CN104143196B - A kind of point target detection method based on many alignment moveout scans extension sampling - Google Patents
A kind of point target detection method based on many alignment moveout scans extension sampling Download PDFInfo
- Publication number
- CN104143196B CN104143196B CN201410319201.8A CN201410319201A CN104143196B CN 104143196 B CN104143196 B CN 104143196B CN 201410319201 A CN201410319201 A CN 201410319201A CN 104143196 B CN104143196 B CN 104143196B
- Authority
- CN
- China
- Prior art keywords
- image
- detector array
- pixel
- detection
- subgraph
- 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.)
- Active
Links
Abstract
A kind of point target detection method based on many alignment moveout scans extension sampling, (1) construction two-wire row detector image-forming device;(2) each detector array carries out point target detection using extension sample mode;(3) N for respectively being gathered each detector arraytGroup view data is processed, and obtains subpixel image;Two width sub-pixed mappings after two detector array treatment carry out Difference Calculation after carrying out Nonuniformity Correction, sub-pix matching, complete background and eliminate;(4) residual image is obtained to difference in step (3) carries out threshold filter, and right using neighborhood constraint criterion extraction positive and negative point therein, and moving spot targets detection is extracted during completing single pass;The time difference is imaged according to different detector array and target speed formation neighborhood constraint criterion isWherein positive and negative region distance Δ D, the target minimum movement speed v of detectionmin, the target maximum movement velocity v of detectionmax, the ground sampled distance GSD of detector array.
Description
Technical field
The invention belongs to target acquisition process field, it is related to a kind of point target based on many alignment moveout scans extension sampling
Detection method.
Background technology
The research of small dim moving target Detection Techniques has important application in civilian, space flight and military affairs under complex background.By
The influence of remote and complex background in image-forming range, target into point-like, lacks enough texture information and shape informations in the picture,
And the target signal to noise ratio in image is very low, this considerably increases the difficulty of dim targets detection.Optical imagery general at present
Detection system typically carries out detection imaging using single detector, to acquired image, using method, base based on filtering
In the method for small echo, based on morphologic method etc., single frames background suppression is carried out, suspected target is carried out on single-frame images and is carried
Take, the mode for being associated using multiframe afterwards carries out target trajectory fitting, so as to realize target detection.This Detecting System is often by multiple
The influence of miscellaneous background, it is impossible to reliably Objective extraction is carried out on single-frame images, while this Detecting System is in Objective extraction mistake
The movable information of target is not made full use of in journey, therefore often there is false alarm rate higher, needed to improve detection accuracy
Using complicated algorithm of target detection, data-handling efficiency is reduced.
The content of the invention
Technology solve problem of the invention is:Overcome the deficiencies in the prior art, propose a kind of based on many alignment moveout scans
The point target detection method of sampling is extended, realizes that moving target quick detection is detected, improve moving target under complex background environment
The ability of detection.
Technical solution of the invention is:A kind of point target detection side based on many alignment moveout scans extension sampling
Method, step is as follows:
(1) many alignment moveout scan detection devices are constructed, the device includes the detection of optical system, sweep mechanism and many alignments
Device;Described sweep mechanism includes pendulum mirror and its drive shaft;Described many detector array include NdIndividual detector array, often
Individual detector array is arranged in parallel, and the distance between two neighboring detector array is di, wherein i=1,2 ..., Nd-1;Optics
Scene in visual field is imaged in focal plane by system and sweep mechanism together, and drive shaft drives pendulum mirror rotation, the field in linear field
With many detector array on certain inswept focal plane of speed in tandem, angular scanning rate is ω to scape imaging;Multiple lines
Row detector is successively imaged to same position scene in visual field, any two alignment imaging time intervalArbitrarily
Two detectable movement velocitys of detector array are more than vTTarget, whereinWherein Δ d is any two
Distance between detector array, f is optical system focal length, and GSD is the ground sampled distance of detector array;The NdValue model
It is N to enclosed≥2;
(2) each detector array carries out point target detection, i.e. detector array and uses N using extension sample modetIt is individual
Detection array is constituted, and the corresponding instantaneous field of view of pixel is IFOV, and two neighboring detection array is arranged in parallel, in vertical scanning direction
Stagger 1/N successivelytIndividual pixel, and set detection array in scanning direction, sample S in a sampling lengthtIt is secondary;The sampling
Length is the corresponding instantaneous field of view of pixel;Described NtMore than or equal to 2;The StSpan is St≥2;
(3) N for respectively being gathered each detector arraytGroup view data is processed, and obtains subpixel image;Choosing
Taking two width sub-pixed mappings after two detector array treatment carries out carrying out Difference Calculation after Nonuniformity Correction, sub-pix are matched,
Background is completed to eliminate;
(4) residual image is obtained to difference in step (3) carries out threshold filter, and is extracted wherein using neighborhood constraint criterion
Positive and negative point it is right, complete single pass during moving spot targets extract;The time difference and target are imaged according to different detector array
Movement velocity forms neighborhood constraint criterionWherein positive and negative region distance Δ D, detection
Target minimum movement speed vmin, the target maximum movement velocity v of detectionmax。
The N for being gathered each detector array in the step (3)tGroup view data is processed, and obtains sub- picture
The mode of first image, detailed process is as follows:
(3.1) N in each detector arraytIndividual detection array is imaged simultaneously, obtains NtGroup view data;Immediately afterwards
It is transferred to step (3.2);Simultaneously, each detector array is according to the setting sampling number corresponding sampling interval in step (2)
Imaging is scanned in scanning direction, imaging every time respectively obtains NtGroup view data, step is transferred to after obtaining data immediately
(3.2);
(3.2) respectively by the N of each detector arraytGroup view data forms frame detection after carrying out alignment splicing
Image;
(3.3) after completing default sampling number in a scanning direction, the frame that each detector array correspondence is obtained is visited
Altimetric image according to the time splice and obtains corresponding subpixel image.
Before the two width subpixel images in the step (3) carry out sub-pix matching, visited with the alignment of first imaging
Survey on the basis of the corresponding subpixel image of device, by the corresponding subpixel image of the detector array of rear imaging in scanning direction forward
It is mobileOK, LΔdSeveral rows are rounded, detector array array pixel dimension is a × a;On this basis again using Asia
The quasi- method of pixel matching is matched.
Filtering in the step (4) is by the way of threshold filter, and detailed process is as follows:
(4.1) each pixel in the difference image obtained after processing step (3) carries out enhancing treatment, obtains image
Isub1;
(4.2) image Isub1Take absolute value, be designated as image Isub2;
(4.3) by Isub2Adjacent several image columns are divided into one group, are divided into NsubIndividual subgraph;The columns L of subgraphsub=
H/Nsub, H is image Isub2Total columns;Simultaneously to image Isub1Using with image Isub2Identical mode is grouped, image
Isub1With Isub2Subgraph correspond;
(4.4) from Isub2The first row of current subgraph is designated as current line and starts, and setting one threshold value initial value T, T take first
The pixel value of individual pixel, or the value more slightly larger than the pixel of first pixel;
(4.5) by the L in current subgraph current linesubThe pixel value of individual pixel is contrasted with threshold value initial value T-phase successively, such as
Fruit pixel value exceedes threshold value initial value T, then improve certain amplitude to T according to formula (a):
T '=T+0.5k(Vi-T) (a)
T=T '
Wherein k is coefficient, is adjusted according to signal noise ratio (snr) of image, ViRepresent i-th pixel of current subgraph current line
Pixel value;
After last pixel treatment of current line, the final filtering threshold T of current subgraph current line r is obtainedr;
(4.6) final threshold value T obtained above is utilizedrTo Isub1The corresponding row of middle correspondence subgraph carries out threshold filter, i.e.,
Pixel in same a line image, will be greater than filtering threshold is labeled as 1;If pixel value is negative, will be less than filtering threshold
The pixel of negative value is labeled as -1;Remaining pixel marks and is;
(4.7) the threshold value T for finally giving step (4.5)rRatio reduction is carried out in formula (b), as current subgraph
The threshold value initial value of next line:
T=(1-0.5m)Tr (b)
Wherein m is proportionality coefficient;
Using current subgraph next line as current line, performed since step (4.5), until all rows of current subgraph
It is disposed;
(4.8) next subgraph is selected as current subgraph, repeat step (4.4)~(4.7), until completing to image
Isub1The threshold filter treatment of all rows of subgraph, is designated as image I 'lab;
(4.9) I ' is traveled throughlabEach row, if continuous more than 2 pixels are denoted as 1, it is constant to retain former mark, otherwise
The pixel is labeled as 0;If continuous less than 2 pixels are marked as -1, retain former mark, will otherwise visit pixel and be labeled as
0;Complete I 'labAfter all pixels traversal, image I is designated aslab。
Enhancing treatment in the step (4.1), i.e., the picture of the pixel value=pending pixel after pending pixel treatment
The pixel value of the neighborhood pixels of element value * α+four, α is enhancing coefficient.
Present invention advantage compared with prior art is:
1st, using New Moving Target Detecting System and data processing technique, Point Target is realized by extending Sampling techniques
The two-dimensional expansion of imaging, and obtain the detection image pair with the time difference using many alignment time difference detectors;By follow-up figure
As sub-pixel matching and Difference Calculation, the suppression of complex background is realized;Echo signal enhancing is carried out using signal enhancing method,
And then by detection method of small target, realize the extraction of moving target.
2nd, the motion feature of moving target is made full use of, according to moving target on two width subpixel image difference images
Positive and negative marked region differentiates extraction to carrying out moving target, can effectively suppress false-alarm, improves target detection accuracy, improves multiple
The ability of moving object detection under miscellaneous background.
3rd, according to the change of detecting movements of objects speed, sweep speed can be adjusted, and combine many alignment arrangement pitches
Difference, arbitrarily image formed by two detector array of selection are processed and Difference Calculation, are made on correspondence difference image just
Negative point realizes that high-speed target is detected to meeting certain neighborhood constraints using two small alignments of arrangement pitch, utilizes
Two big alignments of arrangement pitch realize that slower-velocity target is detected, and thus can realize that friction speed moving target is detected.
4th, by the way of sweep mechanism carries out active scan imaging, it is possible to achieve on a large scale, the mesh in arbitrary motion direction
Mark detection.
5th, when two width subpixel images are matched, directly can be matched using conventional sub-pix matching process, but matching
Efficiency and precision are relatively low;The present invention makes full use of detector array to arrange feature and the imaging mode for setting, with first imaging
On the basis of the corresponding subpixel image of detector array, by the corresponding subpixel image of the detector array of rear imaging in scanning side
To forward movement LΔdOK, two images rapid registering is carried out, is carried out using sub-pix method for registering again on this basis
Match somebody with somebody, reduce two images matching range, improve matching efficiency, reduce matching error, improve matching precision.
Brief description of the drawings
Fig. 1,2 are many alignment moveout scan detection device two ways schematic diagrames of the invention;
Fig. 3 is detector array two field picture splicing schematic diagram of the invention;
Fig. 4 is residual image column split schematic diagram of the invention;
Fig. 5 is neighborhood constraint criterion schematic diagram of the invention;
Specific embodiment
Below in conjunction with the accompanying drawings and example elaborates to the present invention.A kind of many alignment moveout scan moving target detection sides
Method, step is as follows:
(1) many alignment moveout scan detection devices are constructed, the device includes that optical system 1, sweep mechanism 2 and many alignments are visited
Survey device 3;Described sweep mechanism includes pendulum mirror and its drive shaft;Optical system and sweep mechanism together by scene in visual field into
As in focal plane, drive shaft drives pendulum mirror rotation, and the scene imaging in linear field is with certain inswept focal plane of speed
Many detector array in tandem, angular scanning rate ω;Described many detector array include NdIndividual detector array, each
Detector array is arranged in parallel, and the distance between two neighboring detector array is di, wherein i=1,2 ..., Nd-1;Multiple lines
Row detector is successively imaged to same position scene in visual field, any two alignment imaging time intervalArbitrarily
Two detectable movement velocitys of detector array are more than vTTarget, whereinWherein Δ d is any two
Distance between detector array, f is optical system focal length, and GSD is the ground sampled distance of detector array;
Each detector array carries out point target detection, i.e. detector array and uses N using extension sample modetIndividual detection
Array is constituted, and the corresponding instantaneous field of view of pixel is IFOV, and two neighboring detection array is arranged in parallel, in vertical scanning direction successively
Stagger 1/NtIndividual pixel, and set detection array in scanning direction, sample S in a sampling lengthtIt is secondary;The sampling length
It is the corresponding instantaneous field of view of pixel;Described NtMore than or equal to 2;The StSpan is St≥2;Below with NtEnter as a example by=2
Row explanation.
What Fig. 1 was given is that multi-thread column scan detection device is scanned in front end;Emittance information comprising target and background
Incident light converges to focal plane through putting after mirror reflects through optical system 1, forms the picture of scenery, and drive shaft drives pendulum mirror according to pre-
If angular speed rotation, make the picture of scenery inswept each detector array successively.When scenery picture with certain speed it is inswept its
In a detector array when, detector is sampled to the picture of scenery.What Fig. 2 was given is that the multi-thread column scan of rear-end scanning is visited
Survey device.Incident light comprising target and the emittance information of background converges to pendulum mirror through optical system 1, is reflexed to through putting mirror
Focal plane, forms the picture of scenery.Drive shaft drives pendulum mirror to be rotated according to default angular speed, makes the picture of scenery inswept each successively
Individual detector array.When the picture of scenery one of detector array inswept with certain speed, detector is to scenery
As being sampled.
(2) scene in visual field is imaged in focal plane by optical system 1 and sweep mechanism 2 together, and drive shaft drives pendulum mirror
Rotation, the scene imaging in linear field is with the two-wire row detector on certain inswept focal plane of speed in tandem, two
Detector array is successively imaged to same position scene in visual field, imaging time intervalThat is each alignment detection
N in devicetIndividual detection array is imaged simultaneously, obtains NtGroup view data;Step (3) is immediately entered after obtaining view data;In this
Meanwhile, each detector array is still according to setting in a sampling length sampling number in step (1), in scanning direction
Imaging is scanned, imaging every time respectively obtains NtGroup view data, step (3) is transferred to after being similarly obtained view data immediately;
For example, settable St>=2, then it is capable of achieving detector array more than 2 times over-samplings in a scanning direction;
(3) respectively by two N of detector arraytGroup view data forms a frame detection figure after carrying out alignment splicing
Picture, splicing are as shown in figure 3, by NtGroup pattern intersects splicing.Sub-pixed mapping frame detection image is obtained by splicing,
Realize stretching of the target in vertical scanning direction.
(4) after default sampling number, the frame detection image that each detector array correspondence is obtained is entered according to the time
Row splicing obtains two width subpixel images;Default desirable 200~300 rows of sampling number, increasing default sampling number can increase
Scan image details, improves the precision of successive image treatment;Reducing default sampling number can improve the efficiency of data processing, because
This default sampling number can be adjusted according to actual conditions.
(5) synthesize correcting mode using many alignment moveout scan images, heterogeneity school is carried out to two width subpixel images
Just, it is specific as follows:
(5.1) two width subpixel images are carried out into intersection splicing by row, forms the new stitching image I of a widthp1;
(5.2) to stitching image Ip1Each row image carry out Nonuniformity Correction, arranged after the completion of all column processing
To the image I after nonuniformity correctionp2;
(5.3) order of splicing is intersected according to the row of two images in step (5.1), from Ip2It is middle to extract corresponding row respectively
Image, rebuilds two width and completes subpixel image of the row to Nonuniformity Correction;
(5.4) row will be completed to be rotated by 90 ° respectively at same direction to two width subpixel images after nonuniformity correction, is repeated
Step (5.1)~(5.3), image I after being correctedp3;
(5.5) order of splicing is intersected according to the row of two images in step (5.1), from Ip3It is middle to extract corresponding row respectively
Image, rebuilds two width and completes subpixel image of the row to nonuniformity correction;
(5.6) row will be completed to be rotated by the opposite direction that step (5.4) rotates to two width subpixel images of nonuniformity correction
90 °, obtain completing two width subpixel images of row, column both direction nonuniformity correction;
(6) on the basis of the corresponding subpixel image of detector array of first imaging, by the detector array of rear imaging
Corresponding subpixel image is moved forward in scanning directionOK, LΔdRound several rows, detector array array picture
Elemental size is a × a;
(7) two width subpixel images after step (6) treatment are matched using sub-pix image registration algorithm, specifically
It is as follows:
(7.11) it is reference picture with the corresponding subpixel image of the detector array being first imaged, the alignment detection being imaged afterwards
The corresponding subpixel image of device is image subject to registration.
(7.12) gradient largest block is searched in reference picture;The definition of gradient largest block is:Calculate reference frame image
Gradient image, will be greater than the corresponding pixel logic value of gradient of image gradient maximum 4/5 labeled as 1, otherwise labeled as 0, shape
Into image be referred to as logical value image.The appropriately sized window of selection is corresponding containing 1 most windows in logical value image slide
Block is referred to as gradient largest block.Determine the position of gradient largest block by comparing the maximum in sliding window.Window size is built
It is vertical to use 50x50.
(7.13) it is, to picture, using based on the related image matching algorithm of gray scale, to enter to the gradient largest block in (7.12)
Row reference picture is matched with the sub-pix of image subject to registration.By reference picture entered row interpolation obtain more intensive grid come
For target image search, the registration accuracy of sub-pixel is obtained.
(7.14) registration parameter calculated according to (7.13), carries out image conversion, after obtaining registration to image subject to registration
Image.
(8) Difference Calculation is carried out to the two images after step (7) treatment, completes complex background self adaptation and eliminate, obtained
Residual image Isub1;
(9) the image I to being obtained after step (8) treatmentsub1In each pixel carry out enhancing treatment, i.e., pending pixel
The pixel value of the neighborhood pixels of pixel value * α+four of the pixel value=pending pixel after treatment, α typically takes 2~4, after treatment
To image Isub2;
(10) to image Isub2It is filtered, filtering herein can use several filtering modes conventional at present, for example
Iterative threshold segmentation algorithm, naturally it is also possible to use the Promethean threshold filter mode of the application, the method detailed process is as follows:
(10.1) image Isub2Take absolute value, be designated as image Isub3;
(10.2) by Isub3Adjacent several image columns (general 6-10 row) are divided into one group, are divided into NsubIndividual subgraph, such as schemes
Shown in 4.The columns L of subgraphsub=H/Nsub, H is image Isub3Total columns;By Isub2According to above-mentioned Isub3It is identical
Mode be grouped, be similarly obtained NsubIndividual subgraph, Isub2With Isub3In subgraph correspond;
(10.3) to Isub3In the first row (being designated as current line) of current subgraph start, set threshold value initial value a T, T
First pixel value of pixel, or the value more slightly larger than the pixel of first pixel can be taken;
(10.4) by the L in current subgraph current linesubThe pixel value of individual pixel is contrasted with threshold value initial value T-phase successively, such as
Fruit pixel value exceedes threshold value initial value T, then T is improved into certain amplitude, as shown in formula (a):
T '=T+0.5k(Vi-T) (a)
T=T '
Wherein k is usually taken to be 4, k values and can be adjusted according to signal noise ratio (snr) of image, and signal noise ratio (snr) of image is higher, can take k values
Small, signal noise ratio (snr) of image is relatively low, can take greatly k values, ViRepresent the pixel value of i-th pixel of current subgraph current line;
The T obtained after last pixel treatment of current line, the final filtering of as current subgraph current line r
Threshold value Tr;
(10.5) final threshold value T obtained above is utilizedrTo Isub2The corresponding row of middle correspondence subgraph carries out threshold filter,
1 is labeled as in same a line image, will be greater than the pixel of filtering threshold;If pixel value is negative, will be less than filtering threshold
The pixel for being worth negative value is labeled as -1;Remaining pixel marks and is;
(10.6) the threshold value T for finally giving step (10.4)rIn the reduction of a certain ratio, as current subgraph next line
Threshold value initial value, now T computing formula such as formula (b):
T=(1-0.5m)Tr (b)
Wherein m is usually taken to be 4, m values and can be adjusted according to signal noise ratio (snr) of image, and signal noise ratio (snr) of image is higher, can take m values
Greatly, signal noise ratio (snr) of image is relatively low, can take m values small.
Using current subgraph next line as current line, performed since step (10.4), until current subgraph owns
Row is disposed;
(10.7) next subgraph is selected as current subgraph, repeat step (10.3)~(10.6), until completing right
Image Isub2The threshold filter treatment of all rows of subgraph, is designated as image I 'lab;
(10.8) I ' is traveled throughlabEach row, if continuous more than 2 pixels are denoted as 1, it is constant to retain former mark, otherwise
The pixel is labeled as 0;If continuous less than 2 pixels are marked as -1, retain former mark, will otherwise visit pixel and be labeled as
0;Complete I 'labAfter all pixels traversal, image I is designated aslab;
(11) as shown in figure 5, extracting IlabIn paired positive and negative marked region, complete moving target during single pass
Detection is extracted;The neighborhood constraint criterion of positive negative region is that two positive and negative region distances are Δ D in pairs, general desirableTwo of which detector array trace interval Δ t, the target maximum fortune of detection
Dynamic speed vmin, the target maximum movement velocity v of detectionmax, the ground sampled distance GSD of detector array;
(12) moving target extracted according to step (11) carries out Track association analysis by follow-up multiple image, can be true
Determine the status informations such as the direction of motion, the movement locus of moving target.
Unspecified part of the present invention belongs to general knowledge as well known to those skilled in the art.
Claims (5)
1. it is a kind of that the point target detection method sampled is extended based on many alignment moveout scans, it is characterised in that step is as follows:
(1) many alignment moveout scan detection devices are constructed, the device includes optical system, sweep mechanism and many detector array;
Described sweep mechanism includes pendulum mirror and its drive shaft;Described many detector array include NdIndividual detector array, each
Detector array is arranged in parallel, and the distance between two neighboring detector array is di, wherein i=1,2 ..., Nd-1;Optical system
Scene in visual field is imaged in focal plane by system and sweep mechanism together, and drive shaft drives pendulum mirror rotation, the scene in linear field
With many detector array on certain inswept focal plane of speed in tandem, angular scanning rate is ω to imaging;Multiple alignments
Detector is successively imaged to same position scene in visual field, any two alignment imaging time intervalAny two
The individual detectable movement velocity of detector array is more than vTTarget, whereinWherein Δ d is any two line
Distance between row detector, f is optical system focal length, and GSD is the ground sampled distance of detector array;The NdSpan
It is Nd≥2;
(2) each detector array carries out point target detection, i.e. detector array and uses N using extension sample modetIndividual detection battle array
Row composition, the corresponding instantaneous field of view of pixel is IFOV, and two neighboring detection array is arranged in parallel, wrong successively in vertical scanning direction
Open 1/NtIndividual pixel, and set detection array in scanning direction, sample S in a sampling lengthtIt is secondary;The sampling length is
The corresponding instantaneous field of view of pixel;Described NtMore than or equal to 2;The StSpan is St≥2;
(3) N for respectively being gathered each detector arraytGroup view data is processed, and obtains subpixel image;Choose two
Two width sub-pixed mappings after individual detector array treatment carry out Difference Calculation after carrying out Nonuniformity Correction, sub-pix matching, complete
Background is eliminated;
(4) residual image is obtained to difference in step (3) carries out threshold filter, and therein just using the extraction of neighborhood constraint criterion
Negative point is right, and moving spot targets are extracted during completing single pass;The time difference and target motion are imaged according to different detector array
Speed forms neighborhood constraint criterionWherein positive and negative region distance Δ D, the mesh of detection
Mark minimum movement speed vmin, the target maximum movement velocity v of detectionmax。
2. a kind of point target detection method based on many alignment moveout scans extension sampling according to claim 1, it is special
Levy and be:The N for being gathered each detector array in the step (3)tGroup view data is processed, and obtains sub-pixed mapping
The mode of image, detailed process is as follows:
(3.1) N in each detector arraytIndividual detection array is imaged simultaneously, obtains NtGroup view data;It is transferred to immediately afterwards
Step (3.2);Simultaneously, each detector array is being swept according to the sampling number corresponding sampling interval is set in step (2)
Retouch direction and be scanned imaging, imaging every time respectively obtains NtGroup view data, step (3.2) is transferred to after obtaining data immediately;
(3.2) respectively by the N of each detector arraytGroup view data forms a frame detection image after carrying out alignment splicing;
(3.3) after completing default sampling number in a scanning direction, the frame detection figure that each detector array correspondence is obtained
Corresponding subpixel image is obtained as according to the time splice.
3. a kind of point target detection method based on many alignment moveout scans extension sampling according to claim 1, it is special
Levy and be:Before the two width subpixel images in the step (3) carry out sub-pix matching, detected with the alignment of first imaging
On the basis of the corresponding subpixel image of device, by the corresponding subpixel image of the detector array of rear imaging in scanning direction to reach
It is dynamicOK, LΔdSeveral rows are rounded, detector array array pixel dimension is a × a;On this basis again using sub- picture
A plain method for registering is matched.
4. a kind of point target detection method based on many alignment moveout scans extension sampling according to claim 1, it is special
Levy and be:Filtering in the step (4) is by the way of threshold filter, and detailed process is as follows:
(4.1) each pixel in the difference image obtained after processing step (3) carries out enhancing treatment, obtains image Isub1;
(4.2) image Isub1Take absolute value, be designated as image Isub2;
(4.3) by Isub2Adjacent several image columns are divided into one group, are divided into NsubIndividual subgraph;The columns L of subgraphsub=H/
Nsub, H is image Isub2Total columns;Simultaneously to image Isub1Using with image Isub2Identical mode is grouped, image
Isub1With Isub2Subgraph correspond;
(4.4) from Isub2The first row of current subgraph is designated as current line and starts, and setting one threshold value initial value T, T take first picture
The pixel value of element, or the value more slightly larger than the pixel of first pixel;
(4.5) by the L in current subgraph current linesubThe pixel value of individual pixel is contrasted with threshold value initial value T-phase successively, if picture
Element value exceedes threshold value initial value T, then improve certain amplitude to T according to formula (a):
T '=T+0.5k(Vn′-T) (a)
T=T '
Wherein k is coefficient, is adjusted according to signal noise ratio (snr) of image, Vn′Represent the picture of the n-th ' individual pixel of current subgraph current line
Element value;
After last pixel treatment of current line, the final filtering threshold T of current subgraph current line r is obtainedr;
(4.6) final threshold value T obtained above is utilizedrTo Isub1It is middle correspondence subgraph corresponding row carry out threshold filter, i.e., for
1 is labeled as with the pixel that in a line image, will be greater than filtering threshold;If pixel value is negative, will be less than filtering threshold negative value
Pixel be labeled as -1;Remaining pixel marks and is;
(4.7) the threshold value T for finally giving step (4.5)rRatio reduction is carried out in formula (b), as current subgraph next line
Threshold value initial value:
T=(1-0.5m)Tr (b)
Wherein m is proportionality coefficient;
Using current subgraph next line as current line, performed since step (4.5), until all row treatment of current subgraph
Finish;
(4.8) next subgraph is selected as current subgraph, repeat step (4.4)~(4.7), until completing to image Isub1
The threshold filter treatment of all rows of subgraph, is designated as image I 'lab;
(4.9) I ' is traveled throughlabEach row, if continuous more than 2 pixels are denoted as 1, it is constant to retain former mark, otherwise should
Pixel is labeled as 0;If continuous less than 2 pixels are marked as -1, retain former mark, the pixel is otherwise labeled as 0;It is complete
Into I 'labAfter all pixels traversal, image I is designated aslab。
5. a kind of point target detection method based on many alignment moveout scans extension sampling according to claim 4, it is special
Levy and be:Enhancing treatment in the step (4.1), i.e., the pixel of the pixel value=pending pixel after pending pixel treatment
The pixel value of the neighborhood pixels of value * α+four, α is enhancing coefficient.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410319201.8A CN104143196B (en) | 2014-07-04 | 2014-07-04 | A kind of point target detection method based on many alignment moveout scans extension sampling |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410319201.8A CN104143196B (en) | 2014-07-04 | 2014-07-04 | A kind of point target detection method based on many alignment moveout scans extension sampling |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104143196A CN104143196A (en) | 2014-11-12 |
CN104143196B true CN104143196B (en) | 2017-05-31 |
Family
ID=51852364
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410319201.8A Active CN104143196B (en) | 2014-07-04 | 2014-07-04 | A kind of point target detection method based on many alignment moveout scans extension sampling |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104143196B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107886538B (en) * | 2017-10-12 | 2020-06-09 | 北京空间机电研究所 | Star camera star point detection method and system |
CN112729554B (en) * | 2020-12-18 | 2022-07-29 | 中国航空工业集团公司洛阳电光设备研究所 | Self-discrimination variable-frequency sampling multi-sampling method applied to linear detector |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101706961A (en) * | 2009-11-10 | 2010-05-12 | 北京航空航天大学 | Image registration method and image registration device |
CN102663385A (en) * | 2012-03-07 | 2012-09-12 | 北京空间机电研究所 | Detection method for spot target on satellite |
CN103679766A (en) * | 2012-09-25 | 2014-03-26 | 中国航天科工集团第二研究院二〇七所 | A sub-pixel processing method decreasing boundary back substitution errors |
-
2014
- 2014-07-04 CN CN201410319201.8A patent/CN104143196B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101706961A (en) * | 2009-11-10 | 2010-05-12 | 北京航空航天大学 | Image registration method and image registration device |
CN102663385A (en) * | 2012-03-07 | 2012-09-12 | 北京空间机电研究所 | Detection method for spot target on satellite |
CN103679766A (en) * | 2012-09-25 | 2014-03-26 | 中国航天科工集团第二研究院二〇七所 | A sub-pixel processing method decreasing boundary back substitution errors |
Also Published As
Publication number | Publication date |
---|---|
CN104143196A (en) | 2014-11-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Dong et al. | 4D crop monitoring: Spatio-temporal reconstruction for agriculture | |
CN104484648B (en) | Robot variable visual angle obstacle detection method based on outline identification | |
CN109102522B (en) | Target tracking method and device | |
Yang et al. | Concrete defects inspection and 3D mapping using CityFlyer quadrotor robot | |
CN104574393B (en) | A kind of three-dimensional pavement crack pattern picture generates system and method | |
CN110223348A (en) | Robot scene adaptive bit orientation estimation method based on RGB-D camera | |
CN104599258B (en) | A kind of image split-joint method based on anisotropic character descriptor | |
CN104156977B (en) | Point target movement velocity detection method based on multiple linear moveout scanning, extending and sampling | |
US20180308240A1 (en) | Method for estimating the speed of movement of a camera | |
CN111210477B (en) | Method and system for positioning moving object | |
CN109559355B (en) | Multi-camera global calibration device and method without public view field based on camera set | |
CN105741379A (en) | Method for panoramic inspection on substation | |
CN105913435B (en) | A kind of multiscale morphology image matching method and system suitable for big region | |
US20160292883A1 (en) | Method of estimating the speed of displacement of a camera | |
CN102982334A (en) | Sparse parallax obtaining method based on target edge features and gray scale similarity | |
CN106504274A (en) | A kind of visual tracking method and system based under infrared camera | |
CN105182678A (en) | System and method for observing space target based on multiple channel cameras | |
CN104143196B (en) | A kind of point target detection method based on many alignment moveout scans extension sampling | |
Kluger et al. | Temporally consistent horizon lines | |
CN104143179B (en) | Method for enhancing moving target through multi-linear-array time difference scanning expansion sampling | |
CN113034398A (en) | Method and system for eliminating jelly effect in urban surveying and mapping based on artificial intelligence | |
Reid et al. | 3D Trajectories from a Single Viewpoint using Shadows. | |
CN104700359A (en) | Super-resolution reconstruction method of image sequence in different polar axis directions of image plane | |
CN115830064A (en) | Weak and small target tracking method and device based on infrared pulse signals | |
CN115984592A (en) | Point-line fusion feature matching method based on SuperPoint + SuperGlue |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |