CN105160692A - First-moment center of mass calculating method for sliding window with threshold - Google Patents
First-moment center of mass calculating method for sliding window with threshold Download PDFInfo
- Publication number
- CN105160692A CN105160692A CN201510167617.7A CN201510167617A CN105160692A CN 105160692 A CN105160692 A CN 105160692A CN 201510167617 A CN201510167617 A CN 201510167617A CN 105160692 A CN105160692 A CN 105160692A
- Authority
- CN
- China
- Prior art keywords
- center
- mass
- window
- barycenter
- moment
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to a first-moment center of mass calculating method for a sliding window with a threshold. The method includes the following steps: 1) calculating the center of mass of a global image through a first-moment center of mass algorithm, and obtaining an initial center of mass position x0, y0; 2) selecting a square zone with the side length being r as a window by taking the x0, y0 as a center, utilizing the first-moment center of mass algorithm again to calculate the center of mass in the window, and obtaining a center of mass coordinate x1, y1, wherein the dimension of the window is slightly greater than that of a light spot; 3) defining g which equals |x1-x0|, and taking g as the amount of dispersion of the center of mass in an x direction, wherein l=|y1-y0| is the amount of dispersion of the center of mass in a y direction; and 4) determining whether the amount of dispersion of the center of mass g and the amount of dispersion of the center of mass 1 are less than a threshold T, T being 0.1pixel1, taking the x1, y1 as a coordinate of the center of mass of the light spot if g and 1 are less than a threshold T, otherwise, making x0 equal x1 and y0 equal y1, and returning to the step 2. According to the invention, the calculating precision of the center of mass of a light spot target is greatly improved, and the method is suitable for a self-adaptive optical system that needs higher precision.
Description
Technical field
The present invention relates to image processing techniques, is a kind of blind image restoring technology for ADAPTIVE OPTICS SYSTEMS partial correction image, specifically adds the moving window first moment centroid computing method of thresholding.
Background technology
In ADAPTIVE OPTICS SYSTEMS, the CCD camera that usual employing precision is high, reliability is high detects the centroid position in each aperture of spot array as sensitive detection parts, and restore wavefront information according to centroid position, therefore, the centroid detecting accuracy of hot spot and computational accuracy are the main factors of influential system precision.At present, it is also the centroid computing method be most widely used at present that first moment centroid computing method also claims gravity model appoach to be the most basic mass center estimation method, and this algorithm is comparatively simple owing to calculating, and is adopted widely at present.Adopt above centroid computing method, centroid detecting accuracy determined by following factor:
(1) total photoelectron number of signal
(2) summation of dark background and signal to noise ratio (S/N ratio)
(3) each pixel reads noise variance
(4) the detection window size of CCD
(5) intensity distributions of hot spot and equivalent Gaussian width thereof
(6) centroid position of dark background
(7) distance of actual signal facula mass center and dark background centroid position
Change any one factor above all to impact the detecting error of barycenter, therefore a lot of scholar sets about proposing various method to improve the detection accuracy of barycenter from above factor, below will be introduced several representative method.
1. subtract threshold method.Subtract thresholding algorithm and effectively can reduce the impact of reading noise, background dark current noise etc. on barycenter detecting error, but too low threshold value can not completely by noise remove, too high threshold value can cut part effective light spot signal, in actual applications, the impact by the change such as external condition and the gain of camera own is larger.
2. windowing method.Windowing method can effectively reduce the impact of the noise beyond window on centroid detecting accuracy, by suitably changing detection window size to reduce the impact of the pixel away from facula mass center position, to improve centroid detecting accuracy.But the size of window can not reduction simply, when laser image spot vegetarian refreshments can not all in calculation window time, error also will increase, and therefore when carrying out windowing method and carrying out centroid calculation, choosing of window size is most important.
3. weighing first order Moment Methods.Utilize the difference of Gauss's fractions distribution of spot signal and the gray-scale value of spot signal and background and noise signal gray-scale value, the method has larger difficulty when actual computation realizes.
Summary of the invention
For the defect of above centroid computing method, the present invention proposes the moving window first moment centroid computing method adding thresholding, the method takes full advantage of the feature of ADAPTIVE OPTICS SYSTEMS and target, barycenter has stronger convergence, convergence precision is higher, is applicable to require higher ADAPTIVE OPTICS SYSTEMS to computational accuracy.
For achieving the above object, the present invention by the following technical solutions: the moving window first moment centroid computing method adding thresholding, is characterized in that, comprise the following steps:
1) use first moment centroid algorithm to calculate the barycenter of global image, obtain initial centroid position x
0, y
0;
2) with x
0, y
0centered by choose the length of side be the square region of r as window, reuse first moment centroid algorithm and calculate barycenter in this window, obtain center-of-mass coordinate x
1, y
1; Window size is slightly larger than spot size;
3) g=|x is defined
1-x
0| be the barycenter departure in x direction, l=|y
1-y
0| be the barycenter departure in y direction;
4) judge whether barycenter departure g and l is less than threshold value T, T is 0.1pixel, if be less than, by x
1, y
1as the center-of-mass coordinate of hot spot; Otherwise, make x
0=x
1, y
0=y
1, proceed to step 2).
The calculation expression of described first moment centroid algorithm discrete form is
Wherein I
ijfor the gray scale light intensity at (i, j) coordinate place, x
i, y
jfor the coordinate of the x at (i, j) place, y, L, M are the row and column of image respectively.
The present invention substantially increases the centroid calculation precision of hot spot target, is applicable to the ADAPTIVE OPTICS SYSTEMS higher to accuracy requirement.
Accompanying drawing explanation
Fig. 1 is the process flow diagram that the present invention adds the moving window first moment centroid computing method of thresholding;
Fig. 2 is the schematic diagram adopting the method for Fig. 1 to calculate barycenter;
Fig. 3 is the barycenter deviation curve that under different reading noise level, the present invention and first moment algorithm calculate;
Barycenter deviation curve when Fig. 4 is window sliding number of times different under different equivalent Gaussian width;
Fig. 5 be the centroid distance between facula mass center from dark background different time, the change curve of barycenter deviation under different window sliding number of times.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
Embodiment 1
Spot size is 3pixel × 3pixel, and according to the flow process shown in Fig. 1, first use first moment centroid algorithm to calculate the barycenter of global image, formula is as follows
Obtain initial centroid position x
0, y
0; With x
0, y
0centered by choose length of side r be the square region of 5pixel as window, namely window size is 5pixel × 5pixel, reuses first moment centroid algorithm and calculates barycenter in this window, obtain center-of-mass coordinate x
1, y
1; Make g=|x
1-x
0| be the barycenter departure in x direction, l=|y
1-y
0| be the barycenter departure in y direction; Judge whether barycenter departure g and l is less than threshold value T, T is 0.1pixel, if be less than, by x
1, y
1as the center-of-mass coordinate of hot spot; Otherwise, make x
0=x
1, y
0=y
1, repeat above-mentioned steps.
Embodiment 2
Spot size is 3pixel × 3pixel, and first use first moment centroid algorithm to calculate the barycenter of global image, formula is as follows
Obtain initial centroid position x
0, y
0; With x
0, y
0centered by choose length of side r be the square region of 7pixel as window, namely window size is 7pixel × 7pixel, reuses first moment centroid algorithm and calculates barycenter in this window, obtain center-of-mass coordinate x
1, y
1; Make g=|x
1-x
0| be the barycenter departure in x direction, l=|y
1-y
0| be the barycenter departure in y direction; Judge whether barycenter departure g and l is less than threshold value T, T is 0.1pixel, if be less than, by x
1, y
1as the center-of-mass coordinate of hot spot; Otherwise, make x
0=x
1, y
0=y
1, repeat above-mentioned steps.
In specific implementation process, the size of window slightly larger than spot size, should be crossed conference and reduces computational accuracy.
Can see intuitively in fig. 2, (a) uses first moment algorithm to calculate centroid position, and is the center windowing that is window with this centroid position; B () reuses the centroid position calculated of first moment algorithm in window ranges; The centroid offset of (c) twice calculating; D (), using the centroid position of n calculating as the barycenter of window, moving window, reuses first moment algorithm and calculates barycenter, carry out the slip of window by above method in window; E centroid position that () finally restrains.The intensive some place of concentrating can be regarded as discrete light spot place, and the more sparse point that distributes is as noise spot.
Fig. 3 is under different reading noise levels, the barycenter deviation curve figure that algorithm herein and first moment algorithm calculate, simulated conditions: image size 15pixel × 15pixel, hot spot is 3pixel × 3pixel, incoming signal photon number 30, AD conversion coefficient is 1, quantum efficiency is 1, the mean intensity of dark background is 100ADU, facula mass center is 2.6pixel apart from the distance of background barycenter, and be 0.1pixel by threshold value T, the window size of detection is 3pixel × 3pixel, the noise added is average is 0, variance is respectively 0,2,4ADU.Can find out that after the windowing of 3 times, reach convergence, the deviation of centroid detection is tending towards 0 when the variance of reading noise is 0.When reading noise level and increasing, the centroid calculation deviation of final convergence obviously increases.The barycenter that 0th windowing and first moment method calculate, can find out the slip along with repeatedly window, the precision of centroid calculation is significantly improved.
Fig. 4 is under different spot size sizes, the barycenter deviation curve figure that algorithm herein and first moment algorithm calculate.Simulated conditions: image size 15pixel × 15pixel, incoming signal photon number 30, AD conversion coefficient is 1, and quantum efficiency is 1, adding average is 0, variance is the reading noise of 1ADU, and the mean intensity of dark background is 100ADU, and the distance of facula mass center and dark background barycenter is 2.6pixel, be 0.1pixel by threshold value T, the window size of detection is the number that R in 3pixel × 3pixel, figure represents hot spot pixel in the X direction, and R=1 represents that spot size is 1pixel × 1pixel.The reduction that centroid calculation precision is not single along with the reduction of spot size can be found out, when hot spot is greater than window size, because the effective information of hot spot is not added up completely interior, cause barycenter deviation to some extent; When hot spot is less than window size, within the scope of calculation window, noise increases, and also can cause the increase of centroid calculation error, and this is similar to windowing first moment algorithm.Now use the error of the center-of-mass coordinate adding the final convergence of the moving window first moment centroid computing method of thresholding minimum when hot spot equivalence Gaussian width R=3 as can be seen from Figure 4, along with the carrying out of moving window number of times, the moving window first moment centroid computing method adding thresholding is significantly improved than the centroid calculation precision of first moment centroid computing method.
Fig. 5 be the centroid distance between facula mass center from dark background different time, the change curve of barycenter deviation under different window sliding number of times, 0th time moving window is the overall barycenter using first moment method to calculate, and in figure, L represents the centroid distance between facula mass center and dark background.As can be seen from the figure, the barycenter deviation that calculates of first moment along with the distance of facula mass center and dark background barycenter increase and increase.Centroid computing method of the present invention reduces gradually along with window sliding barycenter deviation, and needing more iterative computation just can reach convergence when the distance of the two increases, the result of final convergence shows along with the increase barycenter deviation of facula mass center and dark background barycenter spacing also increases.Simulated conditions: image size 15pixel × 15pixel, incoming signal photon number 30, hot spot is 3pixel × 3pixel, AD conversion coefficient is 1, and quantum efficiency is 1, and reading noise average is 0, variance is 1ADU, the mean intensity of dark background is 100ADU, is 0.1pixel by threshold value T, and window size is 3pixel × 3pixel.
Be more than better embodiment of the present invention, but protection scope of the present invention is not limited thereto.Any those of ordinary skill in the art are in the technical scope disclosed by the present invention, and the conversion expected without creative work or replacement, all should be encompassed within protection scope of the present invention.Therefore the protection domain that protection scope of the present invention should limit with claim is as the criterion.
Claims (2)
1. add the moving window first moment centroid computing method of thresholding, it is characterized in that, comprise the following steps:
1) use first moment centroid algorithm to calculate the barycenter of global image, obtain initial centroid position x
0, y
0;
2) with x
0, y
0centered by choose the length of side be the square region of r as window, reuse first moment centroid algorithm and calculate barycenter in this window, obtain center-of-mass coordinate x
1, y
1; Window size is slightly larger than spot size;
3) g=|x is defined
1-x
0| be the barycenter departure in x direction, l=|y
1-y
0| be the barycenter departure in y direction;
4) judge whether barycenter departure g and l is less than threshold value T, T is 0.1pixel, if be less than, by x
1, y
1as
The center-of-mass coordinate of hot spot; Otherwise, make x
0=x
1, y
0=y
1, proceed to step 2).
2. the moving window first moment centroid computing method adding thresholding according to claim 1, is characterized in that, the calculation expression of described first moment centroid algorithm discrete form is
Wherein I
ijfor the gray scale light intensity at (i, j) coordinate place, x
i, y
jfor the coordinate of the x at (i, j) place, y, L, M are the row and column of image respectively.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510167617.7A CN105160692A (en) | 2015-04-09 | 2015-04-09 | First-moment center of mass calculating method for sliding window with threshold |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510167617.7A CN105160692A (en) | 2015-04-09 | 2015-04-09 | First-moment center of mass calculating method for sliding window with threshold |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105160692A true CN105160692A (en) | 2015-12-16 |
Family
ID=54801534
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510167617.7A Pending CN105160692A (en) | 2015-04-09 | 2015-04-09 | First-moment center of mass calculating method for sliding window with threshold |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105160692A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017113146A1 (en) * | 2015-12-28 | 2017-07-06 | 苏州中启维盛机器人科技有限公司 | Speckle imaging device |
CN108470351A (en) * | 2018-02-01 | 2018-08-31 | 汕头大学 | It is a kind of to track the method, apparatus and storage medium for measuring offset using image patch |
CN112581374A (en) * | 2019-09-29 | 2021-03-30 | 深圳市光鉴科技有限公司 | Speckle sub-pixel center extraction method, system, device and medium |
CN113706468A (en) * | 2021-07-27 | 2021-11-26 | 河北光兴半导体技术有限公司 | Glass defect detection method based on BP neural network |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070007436A1 (en) * | 2005-07-07 | 2007-01-11 | John Maksymowicz | Electro-optical focal plane array digital sensor system |
CN102081738A (en) * | 2011-01-06 | 2011-06-01 | 西北工业大学 | Method for positioning mass center of spatial object star image |
CN104316049A (en) * | 2014-10-28 | 2015-01-28 | 中国科学院长春光学精密机械与物理研究所 | High-precision and low-signal-to-noise-ratio elliptic star spot subdivision location method |
-
2015
- 2015-04-09 CN CN201510167617.7A patent/CN105160692A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070007436A1 (en) * | 2005-07-07 | 2007-01-11 | John Maksymowicz | Electro-optical focal plane array digital sensor system |
CN102081738A (en) * | 2011-01-06 | 2011-06-01 | 西北工业大学 | Method for positioning mass center of spatial object star image |
CN104316049A (en) * | 2014-10-28 | 2015-01-28 | 中国科学院长春光学精密机械与物理研究所 | High-precision and low-signal-to-noise-ratio elliptic star spot subdivision location method |
Non-Patent Citations (1)
Title |
---|
张艳艳等: "加门限的一阶矩光斑质心探测方法", 《光学技术》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017113146A1 (en) * | 2015-12-28 | 2017-07-06 | 苏州中启维盛机器人科技有限公司 | Speckle imaging device |
CN108470351A (en) * | 2018-02-01 | 2018-08-31 | 汕头大学 | It is a kind of to track the method, apparatus and storage medium for measuring offset using image patch |
CN112581374A (en) * | 2019-09-29 | 2021-03-30 | 深圳市光鉴科技有限公司 | Speckle sub-pixel center extraction method, system, device and medium |
CN113706468A (en) * | 2021-07-27 | 2021-11-26 | 河北光兴半导体技术有限公司 | Glass defect detection method based on BP neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Kraus et al. | Uncertainty estimation in one-stage object detection | |
CN105160692A (en) | First-moment center of mass calculating method for sliding window with threshold | |
CN101281648A (en) | Method for tracking dimension self-adaption video target with low complex degree | |
CN104636118A (en) | QR two-dimensional code self-adaptation binarization processing method and device based on light balance | |
CN105894521A (en) | Sub-pixel edge detection method based on Gaussian fitting | |
US11263774B2 (en) | Three-dimensional position estimation device and program | |
CN103617636A (en) | Automatic video-target detecting and tracking method based on motion information and sparse projection | |
CN104915928A (en) | Proper orthogonal decomposition-based velocity field bad vector identification and correction method | |
CN104766079A (en) | Remote infrared weak object detecting method | |
CN101526480B (en) | Real-time detection method of butt weld of thin plates based on visual sense | |
Lee | Pointing accuracy improvement using model-based noise reduction method | |
Rollason et al. | Particle filter for track‐before‐detect of a target with unknown amplitude viewed against a structured scene | |
CN103411562B (en) | A kind of structured light strip center extraction method based on dynamic programming and average drifting | |
CN114202473A (en) | Image restoration method and device based on multi-scale features and attention mechanism | |
CN112308917A (en) | Vision-based mobile robot positioning method | |
McManamon et al. | Exomars rover vehicle perception system architecture and test results | |
CN104331087A (en) | Robust underwater sensor network target tracking method | |
EP4174522A1 (en) | System and method for training a neural network to perform object detection using lidar sensors and radar sensors | |
Xi et al. | Research on the algorithm of noisy laser stripe center extraction | |
CN104236555A (en) | Pulsar timing noise estimation and forecasting method | |
CN113496083B (en) | GPS mobile station vertical speed field optimization method and device | |
CN116416227A (en) | Background image processing method and device | |
CN112987027B (en) | Positioning method of AMCL algorithm based on Gaussian model and storage medium | |
CN108009459A (en) | Character two-dimensional bar code method for rapidly positioning based on triangle polyester fibre symbol | |
CN114895298A (en) | Method and device for measuring and correcting Bernoulli filtering of radar slow-speed weak maneuvering target |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151216 |
|
RJ01 | Rejection of invention patent application after publication |