CN110084841A - A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator - Google Patents
A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator Download PDFInfo
- Publication number
- CN110084841A CN110084841A CN201910354028.8A CN201910354028A CN110084841A CN 110084841 A CN110084841 A CN 110084841A CN 201910354028 A CN201910354028 A CN 201910354028A CN 110084841 A CN110084841 A CN 110084841A
- Authority
- CN
- China
- Prior art keywords
- image
- matching
- weighting
- algorithm
- stereo matching
- 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.)
- Pending
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 19
- 238000001514 detection method Methods 0.000 claims abstract description 10
- 230000011218 segmentation Effects 0.000 claims abstract description 7
- 238000005259 measurement Methods 0.000 claims abstract description 4
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000013139 quantization Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 4
- 230000001133 acceleration Effects 0.000 claims description 3
- 238000004040 coloring Methods 0.000 claims description 3
- 238000005314 correlation function Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000002922 simulated annealing Methods 0.000 claims description 3
- 238000011524 similarity measure Methods 0.000 abstract description 4
- 238000000034 method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000006116 polymerization reaction Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to Stereo Matching Algorithm technical fields, the weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator that the invention discloses a kind of, the selection of matching and algorithm including weighted template, the matching of weighted template simultaneously is divided into the detection and machine vision of the setting of parameter, the segmentation of image, filtering, and the selection of algorithm includes the selection of feature space, cost calculating, similarity measurement and search space and search strategy.The present invention carries out cost calculating using the mode that two kinds of AD, gradient similarity measures combine, and is filtered using the improvement guidance figure being weighted based on LOG operator, to realize efficient, high-precision Stereo matching.
Description
Technical field
The present invention relates to Stereo Matching Algorithm technical field more particularly to a kind of weighting guidance figure filters based on LOG operator
Wave Stereo Matching Algorithm.
Background technique
In the Stereo matching research of early stage, people often carry out cost calculating using single similarity measure,
As a result vulnerable to the influence of environmental change, traditional sectional perspective matching algorithm is more when carrying out cost polymerization, using fixed window
The mode of mouth, this mode is simple and effective, but in depth discontinuity zone, matching effect is very poor, which needs to calculate
The weight of each pixel, the speed of service are slower in rectangular window.
Summary of the invention
The weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator that the invention proposes a kind of, to solve above-mentioned back
The problem of being proposed in scape technology.
The weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator that the invention proposes a kind of, specific step is such as
Under:
The matching of S1, weighted template: while the matching of weighted template includes the following steps:
1), the setting of parameter: object adopts image under different care intensity, through image capture device
Acquired image, is compared and is carried out by way of encoding to image the setting of parameter by collection；
2), the segmentation of image: the segmentation of image includes the sampling and quantization, the analysis of image and image shape of image again
Description and classification；
3), the detection filtered: filtering is detected and recorded using BGA package detection and surface defects detection；
4), machine vision；
The selection of S2, algorithm: the matching of algorithm specifically includes following steps:
The selection of M1, feature space: the image of original image and template is subjected to Characteristic Contrast and selects matched spy
Space is levied, while improving using the feature space of selection the matching performance of algorithm template and reducing search space and reduction figure
As noise；
M2, cost calculate；
M3, similarity measurement: the form that matched algorithm is defined as function is showed in the program of computer, together
When function include correlation function, distance function and mutual information function, measure characteristic matching image so as to effectively improve
Between template image feature whether similar accuracy；
M4, search space and search strategy.
Preferably, 2 in the S1) in image sampling and quantization specifically include by sample devices obtain image, warp
Image processing apparatus is crossed to store the image data in computer in a manner of array.
Preferably, 2 in the S1) in the analysis of image specifically refer to sentence by the color of image, brightness and texture
Disconnected image whether there is similarity, analyze whether divided image can be modified and merge.
Preferably, 2 in the S1) in image shape description and classification refer to by image by computer will be original
Image is compiled into corresponding coding, while coding quickly being classified by the controller of computer-internal.
Preferably, 4 in the S1) in machine vision refer to object is tracked using matched template, object
The posture of the cutting of body, the extraction of body surface information data and resistance is examined.
Preferably, the cost in the M2 calculates specific as follows: AD being used to mostly use single-pass as the algorithm of matching cost
Road, AD or mean value AD, wherein mean value AD usually first calculates separately AD in red, green, blue (R, G, B) triple channel, then asks again
The average value of three distributes different weights for R, G, B triple channel respectively, leads to for the color information for being sufficiently reserved original image
It crossing weighted sum and obtains AD matching cost, A weighting D can preferably keep the colouring information of original image as matching cost,
It is significantly better than using A weighting D as the effect that matching cost generates disparity map and uses single channel.
Preferably, the search space in the M2 and search strategy include that a corresponding collection is formulated according to the parameter of selection
Close, containing in the set can be such that the instruction of image registration transformation operates, wherein when geometric transformation search space it is main because
Element, while the form that the set deformation of image is divided into whole deformation, local deformation and global displacement deformation exists；Search strategy
It include exhaustive search, hierarchical search, simulated annealing, Directional acceleration and SSD algorithm, so as to many algorithms pair
The coding of image is calculated and is searched for.
A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator proposed by the present invention, beneficial effect are:
This kind is combined based on the weighting guidance figure filtering Stereo Matching Algorithm of LOG operator using two kinds of AD, gradient similarity measures
Mode carries out cost calculating, is filtered using the improvement guidance figure being weighted based on LOG operator, to realize efficient, high
The Stereo matching of precision.
Specific embodiment
It is next combined with specific embodiments below that the present invention will be further described.
A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator, specific steps are as follows:
The matching of S1, weighted template: while the matching of weighted template includes the following steps:
1), the setting of parameter: object adopts image under different care intensity, through image capture device
Acquired image, is compared and is carried out by way of encoding to image the setting of parameter, the sampling of image by collection
And quantization specifically includes and obtains image by sample devices, by image processing apparatus by the image data in computer with array
Mode stored；The analysis of image specifically refers to whether there is by the color of image, brightness and texture estimation image
Similarity, analyzes whether divided image can be modified and merge；The description and classification of image shape refer to and will scheme
As original image is compiled into corresponding coding by computer, while coding being carried out fastly by the controller of computer-internal
The classification of speed.
2), the segmentation of image: the segmentation of image includes the sampling and quantization, the analysis of image and image shape of image again
Description and classification；
3), the detection filtered: filtering is detected and recorded using BGA package detection and surface defects detection；
4), machine vision, refer to object is tracked using matched template, the cutting of object, body surface information
The extraction of data and the posture of resistance are examined；
By multiple steps of above-mentioned S1, so that be effectively retained the color information and structural information of image, improve
Reliability with cost.
The selection of S2, algorithm: the matching of algorithm specifically includes following steps:
The selection of M1, feature space: the image of original image and template is subjected to Characteristic Contrast and selects matched spy
Space is levied, while improving using the feature space of selection the matching performance of algorithm template and reducing search space and reduction figure
As noise reduces the difficulty to image procossing to effectively raise the quality of image.
M2, cost calculate: using AD to mostly use single channel, AD or mean value AD as the algorithm of matching cost, wherein
Value AD usually first calculates separately AD in red, green, blue (R, G, B) triple channel, seeks the average value of three, again then to be sufficiently reserved
The color information of original image distributes different weights for R, G, B triple channel respectively, obtains AD by weighted sum and matches generation
Valence, A weighting D can preferably keep the colouring information of original image as matching cost, using A weighting D as matching cost
Generate disparity map effect be significantly better than use single channel, using the mode that two kinds of AD, gradient similarity measures combine come into
Row initial cost calculates, and is filtered using the improvement guidance figure being weighted based on LOG operator, to realize efficiently, in high precision
Stereo matching.
M3, similarity measurement: the form that matched algorithm is defined as function is showed in the program of computer, together
When function include correlation function, distance function and mutual information function, measure characteristic matching image so as to effectively improve
Between template image feature whether similar accuracy.
M4, search space and search strategy search space and search strategy include that one is formulated according to the parameter of selection accordingly
Set, containing in the set can be such that the instruction of image registration transformation operates, wherein when geometric transformation search space master
Factor is wanted, while the form that the set deformation of image is divided into whole deformation, local deformation and global displacement deformation exists；Search
Strategy includes exhaustive search, hierarchical search, simulated annealing, Directional acceleration and SSD algorithm, so as to a variety of calculations
Method is calculated and is searched for the coding of image.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Anyone skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (7)
1. a kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator, which is characterized in that specific steps are as follows:
The matching of S1, weighted template: while the matching of weighted template includes the following steps:
1), the setting of parameter: object is acquired image under different care intensity, through image capture device, will
Acquired image compares and carries out the setting of parameter by way of encoding to image；
2), the segmentation of image: the segmentation of image includes the sampling of image and retouching for quantization, the analysis of image and image shape again
It states and classifies；
3), the detection filtered: filtering is detected and recorded using BGA package detection and surface defects detection；
4), machine vision；
The selection of S2, algorithm: the matching of algorithm specifically includes following steps:
The selection of M1, feature space: the image of original image and template is subjected to Characteristic Contrast and selects matched feature empty
Between, while improving using the feature space of selection the matching performance of algorithm template and reducing search space and reduce image and making an uproar
Sound；
M2, cost calculate；
M3, similarity measurement: the form that matched algorithm is defined as function is showed in the program of computer, while letter
Number includes correlation function, distance function and mutual information function, measures characteristic matching image and mould so as to effectively improve
Between plate characteristics of image whether similar accuracy；
M4, search space and search strategy.
2. a kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator according to claim 1, feature exist
In, in the S1 2) in image sampling and quantization specifically include by sample devices obtain image, filled by image procossing
It sets and stores the image data in computer in a manner of array.
3. a kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator according to claim 1, feature exist
In, in the S1 2) in the analysis of image specifically refer to whether deposit by the color of image, brightness and texture estimation image
In similarity, analyze whether divided image can be modified and merge.
4. a kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator according to claim 1, feature exist
In, in the S1 2) in image shape description and classification refer to original image be compiled into phase by computer by image
The coding answered, while coding quickly being classified by the controller of computer-internal.
5. a kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator according to claim 1, feature exist
In, in the S1 4) in machine vision refer to object is tracked using matched template, the cutting of object, object
The extraction of surface information data and the posture of resistance are examined.
6. a kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator according to claim 1, feature exist
In the cost in the M2 calculates specific as follows: AD being used to mostly use single channel, AD or mean value as the algorithm of matching cost
AD, wherein mean value AD usually first calculates separately AD in red, green, blue (R, G, B) triple channel, then seeks the average value of three again,
For the color information for being sufficiently reserved original image, different weights is distributed respectively for R, G, B triple channel, is obtained by weighted sum
AD matching cost, A weighting D can preferably keep the colouring information of original image as matching cost, using A weighting D conduct
The effect that matching cost generates disparity map, which is significantly better than, uses single channel.
7. a kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator according to claim 1, feature exist
In search space and search strategy in the M2 include formulating a corresponding set according to the parameter of selection, in the set
Containing can be such that the instruction of image registration transformation operates, wherein when geometric transformation search space principal element, while image
Set deformation be divided into whole deformation, local deformation and global displacement deformation form exist；Search strategy includes exhaustive
Search, hierarchical search, simulated annealing, Directional acceleration and SSD algorithm, so as to many algorithms to the coding of image
It is calculated and is searched for.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910354028.8A CN110084841A (en) | 2019-04-29 | 2019-04-29 | A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910354028.8A CN110084841A (en) | 2019-04-29 | 2019-04-29 | A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110084841A true CN110084841A (en) | 2019-08-02 |
Family
ID=67417707
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910354028.8A Pending CN110084841A (en) | 2019-04-29 | 2019-04-29 | A kind of weighting guidance figure filtering Stereo Matching Algorithm based on LOG operator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110084841A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111351449A (en) * | 2020-02-14 | 2020-06-30 | 广东工业大学 | Stereo matching method based on cost aggregation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020126915A1 (en) * | 2001-01-18 | 2002-09-12 | Shang-Hong Lai | Method for image alignment under non-uniform illumination variations |
| US20120163704A1 (en) * | 2010-12-23 | 2012-06-28 | Electronics And Telecommunications Research Institute | Apparatus and method for stereo matching |
| CN104361590A (en) * | 2014-11-12 | 2015-02-18 | 河海大学 | High-resolution remote sensing image registration method with control points distributed in adaptive manner |
| CN106530333A (en) * | 2016-10-10 | 2017-03-22 | 天津大学 | Hierarchy optimization stereo matching method based on binding constraint |
| CN107301664A (en) * | 2017-05-25 | 2017-10-27 | 天津大学 | Improvement sectional perspective matching process based on similarity measure function |
| CN107392943A (en) * | 2017-07-14 | 2017-11-24 | 天津大学 | Parallax refining algorithm based on multiple dimensioned weight guiding filtering |
| CN108520534A (en) * | 2018-04-23 | 2018-09-11 | 河南理工大学 | A kind of adaptive multimodality fusion Stereo Matching Algorithm |
-
2019
- 2019-04-29 CN CN201910354028.8A patent/CN110084841A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020126915A1 (en) * | 2001-01-18 | 2002-09-12 | Shang-Hong Lai | Method for image alignment under non-uniform illumination variations |
| US20120163704A1 (en) * | 2010-12-23 | 2012-06-28 | Electronics And Telecommunications Research Institute | Apparatus and method for stereo matching |
| CN104361590A (en) * | 2014-11-12 | 2015-02-18 | 河海大学 | High-resolution remote sensing image registration method with control points distributed in adaptive manner |
| CN106530333A (en) * | 2016-10-10 | 2017-03-22 | 天津大学 | Hierarchy optimization stereo matching method based on binding constraint |
| CN107301664A (en) * | 2017-05-25 | 2017-10-27 | 天津大学 | Improvement sectional perspective matching process based on similarity measure function |
| CN107392943A (en) * | 2017-07-14 | 2017-11-24 | 天津大学 | Parallax refining algorithm based on multiple dimensioned weight guiding filtering |
| CN108520534A (en) * | 2018-04-23 | 2018-09-11 | 河南理工大学 | A kind of adaptive multimodality fusion Stereo Matching Algorithm |
Non-Patent Citations (2)
| Title |
|---|
| 周博等: "基于高斯拉普拉斯算子的加权引导图滤波立体匹配算法", 《激光与光电子学进展》 * |
| 汤勃等: "机器视觉表面缺陷检测综述", 《中国图象图形学报》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111351449A (en) * | 2020-02-14 | 2020-06-30 | 广东工业大学 | Stereo matching method based on cost aggregation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108549873B (en) | Three-dimensional face recognition method and three-dimensional face recognition system | |
| CN107578418B (en) | Indoor scene contour detection method fusing color and depth information | |
| CN108197587B (en) | Method for performing multi-mode face recognition through face depth prediction | |
| CN105956582B (en) | A kind of face identification system based on three-dimensional data | |
| CN107248159A (en) | A kind of metal works defect inspection method based on binocular vision | |
| CN105574527B (en) | A kind of quick object detecting method based on local feature learning | |
| US6701005B1 (en) | Method and apparatus for three-dimensional object segmentation | |
| CN101443817B (en) | Method and device for determining correspondence, preferably for the three-dimensional reconstruction of a scene | |
| CN104835175B (en) | Object detection method in a kind of nuclear environment of view-based access control model attention mechanism | |
| CN108710913A (en) | A kind of switchgear presentation switch state automatic identification method based on deep learning | |
| CN107403168A (en) | A kind of facial-recognition security systems | |
| CN103824059A (en) | Facial expression recognition method based on video image sequence | |
| CN109118528A (en) | Singular value decomposition image matching algorithm based on area dividing | |
| CN106155299B (en) | A kind of pair of smart machine carries out the method and device of gesture control | |
| CN109583376B (en) | Ancient ceramic source breaking and generation breaking method based on multi-feature information fusion | |
| CN111598172B (en) | Dynamic target grabbing gesture rapid detection method based on heterogeneous depth network fusion | |
| CN102163281A (en) | Real-time human body detection method based on AdaBoost frame and colour of head | |
| CN112288758B (en) | Infrared and visible light image registration method for power equipment | |
| CN105957107A (en) | Pedestrian detecting and tracking method and device | |
| CN107886110A (en) | Method for detecting human face, device and electronic equipment | |
| CN105956544A (en) | Remote sensing image road intersection extraction method based on structural index characteristic | |
| CN112085675A (en) | Depth image denoising method, foreground segmentation method and human motion monitoring method | |
| CN106529441A (en) | Fuzzy boundary fragmentation-based depth motion map human body action recognition method | |
| CN108242061B (en) | Supermarket shopping cart hand identification method based on Sobel operator | |
| CN104463896B (en) | Image corner point detection method and system based on kernel similar region distribution characteristics |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190802 |