CN102322820B - Automatic separation method for front and rear surface reflected light spots in surface shape detection system - Google Patents
Automatic separation method for front and rear surface reflected light spots in surface shape detection system Download PDFInfo
- Publication number
- CN102322820B CN102322820B CN 201110270339 CN201110270339A CN102322820B CN 102322820 B CN102322820 B CN 102322820B CN 201110270339 CN201110270339 CN 201110270339 CN 201110270339 A CN201110270339 A CN 201110270339A CN 102322820 B CN102322820 B CN 102322820B
- Authority
- CN
- China
- Prior art keywords
- hot spot
- image
- spot
- wave filter
- data
- 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
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an automatic separation method for front and rear surface reflected light spots in a surface shape detection system. The method comprises the following steps of: A1, acquiring a current front and rear surface reflected light spot image F1 by a charge coupled device (CCD); A2, filtering noise by adopting a method combining background subtraction and multiplication filtration, and performing iterative threshold segmentation to obtain a binary image F2; A3, detecting all light spots in the image F2, and extracting the centroid of all the light spots; A4, tracking with a Kalman filter; A5, searching a nearest light spot centroid around the last predicted centroid, outputting the nearest light spot centroid, transmitting all found front surface reflected light spot centroid data to the Kalman filter, and predicting the position of the light spot centroid of the next frame by using the last predicted data and the actual extracted data through the Kalman filter; and A6, if the position is the data of the last frame, ending. The Kalman filter-based tracking algorithm accurately distinguishes the front surface reflected light spots, and the experimental result statistics of multiple times shows that the tracking accuracy is more than 95 percent.
Description
Technical field
The present invention relates to front and rear surfaces flare automatic separation method in a kind of the shape detection system, belong to optical module face shape detection technique and image processing field.
Background technology
In the face shape testing process that adopts based on the angular difference method, detection laser forms the front and rear surfaces hot spot at CCD after the reflection of optical module front and rear surfaces is by optical system.When the optical element front and rear surfaces depth of parallelism is very high, make that the front and rear surfaces measurement reflected light angle of same measurement point correspondence is very little, overlap hot spot at CCD.Carry out centroid calculation if overlapping hot spot is regarded as integral body, will introduce the bigger error of calculation, cause surface shape measurement result's error.Simultaneously in the reconstruct of face shape, can only use the front surface hot spot, therefore the front and back flare must be separated automatically, to reject the surperficial hot spot of the back of the body.Therefore, the front and back hot spot automatic separation method that needs the practical application of a kind of energy of research.
When carrying out optical module face shape and detect, behind the detected optics assembly of L bundle laser incident, after the reflection of optical module front and rear surfaces can CCD form K (the individual front and rear surfaces laser facula of L≤K≤2L), as shown in Figure 1.Hot spot can be divided into 3 class hot spots such as A, B, C, according to criterion, realize extraction and the classification of front surface hot spot in conjunction with the gray feature of front and rear surfaces flare.Concrete grammar is as follows:
(1) manually reject back of the body surface hot spot in first sampled point light spot image, detect center-of-mass coordinate and the classification of all front surface hot spots in this image, with serve as to classify with reference to hot spot;
(2) since the 2nd sampled point, cut apart containing the image of carrying on the back surperficial hot spot, extract the barycenter of all hot spots and gray scale and, form the hot spot S set;
(3) find out the light spot group intersection that meets formula C class from the S set, from these combinations, choose one group of the gray-scale value maximum again as 4 C class front surface hot spots; From the S set, remove these combinations;
(4) if current scan point and reference hot spot different rows utilize formula category-B hot spot the 2nd decision rule, upgrading with reference to facula information is current scan point; As the colleague, utilize category-B hot spot the 2nd decision rule to differentiate.The light spot group cooperation that will meet this criterion is the candidate, finds out one group of the gray-scale value maximum as 2 category-B front surface hot spots.From the S set, remove the hot spot combination of finding;
(5) last, utilize and differentiate category-A candidate hot spot, find out the real category-A hot spot of conduct of gray scale maximum;
There is following shortcoming in this method:
(1) needs rear surface flare and type in first two field picture are manually marked.
(2) separation of hot spot depends on the threshold value in the decision rule before and after;
(3) have overlapping the time just helpless when the front and back hot spot.
The rejecting of general spot backlight all is that the method by physically solves, and is the most frequently used and effectively carry on the back the surface at optical module and be coated with last layer vaseline.But optical module debug put on the shelf after, this mode is difficult to carry out because when optical module debug put on the shelf and move after, be not allow vaseline is smeared in the rear surface of optical module.
Summary of the invention
Technical matters to be solved by this invention is to provide front and rear surfaces flare automatic separation method in a kind of the shape detection system at the deficiencies in the prior art.
Front and rear surfaces flare automatic separation method in a kind of the shape detection system may further comprise the steps:
A1:CCD gathers the light spot image F1 of current front and rear surfaces reflection; If first two field picture is then manually wiped back of the body surface reflection spot, follow the tracks of for the back and select correct target;
A2: the method that adopts background subtraction to combine with multiplication filtering comes the filtering noise row iteration threshold value of going forward side by side to cut apart, to isolate laser facula; Obtain bianry image F2;
A3: all hot spots among the detected image F2, if the hot spot adhesion is arranged, then carry out the separation of adhesion hot spot, extract the barycenter of all hot spots;
A4: utilize the Kalman wave filter to follow the tracks of.If first two field picture with facula mass center data initialization Kalman wave filter, and is done prediction for the first time, if first frame changes steps A 1, otherwise, then adopt the Kalman wave filter to follow the tracks of, namely change steps A 5;
A5: seek nearest facula mass center and output on every side at the last barycenter of once predicting, all front surface reflection facula mass center data that find are passed to the Kalman wave filter, and the last predicted data of Kalman wave filter utilization and this actual amount paid be it is predicted the position of next frame facula mass center;
A6: if the last frame data then finish.
Front and rear surfaces flare automatic separation method in described the shape detection system, in the described steps A 3, for the hot spot that adhesion is arranged, the method that its hot spot separates is as follows:
A31: bianry image F2 is carried out range conversion, obtain image F3, make overlapping hot spot concave point more obvious;
A32: the changing image F3 that adjusts the distance adopts the structural element that increases gradually to carry out the morphology ON operation, calculates image behind each ON operation and the difference of original image, obtains spot radius R;
A33: utilize overlapping differentiation, from F3, extract all overlapping regions and constitute overlapping region bianry image F4;
A34: to each overlapping region among the F4, follow the tracks of the overlapping region profile, extract chain code; According to chain code and radius R, ask for the centre point of each point correspondence, be gathered into the central area of each hot spot;
A35: with the slightly little circular configuration element of radius ratio R F4 is corroded and to obtain F5, and central area image and F5 are carried out intersection operation, depart from spot center centre point far away thereby delete those, obtain center of circle area image F6;
A36: to each zone, the hot spot center of circle, be that the circular configuration element bigger slightly than R carries out an expansive working with radius, make intersection operation with F2 then, its result is just as the shape estimation to the corresponding hot spot of zone, this center of circle institute.
The present invention utilizes the kalman track algorithm, in first two field picture of each experiment, will carry on the back surface reflection spot people for wiping, and follows the tracks of the hot spot of all front surface reflections, thereby reaches the purpose of classification front and rear surfaces hot spot; For the front and rear surfaces hot spot situation of coincidence is arranged, adopt based on the overlapping cutting techniques of the hot spot of range conversion, realize the separation of overlapping hot spot well; Employing can right area have been told the hot spot of front surface reflection based on the track algorithm of Kalman wave filter, and repeatedly the experimental result statistics shows and follows the tracks of accuracy more than 95%.
Description of drawings
Fig. 1 is surperficial hot spot separation method synoptic diagram of the prior art;
Fig. 2 is the hot spot separation process that the present invention is based on the Kalman wave filter;
Fig. 3 is the experiment effect figure that the present invention is based on the hot spot separation of Kalman wave filter; 3-1 is first frame, and 3-2 is second frame, and 3-3 is the 4th frame, and 3-4 is the 6th frame, and 3-5 is the 8th frame, and 3-6 is the 12 frame;
The experimental result picture that Fig. 4 carries out overlapping separation for the present invention to three class laser faculas.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.
Embodiment 1
Front and back flare separation method based on kalman filtering
When carrying out the test of face shape, will carry on the back surface reflection spot people for wiping to first two field picture that obtains, follow the tracks of the hot spot of all front surface reflections, thereby reach the purpose of classification front and rear surfaces hot spot.With reference to figure 2, its key step is:
A1:CCD gathers the light spot image F1 of current front and rear surfaces reflection; If first two field picture is then manually wiped back of the body surface reflection spot, follow the tracks of for the back and select correct target;
A2: the method that adopts background subtraction to combine with multiplication filtering comes the filtering noise row iteration threshold value of going forward side by side to cut apart, to isolate laser facula; Obtain bianry image F2;
A3: all hot spots among the detected image F2, if the hot spot adhesion is arranged, then carry out the separation of adhesion hot spot, extract the barycenter of all hot spots.
For the hot spot that adhesion is arranged, the main flow process of its hot spot separation method is as follows:
A31: bianry image F2 is carried out range conversion (seeing formula 1), obtain image F3, make overlapping hot spot concave point more obvious.
d＝|m-k|+|n-l| (1)
Wherein, (k l) is boundary pixel, and (m n) is area pixel.
A32: the changing image F3 that adjusts the distance adopts the structural element that increases gradually to carry out the morphology ON operation, calculates image behind each ON operation and the difference of original image, obtains spot radius R.
A33: utilize overlapping differentiation (seeing formula 2), from F3, extract all overlapping regions and constitute overlapping region bianry image F4.
Decision rule is defined as follows:
P wherein
0Be the overlapping discrimination threshold of hot spot, PE=4 π A/C
2(0＜PE≤1), A is the area of hot spot, C is the girth of hot spot.
A34: to each overlapping region among the F4, follow the tracks of the overlapping region profile, extract chain code.According to chain code and radius R, ask for the centre point of each point correspondence, be gathered into the central area of each hot spot.
A35: with the slightly little circular configuration element of radius ratio R F4 is corroded and to obtain F5, and central area image and F5 are carried out intersection operation, depart from spot center centre point far away thereby delete those, obtain center of circle area image F6.
A36: to each zone, the hot spot center of circle, be that the circular configuration element bigger slightly than R carries out an expansive working with radius, make intersection operation with F2 then, its result is just as the shape estimation to the corresponding hot spot of zone, this center of circle institute.
A4: utilize the Kalman wave filter to follow the tracks of.If first two field picture with facula mass center data initialization Kalman wave filter, and is done prediction for the first time, if first frame changes steps A 1, otherwise, then adopt the Kalman wave filter to follow the tracks of, namely change steps A 5.
A5: seek nearest facula mass center and output (being the hot spot of front surface reflection) on every side at the last barycenter of once predicting, all front surface reflection facula mass center data that find are passed to the Kalman wave filter, and the last predicted data of Kalman wave filter utilization and this actual amount paid be it is predicted the position of next frame facula mass center;
A6: if the last frame data then finish;
Embodiment 2
Employing is based on tracking effect such as Fig. 3 of Kalman wave filter, the hot spot of hot spot among Fig. 3 in the rectangle frame for detecting, the front surface reflection hot spot of hot spot for following the tracks of in the circular frame, can find under the situation that the front and rear surfaces flare occurs, employing can right area have been told the hot spot of front surface reflection based on the track algorithm of Kalman wave filter, and repeatedly the experimental result statistics shows and follows the tracks of accuracy more than 95%.
Embodiment 3
Overlapping hot spot separating experiment result based on range conversion
In the experiment three class laser faculas have been carried out overlapping separation: (1) big or small basically identical, overlapping not serious hot spot (Fig. 4-1); (2) the overlapping hot spot that causes not of uniform size (Fig. 4-2); (3) big or small basically identical, overlapping more serious hot spot (Fig. 4-3).Wherein, be followed successively by former figure, segmentation effect figure, center of circle areal map, separating resulting figure from left to right among Fig. 4-1, Fig. 4-2 and Fig. 4-3.Experimental result shows, by method of the present invention, can realize the separation of overlapping hot spot well.
Should be understood that, for those of ordinary skills, can be improved according to the above description or conversion, and all these improvement and conversion all should belong to the protection domain of claims of the present invention.
Claims (2)
1. front and rear surfaces flare automatic separation method in the face shape detection system is characterized in that, may further comprise the steps:
A1:CCD gathers the light spot image F1 of current front and rear surfaces reflection; If first two field picture is then manually wiped back of the body surface reflection spot, follow the tracks of for the back and select correct target;
A2: the method that adopts background subtraction to combine with multiplication filtering comes the filtering noise row iteration threshold value of going forward side by side to cut apart, to isolate laser facula; Obtain bianry image F2;
A3: all hot spots among the detected image F2, if the hot spot adhesion is arranged, then carry out the separation of adhesion hot spot, extract the barycenter of all hot spots;
A4: utilize the Kalman wave filter to follow the tracks of; If first two field picture with facula mass center data initialization Kalman wave filter, and is done prediction for the first time, if first frame changes steps A 1, otherwise, then adopt the Kalman wave filter to follow the tracks of, namely change steps A 5;
A5: seek nearest facula mass center and output on every side at the last barycenter of once predicting, all front surface reflection facula mass center data that find are passed to the Kalman wave filter, and the last predicted data of Kalman wave filter utilization and this actual amount paid be it is predicted the position of next frame facula mass center;
A6: if the last frame data then finish, otherwise, forward steps A 1 to.
2. front and rear surfaces flare automatic separation method in according to claim 1 shape detection system is characterized in that, in the described steps A 3, for the hot spot that adhesion is arranged, the method that its hot spot separates is as follows:
A31: bianry image F2 is carried out range conversion, obtain image F3, make overlapping hot spot concave point more obvious;
A32: the changing image F3 that adjusts the distance adopts the structural element that increases gradually to carry out the morphology ON operation, calculates image behind each ON operation and the difference of original image, obtains spot radius R;
A33: utilize overlapping differentiation, from F3, extract all overlapping regions and constitute overlapping region bianry image F4;
A34: to each overlapping region among the F4, follow the tracks of the overlapping region profile, extract chain code; According to chain code and radius R, ask for the centre point of each point correspondence, be gathered into the central area of each hot spot;
A35: with the slightly little circular configuration element of radius ratio R F4 is corroded and to obtain F5, and central area image and F5 are carried out intersection operation, depart from spot center centre point far away thereby delete those, obtain center of circle area image F6;
A36: to each zone, the hot spot center of circle, be that the circular configuration element bigger slightly than R carries out an expansive working with radius, make intersection operation with F2 then, its result is just as the shape estimation to the corresponding hot spot of zone, this center of circle institute.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110270339 CN102322820B (en) | 2011-09-14 | 2011-09-14 | Automatic separation method for front and rear surface reflected light spots in surface shape detection system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110270339 CN102322820B (en) | 2011-09-14 | 2011-09-14 | Automatic separation method for front and rear surface reflected light spots in surface shape detection system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102322820A CN102322820A (en) | 2012-01-18 |
| CN102322820B true CN102322820B (en) | 2013-07-10 |
Family
ID=45450616
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110270339 Expired - Fee Related CN102322820B (en) | 2011-09-14 | 2011-09-14 | Automatic separation method for front and rear surface reflected light spots in surface shape detection system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102322820B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8898487B2 (en) | 2010-12-27 | 2014-11-25 | Microsoft Corporation | Power management via coordination and selective operation of timer-related tasks |
| CN104019764B (en) * | 2014-06-20 | 2016-08-24 | 中国工程物理研究院激光聚变研究中心 | 2 × 2 array light source scan-type surface shape measurement optical system scaling methods |
| CN108981585A (en) * | 2017-06-01 | 2018-12-11 | 上海砺晟光电技术有限公司 | It can accurately measure the laser displacement sensor of curved surface displacement of targets |
| CN107452026A (en) * | 2017-08-08 | 2017-12-08 | 何佳芮 | A kind of processing system and its method for image spot barycenter |
| CN110766700B (en) * | 2019-10-23 | 2022-01-28 | 吉林大学 | ICP-AES spectral image processing method based on digital micromirror |
| CN112150545B (en) * | 2020-09-28 | 2024-07-12 | 中国科学院空间应用工程与技术中心 | Tobit Kalman filter-based facula centroid obtaining method and Tobit Kalman filter-based facula centroid obtaining device |
| CN112634304B (en) * | 2020-12-31 | 2022-09-13 | 上海易维视科技有限公司 | Method for removing reflection light spots in 3D format video or image |
| CN113155756B (en) * | 2021-03-31 | 2022-10-04 | 中国科学院长春光学精密机械与物理研究所 | Light spot online calibration method |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US3879131A (en) * | 1974-02-06 | 1975-04-22 | Bell Telephone Labor Inc | Photomask inspection by real time diffraction pattern analysis |
| US6201601B1 (en) * | 1997-09-19 | 2001-03-13 | Kla-Tencor Corporation | Sample inspection system |
| CN102128608A (en) * | 2010-12-07 | 2011-07-20 | 北京航空航天大学 | Highly dynamic two-dimensional attitude angle measuring method and system |
-
2011
- 2011-09-14 CN CN 201110270339 patent/CN102322820B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN102322820A (en) | 2012-01-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102322820B (en) | Automatic separation method for front and rear surface reflected light spots in surface shape detection system | |
| Gamba et al. | Improving urban road extraction in high-resolution images exploiting directional filtering, perceptual grouping, and simple topological concepts | |
| CN102855622B (en) | A kind of infrared remote sensing image sea ship detection method based on significance analysis | |
| CN102298698B (en) | Remote sensing image airplane detection method based on fusion of angle points and edge information | |
| Shi et al. | Real-time traffic light detection with adaptive background suppression filter | |
| CN111626290A (en) | Infrared ship target detection and identification method under complex sea surface environment | |
| CN112215819B (en) | Airport pavement crack detection method based on depth feature fusion | |
| CN105740809A (en) | Expressway lane line detection method based on onboard camera | |
| CN113420607A (en) | Multi-scale target detection and identification method for unmanned aerial vehicle | |
| Cygert et al. | Toward robust pedestrian detection with data augmentation | |
| CN109559324A (en) | A kind of objective contour detection method in linear array images | |
| CN104615986A (en) | Method for utilizing multiple detectors to conduct pedestrian detection on video images of scene change | |
| Abbiramy et al. | Spermatozoa detection, counting and tracking in video streams to detect asthenozoospermia | |
| CN111091095A (en) | Method for detecting ship target in remote sensing image | |
| CN103955949A (en) | Moving target detection method based on Mean-shift algorithm | |
| CN113378763A (en) | SAR image-oriented target automatic detection method and system | |
| CN104237121A (en) | Sperm motility performance detection method | |
| CN104182992A (en) | Method for detecting small targets on the sea on the basis of panoramic vision | |
| CN102663385A (en) | Detection method for spot target on satellite | |
| CN101976342B (en) | Space-time collaborative determination-based cat eye effect target identification method | |
| Li et al. | Efficient multi-lane detection based on large-kernel convolution and location | |
| Ghahremani et al. | Cascaded CNN method for far object detection in outdoor surveillance | |
| Peng et al. | Real-time traffic sign text detection based on deep learning | |
| CN115797366A (en) | Hough transform-based moving target extraction image processing method and system | |
| Yang et al. | Method for building recognition from FLIR images |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130710 Termination date: 20140914 |
|
| EXPY | Termination of patent right or utility model |