CN101520891B - Starry sky image object track-detecting method - Google Patents
Starry sky image object track-detecting method Download PDFInfo
- Publication number
- CN101520891B CN101520891B CN2009100215657A CN200910021565A CN101520891B CN 101520891 B CN101520891 B CN 101520891B CN 2009100215657 A CN2009100215657 A CN 2009100215657A CN 200910021565 A CN200910021565 A CN 200910021565A CN 101520891 B CN101520891 B CN 101520891B
- Authority
- CN
- China
- Prior art keywords
- image
- piece
- theta
- value
- fritter
- 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
Abstract
The invention discloses a starry sky image object track-detecting method. The method comprise the following steps: firstly, carrying out frame difference accumulation on five continuous images among sequence images and adopting a self-adaptive threshold value t to carry out binarization processing of 0 and 1 on the accumulated images; evenly dividing a 1,024*1,024 image I into 32*32 small blocks; combining and connecting all pixels on the image blocks one by one, and calculating theta and rho to establish a nonuniform quantizing accumulator array H(rho, theta) of a parameter space; moreover, adding 1 to the address of a corresponding accumulator array H(rho, theta), namely H(rho, theta) is equal to H(rho, theta) plus 1, till the conversion of all points is finished; and finding the local maximum value of the accumulator array H(rho, theta) in the parameter space and finding out the values of linear parameters theta andrRho to determine a straight line. By dividing an object track into smaller blocks and adopting nonuniformization to improve the prior Hough conversion, the starry sky image object track-detecting method realizes the detection of small objects in starry sky and reduces calculation amount and memory space.
Description
Technical field
The present invention relates to small and weak moving-target movement locus detection method in a kind of astronomical image processing method, particularly starry sky background image.
Background technology
Imageable target makes target less at the area of image, and does not have shape and structure information and texture information owing to far apart from imaging device under the starry sky background, because detecting devices self influences, the signal to noise ratio (S/N ratio) of image is lower simultaneously.Therefore carrying out moving object detection under the condition of low signal-to-noise ratio has important Research Significance.
Document " a kind of detection and tracking of rapid serial image low signal-to-noise ratio point target; Xian Electronics Science and Technology University's journal; 1999.12; 732-736 " discloses a kind of based on multistage hypothesis space track search method, the method is by means of the thinking of sequential classification, by the gray scale of potential target and the gray scale of its possible position in the neighborhood of next frame are superposeed, and judge whether it is target by sequential threshold value.This method is being set under verification and measurement ratio and the false alarm rate, and it is relatively good to detect effect, but owing to need carry out the search in space, calculated amount and memory space are big, and real-time is poor, and generally require target speed less than 1 pixel, otherwise it is big that the hunting zone becomes, and calculated amount is exponential rising.
Summary of the invention
In order to overcome the big deficiency of prior art calculated amount and memory space; the invention provides a kind of starry sky image object track-detecting method; by target trajectory being divided into less sub-piece; adopt non-homogeneousization that traditional Hough conversion is improved; can realize that the Weak target under the starry sky background detects, and reduces calculated amount and memory space.
The technical solution adopted for the present invention to solve the technical problems: a kind of starry sky image object track-detecting method is characterized in comprising the steps:
(a) N continuous in the sequence image is opened image and carry out the frame difference and add up, and adopt adaptive threshold t that the image after adding up is carried out 0 and 1 binary conversion treatment, N=5;
(b) image I of 1024 * 1024 sizes evenly is divided into 32 * 32 fritter, and is 32 * 32 fritter B each size
I, jIntermediate value is 1 counts and add up;
I is the horizontal ordinate of fritter in image, 1≤i≤32, and j is the ordinate of fritter in image, 1≤j≤32;
(c) value according to statistics is 1 counts, and finds out fritter B
I, jAnd the interior (B of eight neighborhoods
I-1, j-1, B
I-1, j, B
I-1, j+1, B
I, j-1, B
I, j+1, B
I+1, j-1, B
I+1, j, B
I+1, j+1) the maximum value piece, according to maximum value and specified threshold value t_num_ones relatively, tentatively judge whether there is target trajectory in the neighborhood piece, if more than or equal to threshold value, then may have target trajectory, if less than threshold value, then do not have target trajectory; T_num_ones=20;
(d),, calculate three pieces, i.e. B being positioned at place, four right angles in its eight neighborhood respectively to the local maximum piece to there being the fritter of target trajectory
I-1, j-1+ B
I-1, j+ B
I, j-1, piece B
I, j-1+ B
I+1, j-1+ B
I+1, j, piece B
I-1, j+ B
I-1, j+1+ B
I, j+1With piece B
I, j+1+ B
I+1, j+ B
I+1, j+1Value is 1 counts, and finds out the maximum piece of wherein counting, and and B
I, jMerge, form 64 * 64 image block;
(e) establish that space line section starting point coordinate is (x on the image block
0, y
0), terminal point coordinate is (x
1, y
1), and x
0≠ x
1, according to
k=(y
1-y
0)/(x
1-x
0) (1)
θ=arctan(k) (2)
ρ=(y
0-kx
0)cosθ (3)
All pixels on the image block are made up line one by one and calculate θ and ρ, respectively the parameter space horizontal direction is carried out the quantification of angle and the quantification of vertical direction distance according to result of calculation, set up parameter space non-uniform quantizing totalizer array H (ρ, θ);
(f) with totalizer array H (ρ, θ) zero setting, to choose in the step (c) and B
I, j64 * 64 the piece intermediate value that merges that the back forms is that 1 point carries out conversion, finds these θ and ρ values in the parameter space correspondence, and with corresponding accumulator array H (ρ, position θ) adds 1, promptly H (ρ, θ)=H (ρ, θ)+1, finish up to all equal conversion of point;
(g) (ρ, local maximum θ) find straight line parameter θ and ρ value, determine straight line to seek on the parameter space totalizer array H.
The invention has the beneficial effects as follows: owing to, adopt non-homogeneousization that traditional Hough conversion is improved, realized that the Weak target under the starry sky background detects, reduced calculated amount and memory space by target trajectory being divided into less sub-piece.
Below in conjunction with drawings and Examples the present invention is elaborated.
Description of drawings
Accompanying drawing is that the inventive method piecemeal merges synoptic diagram.
Embodiment
Accumulated image piecemeal and preliminary judgement.
Target trajectory length depends on the movement velocity of number of image frames and target in the sequence frame difference accumulated image under the starry sky background.Usually the target spot size is generally about 4 * 4 pixels, and its movement velocity is uncertain.If the tangential movement speed and the vertical motion speed of target are 1 pixel/frame on the hypothesis image, then the formed course length of hot spot is about 25~36 pixels, and the starry sky background image is generally 1024 * 1024, and with respect to the image size, the track of target is very short.Exist a large amount of random noises again on the image simultaneously, its number and gray scale all surpass target.If directly the frame difference accumulated image under the starry sky background is carried out the Hough conversion, is difficult to find target trajectory.
The present invention is divided into target trajectory in the less sub-piece and detects, and has solved because the problem that the track detection that target trajectory existence too short with respect to entire image and much noise causes lost efficacy.
At first the N continuous in the sequence image is opened image and carry out the frame difference and add up, and adopt adaptive threshold t the image after adding up to be carried out 0 and 1 binary conversion treatment.N=5。
Secondly, the image I of 1024 * 1024 sizes evenly is divided into 32 * 32 fritter, adds up to 32 * 32, and be 32 * 32 fritter B each size
I, jIntermediate value is 1 counts and add up.
I is the horizontal ordinate of fritter in image, 1≤i≤32, and j is the ordinate of fritter in image, 1≤j≤32.
Then, be 1 count according to the value of statistics, find out fritter B
I, jAnd the interior (B of eight neighborhoods
I-1, j-1, B
I-1, j, B
I-1, j+1, B
I, j-1, B
I, j+1, B
I+1, j-1, B
I+1, j, B
I+1, j+1) the maximum value piece, according to maximum value and specified threshold value t_num_ones (t_num_ones=20) relatively, tentatively judge whether there is target trajectory in the neighborhood piece, if more than or equal to threshold value, then may there be target trajectory,, then do not have target trajectory if less than threshold value;
At last, to there being the fritter of target trajectory, promptly maximum value then to the local maximum piece, is calculated three pieces, i.e. B being positioned at place, four right angles in its eight neighborhood greater than threshold value t_num_ones (t_num_ones=20) respectively
I-1, j-1+ B
I-1, j+ B
I, j-1, piece B
I, j-1+ B
I+1, j-1+ B
I+1, j, piece B
I-1, j+ B
I-1, j+1+ B
I, j+1With piece B
I, j+1+ B
I+1, j+ B
I+1, j+1Value is 1 counts, and finds out the maximum piece of wherein counting, and and B
I, jMerge, form 64 * 64 image block, size is the process object that 64 * 64 image blocks are follow-up track detection after merging, and sees accompanying drawing.
Target trajectory detects.
What tradition Hough conversion was adopted is the uniform quantization of parameter space, promptly in the parameter space horizontal direction angle θ is carried out uniform quantization in [0, π] interval, and ρ is at [ρ for the vertical direction pair pitch parameters
Min, ρ
Max] carrying out uniform quantization in (concrete numerical value relevant) interval with image expression, Δ θ and Δ ρ are constant during the parameter uniform quantization;
Wherein when image carries out coordinate representation, initial point is placed the center of image, rather than the upper left corner of image or the lower left corner, the line segment inclination alpha, α ∈ [pi/2, pi/2), suppose that picture traverse is W, highly be H, being centered close to of image
The horizontal ordinate leftmost side and the rightmost side are respectively
Ordinate lower side and top side are respectively
When the width of image with when highly enough big, x
0+ 1 and x
0Between difference just very little, can all use x
0Expression x
+, in like manner can use y
0Expression y
+, as seen
The parameter space uniform quantization causes peak value occurring in the adjacent angle of parameter space totalizer, cause peak value fuzzy, and non-uniform quantizing is exactly to seeking suitable Δ θ and Δ ρ, to avoid the peak value obfuscation.
The present invention adopts non-homogeneousization that traditional Hough conversion is improved, and concrete steps are as follows:
If space line section starting point coordinate is (x on the image
0, y
0), terminal point coordinate is (x
1, y
1), and x
0≠ x
1, then
k=(y
1-y
0)/(x
1-x
0) (4)
θ=arctan(k) (5)
ρ=(y
0-kx
0)cosθ (6)
According to formula (1) (2) (3), all pixels on the image are made up line one by one and calculate θ and ρ, θ and ρ according to the result of calculation gained, respectively the parameter space horizontal direction is carried out the quantification of angle and the quantification of vertical direction distance, thereby set up parameter space non-uniform quantizing totalizer array H (ρ, θ);
With totalizer array H (ρ, θ) zero setting, to choose in the step 1 and B
I, j64 * 64 the piece intermediate value that merges that the back forms is that 1 point carries out conversion, finds these θ and ρ values in the parameter space correspondence, and with corresponding accumulator array H (ρ, position θ) adds 1, promptly H (ρ, θ)=H (ρ, θ)+1, finish up to all equal conversion of point;
Totalizer array H on the searching parameter space (ρ, local maximum θ), thus find straight line parameter θ and ρ value, and then definite straight line.
Claims (1)
1. a starry sky image object track-detecting method is characterized in that comprising the steps:
(a) N continuous in the sequence image is opened image and carry out the frame difference and add up, and adopt adaptive threshold t that the image after adding up is carried out 0 and 1 binary conversion treatment, N=5;
(b) image I of 1024 * 1024 sizes evenly is divided into 32 * 32 fritter, and is 32 * 32 fritter B each size
I, jIntermediate value is 1 counts and add up;
I is the horizontal ordinate of fritter in image, 1≤i≤32, and j is the ordinate of fritter in image, 1≤j≤32;
(c) value according to statistics is 1 counts, and finds out fritter B
I, jAnd the interior (B of eight neighborhoods
I-1, j-1, B
I-1, j, B
I-1, j+1, B
I, j-1, B
I, j+1, B
I+1, j-1, B
I+1, j, B
I+1, j+1) the maximum value piece, according to maximum value and specified threshold value t_num_ones relatively, tentatively judge whether there is target trajectory in the neighborhood piece, if more than or equal to threshold value t_num_ones, then may there be target trajectory,, then do not have target trajectory if less than threshold value t_num_ones; T_num_ones=20;
(d),, calculate three pieces, i.e. B being positioned at place, four right angles in its eight neighborhood respectively to the local maximum piece to there being the fritter of target trajectory
I-1, j-1+ B
I-1, j+ B
I, j-1, piece B
I, j-1+ B
I+1, j-1+ B
I+1, j, piece B
I-1, j+ B
I-1, j+1+ B
I, j+1With piece B
I, j+1+ B
I+1, j+ B
I+1, j+1Value is 1 counts, and finds out the maximum piece of wherein counting, and and B
I, jMerge, form 64 * 64 image block;
(e) establish that space line section starting point coordinate is (x on the image block
0, y
0), terminal point coordinate is (x
1, y
1), and x
0≠ x
1, according to
k=(y
1-y
0)/(x
1-x
0) (1)
θ=arctan(k) (2)
ρ=(y
0-kx
0)cosθ (3)
All pixels on the image block are made up line one by one and calculate θ and ρ, respectively the parameter space horizontal direction is carried out the quantification of angle and the quantification of vertical direction distance according to result of calculation, set up parameter space non-uniform quantizing totalizer array H (ρ, θ);
(f) with totalizer array H (ρ, θ) zero setting, to choose in the step (d) and B
I, j64 * 64 the piece intermediate value that merges that the back forms is that 1 point carries out conversion, finds these θ and ρ values in the parameter space correspondence, and with corresponding totalizer array H (ρ, position θ) adds 1, promptly H (ρ, θ)=H (ρ, θ)+1, finish up to all equal conversion of point;
(g) (ρ, local maximum θ) find straight line parameter θ and ρ value, determine straight line to seek on the parameter space totalizer array H.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009100215657A CN101520891B (en) | 2009-03-17 | 2009-03-17 | Starry sky image object track-detecting method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009100215657A CN101520891B (en) | 2009-03-17 | 2009-03-17 | Starry sky image object track-detecting method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101520891A CN101520891A (en) | 2009-09-02 |
CN101520891B true CN101520891B (en) | 2011-08-17 |
Family
ID=41081464
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2009100215657A Expired - Fee Related CN101520891B (en) | 2009-03-17 | 2009-03-17 | Starry sky image object track-detecting method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101520891B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102116626B (en) * | 2009-12-31 | 2012-05-16 | 北京控制工程研究所 | Prediction and correction method of node of star point track image |
CN102081793A (en) * | 2011-01-06 | 2011-06-01 | 西北工业大学 | Method for eliminating smear effect bright line of starry sky background image frame transfer type CCD sensor |
CN102081800B (en) * | 2011-01-06 | 2012-07-25 | 西北工业大学 | Method for detecting spatial weak moving target |
CN103472256B (en) * | 2013-09-25 | 2015-09-16 | 东南大学 | Based on flowing two-dimension speed field measurement method and the device of area array CCD spatial filter |
CN105761279B (en) * | 2016-02-18 | 2019-05-24 | 西北工业大学 | Divide the method for tracking target with splicing based on track |
CN106023148B (en) * | 2016-05-06 | 2018-01-12 | 北京航空航天大学 | A kind of sequence focuses on star image point position extracting method under observation mode |
CN111353991A (en) * | 2020-03-10 | 2020-06-30 | 北京市商汤科技开发有限公司 | Target detection method and device, electronic equipment and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1570957A (en) * | 2004-05-13 | 2005-01-26 | 上海交通大学 | Method for detecting infrared weak and small object under antiaerial imaging fluctuation background |
US20070047822A1 (en) * | 2005-08-31 | 2007-03-01 | Fuji Photo Film Co., Ltd. | Learning method for classifiers, apparatus, and program for discriminating targets |
CN101241599A (en) * | 2008-02-28 | 2008-08-13 | 上海交通大学 | Row based weak target detection method in infra-red ray row detector image-forming |
-
2009
- 2009-03-17 CN CN2009100215657A patent/CN101520891B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1570957A (en) * | 2004-05-13 | 2005-01-26 | 上海交通大学 | Method for detecting infrared weak and small object under antiaerial imaging fluctuation background |
US20070047822A1 (en) * | 2005-08-31 | 2007-03-01 | Fuji Photo Film Co., Ltd. | Learning method for classifiers, apparatus, and program for discriminating targets |
CN101241599A (en) * | 2008-02-28 | 2008-08-13 | 上海交通大学 | Row based weak target detection method in infra-red ray row detector image-forming |
Also Published As
Publication number | Publication date |
---|---|
CN101520891A (en) | 2009-09-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101520891B (en) | Starry sky image object track-detecting method | |
He et al. | An improved Canny edge detector and its realization on FPGA | |
CN109559324B (en) | Target contour detection method in linear array image | |
CN102521589B (en) | Method and system for detecting lane marked lines | |
US20170132803A1 (en) | Apparatus and method for processing a depth image | |
CN104835166B (en) | Liquid medicine bottle foreign matter detecting method based on machine vision detection platform | |
CN101241223B (en) | Automatic focusing process | |
CN106488115B (en) | For tracking and sensing the control device and method of the image sensing apparatus of tracking target | |
CN112446838A (en) | Image noise detection method and device based on local statistical information | |
CN108022249A (en) | A kind of remote sensing video satellite moving vehicle target region of interest extraction method | |
CN104537342B (en) | A kind of express lane line detecting method of combination ridge border detection and Hough transformation | |
CN109001757A (en) | A kind of parking space intelligent detection method based on 2D laser radar | |
CN116309565B (en) | High-strength conveyor belt deviation detection method based on computer vision | |
CN105654428A (en) | Method and system for image noise reduction | |
US20040057601A1 (en) | Method for image processing | |
US9600700B1 (en) | Portable electronic device and operation method thereof | |
CN104766079A (en) | Remote infrared weak object detecting method | |
CN104613869A (en) | Method and system for detecting elliptical hole group based on maximum inscribed circle | |
CN105300390B (en) | The determination method and device of obstructing objects movement locus | |
CN109754034A (en) | A kind of terminal device localization method and device based on two dimensional code | |
CN114882041B (en) | Voting and identification method for burrs inside cast iron pipeline | |
CN110807354B (en) | Industrial assembly line product counting method | |
CN112184644A (en) | Air tightness bubble detection method and device based on multiple illumination intensities | |
CN106067031A (en) | Cooperate with the degree of depth learning network Machine Vision Recognition system based on artificial mechanism for correcting errors | |
CN102004585B (en) | Multi-area identification method for touch screen |
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 | ||
C17 | Cessation of patent right | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20110817 Termination date: 20140317 |