motion.Because except considering mass motion, the motion of shape has also become the motion of piece self probably.At this moment, from C class piece, select a motion vector and from the piece of former frame same position select a motion vector as father's chromosome after, can utilize and for example intersect or the algorithm of conversion is searched for the small movements value.
In the present invention, in order to predict the optimal motion vector, the feature of piece image has been adopted heredity Algorithm. The most accurate method that is used for motion vectors is all direction search method, and it needs too many fortune Calculate. In order to address the above problem, the method for a simplification is employed, but lacks accuracy. That is, The total optimization value may not be found, and only concentrates on local optimum. And the application genetic algorithm, can Find the total optimization value, but do not have incorrect hypothesis. Yet, in genetic algorithm, have relatively a large amount of Computing. For this reason, in the block matching method according to the feature of image block, adopted genetic algorithm. This method, Billy based on the method for the genetic algorithm of adjacent block motion vector, produces optimum fortune with only The amount of calculation of dynamic vector is little, and uses this method and can calculate correct motion vector.
Claims (2)
1. initial method in using the block matching method of genetic algorithm comprises:
(a) from the reference frame of piece image, detect a reservation shape, and this image segmentation is become predetermined piece;
(b), divide by each piece of segmented image for the piece irrelevant, overlapping piece and the piece in detecting shape are arranged with the outline line that detects shape with detecting shape according to the relationship characteristic of each piece and this shape; With
(c) in each piece of dividing according to described feature, come the initialization motion vector by using suitable motion vector.
2. initial method as claimed in claim 1, wherein, at step (c), the piece irrelevant with detecting shape adopted the dynamic population control method, the piece of the outline line of the overlapping shape that detects is adopted the motion vector of the adjacent block with weighted value, the piece in detecting shape is adopted the motion vector of adjacent block.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1019970076399A KR100292348B1 (en) | 1997-12-29 | 1997-12-29 | Initializing method according to the feature of a detected shape in block matching algorithm to apply genetic algorithm |
| KR76399/1997 | 1997-12-29 | ||
| KR76399/97 | 1997-12-29 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1222041A CN1222041A (en) | 1999-07-07 |
| CN1139257C true CN1139257C (en) | 2004-02-18 |
Family
ID=19529250
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB981184901A Expired - Fee Related CN1139257C (en) | 1997-12-29 | 1998-08-20 | Initializing method according to feature block of detected shape in block matching method to which genetic algorithm is applied |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP3025481B2 (en) |
| KR (1) | KR100292348B1 (en) |
| CN (1) | CN1139257C (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4708712B2 (en) * | 2004-02-04 | 2011-06-22 | キヤノン株式会社 | Image processing apparatus, control method therefor, and program |
| KR101522306B1 (en) * | 2013-05-09 | 2015-05-26 | 서울대학교산학협력단 | A system and control method for a meta-heuristic algorithm utilizing similarity for performance enhancement |
| CN114040203B (en) * | 2021-11-26 | 2024-07-12 | 京东方科技集团股份有限公司 | Video data processing method, apparatus, device and computer storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2625356B2 (en) * | 1993-09-09 | 1997-07-02 | 日本電気株式会社 | Image processing device |
| JPH07123276A (en) * | 1993-10-21 | 1995-05-12 | Nippon Telegr & Teleph Corp <Ntt> | Digital compression coding method for picture signal |
-
1997
- 1997-12-29 KR KR1019970076399A patent/KR100292348B1/en not_active IP Right Cessation
-
1998
- 1998-08-20 CN CNB981184901A patent/CN1139257C/en not_active Expired - Fee Related
- 1998-09-02 JP JP24856098A patent/JP3025481B2/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| JP3025481B2 (en) | 2000-03-27 |
| JPH11205804A (en) | 1999-07-30 |
| KR100292348B1 (en) | 2001-06-01 |
| KR19990056404A (en) | 1999-07-15 |
| CN1222041A (en) | 1999-07-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110087087B (en) | VVC inter-frame coding unit prediction mode early decision and block division early termination method | |
| JP3242409B2 (en) | Method for moving grid of target image and apparatus using the same, method for estimating compression / motion using the same and apparatus therefor | |
| Wu et al. | SVM based fast CU partitioning algorithm for VVC intra coding | |
| CN111462261B (en) | Fast CU partitioning and intra-frame decision method for H.266/VVC | |
| CN111654698B (en) | Fast CU partition decision method for H.266/VVC | |
| CN1130847A (en) | Apparatus for encoding contour of regions contained in video signal | |
| CN110351561B (en) | Efficient reinforcement learning training method for video coding optimization | |
| Kamble et al. | Modified three-step search block matching motion estimation and weighted finite automata based fractal video compression | |
| CN1139257C (en) | Initializing method according to feature block of detected shape in block matching method to which genetic algorithm is applied | |
| CN107690069A (en) | A kind of cascade method for video coding of data-driven | |
| CN114257819A (en) | VCC coding unit fast dividing method based on space-time depth information | |
| CN100385957C (en) | Kinematic vector predicting method | |
| CN110225339A (en) | A kind of HEVC video sequence coding/decoding accelerated method | |
| CN110198442A (en) | A kind of intra-frame prediction method, device and storage medium for Video coding | |
| Wang et al. | Fast CU partitioning algorithm for VVC based on multi-stage framework and binary subnets | |
| CN114143536B (en) | Video coding method of SHVC (scalable video coding) spatial scalable frame | |
| CN112070851B (en) | Index map prediction method based on genetic algorithm and BP neural network | |
| CN110351555A (en) | Multipass based on intensified learning goes through video frequency coding rate distribution and control optimization method | |
| CN109982079B (en) | Intra-frame prediction mode selection method combined with texture space correlation | |
| Yuelei et al. | A genetic search algorithm for motion estimation | |
| Li et al. | Resnet approach for coding unit fast splitting decision of hevc intra coding | |
| CN115130483B (en) | Neural architecture searching method based on multi-target group intelligent algorithm and application | |
| Wang et al. | Fast Learning-Based Split Type Prediction Algorithm for VVC | |
| CN116958739B (en) | Attention mechanism-based carbon fiber channel real-time dynamic numbering method | |
| Lin et al. | CU partition prediction scheme for x265 intra coding using neural networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C06 | Publication | ||
| PB01 | Publication | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20040218 Termination date: 20090921 |