CN104463911A - Small infrared moving target detection method based on complicated background estimation - Google Patents
Small infrared moving target detection method based on complicated background estimation Download PDFInfo
- Publication number
- CN104463911A CN104463911A CN201410744338.8A CN201410744338A CN104463911A CN 104463911 A CN104463911 A CN 104463911A CN 201410744338 A CN201410744338 A CN 201410744338A CN 104463911 A CN104463911 A CN 104463911A
- Authority
- CN
- China
- Prior art keywords
- image
- image block
- motion
- background
- template
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/215—Motion-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30212—Military
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a small infrared moving target detection method based on complicated background estimation. The method comprises the steps that a template matching method is utilized for estimating background motion vector parameters between two frames of images, the background motion vector parameters are added to the first frame of image to obtain a motion compensation image, a motion compensation image is removed from the other frame of image to obtain a foreground image with the background removed, saliency detection is carried out on the foreground image to obtain a saliency confidence picture, and thresholding is carried out on the saliency confidence picture to extract a small moving target. According to the small infrared moving target detection method, the small moving target is detected from the complicated infrared scene with no priori knowledge, and the small infrared moving target can be rapidly, accurately and automatically detected.
Description
Technical field
The present invention relates to a kind of infrared moving small target detection method estimated based on complex background, belong to the cross-application technical fields such as computer vision, pattern-recognition, image procossing.
Background technology
As the gordian technique of in infrared imaging detection system, the research of infrared small target detection receives the concern of Chinese scholars always.There is a target, from the process that infrared imaging device is formed, from the point-like Small object that remote is faint, develop into brighter, a stable spot, finally form a larger Area Objects.Obviously, on remote, find target, to winning initiative property, tool is marginal.When target range infrared eye is too far away, infrared target small-sized, contrast is very low, there is no the features such as obvious texture, structure, make the identification of target quite difficult.And infrared imaging image is gray level image, therefore in general infrared image, the edge of object and background can be fuzzyyer, and target texture is not obvious, and signal to noise ratio (S/N ratio) is low.In addition, due under normal circumstances, the background residing for target is extremely complicated, target by the clutter that occurs in a large number and noise pollute, make the process of infrared target more difficult.But target range is far away, its brightness is more weak, and area is less, and feature is more not obvious, and the difficulty of detection is larger.Therefore, the key of infrared target detection is the detection of infrared small target.
For the detection of infrared motion target, its task from each frame of image sequence, the position that Moving Objects exists detected and splits its region occupied, and as far as possible intactly it extracted from background.In the research detected moving target under static background, method of difference is a kind of comparatively conventional moving target detecting method, and people often it can be used as one of instrument processing all kinds of Target Tracking Problem.Its thought is the difference by consecutive frame, utilizes the strong correlation between image sequence consecutive frame to carry out change and detects, and extract moving target from background.Not only there is the internal noise of infrared eye itself in complicated infrared background, and more seriously there is the varying background clutter caused by cloud layer etc.In order to effectively detect the Small object of weak signal from above-mentioned background, there has been proposed background suppress technology.Classical way is the estimated parameter managing to obtain background based on image sequence, and then from input picture, subtracting background is estimated, just obtains a width signal and strengthens image.Finally, utilize thresholding or sequential gate method, correct for Small object is detected.Due to the motion of infrared seeker, in infrared imagery technique, background and target all may be in the state of a motion, will be subject to the motion of background and influenced based on traditional background estimating and detection method.
In actual applications, cannot obtain their priori for some infrared image and moving small target, be first the characteristic that can not obtain target, namely cannot be detected by the method for modeling; Secondly, due to the restriction of infrared image image-forming condition and low signal-to-noise ratio, make Small object not have the features such as obvious texture, structure, this gives the larger difficulty of detection of target; Finally, the servo-actuated of background brings larger interference to target.
Summary of the invention
The invention provides a kind of infrared moving small target detection method estimated based on complex background, achieve and detect moving small target from complicated Infrared Scene with without any priori condition, automatically can detect infrared moving small target fast and accurately.
In order to achieve the above object, the invention provides a kind of infrared moving small target detection method estimated based on complex background, comprise following steps:
Step S1, the background motion vector parameter utilizing template matching method to estimate between two two field pictures;
Step S2, the first two field picture is added background motion vector parameter, obtain motion compensated image, by next frame figure image subtraction motion compensated image, obtain the foreground image after removing background;
Step S3, foreground image done to conspicuousness and detect, obtain conspicuousness confidence map, thresholding process is done to conspicuousness confidence map and extracts moving small target.
Described step S1 comprises following steps:
Step S1.1, from arbitrary continuation two two field picture, first from the first two field picture, choose N number of image block;
Step S1.2, using any one image block in the first frame as template image block, adopt based on squared difference and template matching method, search for the accurate location of this template image block in next frame image;
Step S1.3, obtain the motion vector of this template image block between two two field pictures by image block matching primitives, this motion vector comprises direction of motion and shift value;
Step S1.4, all image blocks in the first two field picture are carried out to the operation of step S1.2 and step S1.3, obtain the direction of motion of all image blocks between two two field pictures and shift value;
Step S1.5, moving displacement value to be sorted from big to small, remove the image block that largest motion shift value is corresponding, statistical study is carried out to the direction of motion of residual image block and shift value, obtains direction of primary motion and the main motion shift value of background motion, form background motion vector parameter.
In described step S1.1, the N number of image block chosen can all regions of overlay image.
As claimed in claim 2 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, in described step S1.2, adopt based on squared difference and template matching method, the method for searching for the accurate location of template image block in next frame image in the first two field picture comprises:
The quadratic sum of employing difference represents the similarity between two two field pictures:
（1）
Above formula is launched, obtains:
（2）
After abbreviation, obtain:
（3）
Wherein, the correlativity of each image block in template image block and image to be matched is:
（4）
Wherein,
represent template image block,
it is the coordinate position of template image;
the sampled images block in image to be matched,
be step-size in search, refer to and to search at the neighborhood position of template image, move x and y pixel respectively in X-direction and Y-direction and obtain new image block
;
The image block that in image to be matched, correlation values is maximum is exactly the accurate location of template image block in next frame image.
In described step S1.5, the described direction of motion to residual image block and shift value carry out statistical study and comprise: set up two-dimensional coordinate system, the direction of motion of image block is divided into four direction, the X-direction moving displacement amount of the image block on first direction is greater than 0, and Y-direction moving displacement amount is greater than 0; The X-direction moving displacement amount of the image block in second direction is less than 0, and Y-direction moving displacement amount is greater than 0; The X-direction moving displacement amount of the image block on the 3rd direction is less than 0, and Y-direction moving displacement amount is less than 0; The X-direction moving displacement amount of the image block on four direction is greater than 0, and Y-direction moving displacement amount is less than 0; Add up total number of each direction epigraph block, directions maximum for image block number is defined as motion principal direction, using the mean value of the motion excursion amount of image blocks all on direction of primary motion as main motion shift value.
In described step S3, do conspicuousness to foreground image and detect, the method obtaining conspicuousness confidence map comprises:
To image
carry out Fourier transform, obtain its spectral amplitude
for:
（5）
Wherein,
ask image
fourier transform;
Obtain the phase spectrum of image further:
（6）
The spectral amplitude based on logarithm is asked to be:
（7）
The spectrum residual error of computed image
for:
（8）
To spectrum residual error
carry out Fourier inversion, then carry out a Gaussian blur filter and just obtain salient region, the final conspicuousness confidence map obtained is:
（9）
Wherein,
gaussian blurring function,
for inverse Fourier transform.
In described step S3, method conspicuousness confidence map being done to thresholding process extraction moving small target comprises:
To conspicuousness confidence map
do Threshold segmentation, obtain real Small object region:
（10）
Wherein, T is segmentation threshold;
In Small object region value be 1 region think in this region, to judge the region at target place the actual position of target, complete the extraction of moving small target.
The span of described segmentation threshold T is 0.3≤T≤0.6.
The present invention is by the estimation to context parameter, obtain the compensation image of preceding frame image, then foreground image is obtained in conjunction with two inter-frame relation, according to the moving target characteristic under complex scene, moving small target is comprised in foreground image, according to conspicuousness Detection and Extraction moving small target, achieve and detect moving small target from complicated Infrared Scene with without any priori condition, automatically can detect infrared moving small target fast and accurately.
Accompanying drawing explanation
Fig. 1 is process flow diagram of the present invention.
Fig. 2 is the template matching method schematic diagram based on image block.
Fig. 3 is the optimum position figure based on template matches.
Fig. 4 is background motion model parameters estimation method schematic diagram.
Fig. 5 determines motion principal direction and principal direction shift value schematic diagram.
Fig. 6 is based on background motion model parameters estimation experiment effect figure.
Fig. 7 is the process schematic from prospect extracting target from images.
Embodiment
Following according to Fig. 1 ~ Fig. 7, illustrate preferred embodiment of the present invention.
As shown in Figure 1, the invention provides a kind of infrared moving small target detection method estimated based on complex background, comprise following steps:
Step S1, the background motion vector parameter utilizing template matching method to estimate between two two field pictures;
Step S2, the first two field picture is added background motion vector parameter, obtain motion compensated image, by next frame figure image subtraction motion compensated image, obtain the foreground image after removing background;
Step S3, foreground image done to conspicuousness and detect, obtain conspicuousness confidence map, thresholding process is done to conspicuousness confidence map and extracts moving small target.
Described step S1 comprises following steps:
Step S1.1, from arbitrary continuation two two field picture, first from the first two field picture, choose N number of image block;
Generally can carry out 36 or 48 deciles to image, also directly can choose arbitrarily 50 image blocks, the principle chosen is all regions of overlay image as far as possible;
Step S1.2, using any one image block in the first frame as template image block, adopt based on squared difference and template matching method, search for the accurate location of this template image block in next frame image;
Step S1.3, obtain the motion vector of this template image block between two two field pictures (comprising direction of motion and shift value h) by image block matching primitives;
Step S1.4, all image blocks in the first two field picture are carried out to the operation of step S1.2 and step S1.3, obtain the direction of motion of all image blocks between two two field pictures and shift value h
1h
n;
Step S1.5, moving displacement value to be sorted from big to small, remove the image block that maximum deviation value is corresponding, statistical study is carried out to the direction of motion of residual image block and shift value, obtains direction of primary motion and the main motion shift value of background motion, form background motion vector parameter;
The described direction of motion to residual image block and shift value carry out statistical study and comprise: set up two-dimensional coordinate system, the direction of motion of image block is divided into four direction.First determine that the moving displacement amount of each image block is in X-direction, the positive negativity of Y-direction, wherein X-direction side-play amount is greater than 0, and Y-direction side-play amount is greater than 0, is designated as direction one; X-direction side-play amount is less than 0, and Y-direction side-play amount is greater than 0, is designated as direction two; X-direction side-play amount is less than 0, and Y-direction side-play amount is less than 0, is designated as direction three; X-direction side-play amount is greater than 0, and Y-direction side-play amount is less than 0, is designated as direction four.Then which direction each image block of statistical computation belongs to, and adds up the number of each direction epigraph block, and that direction that image block number is maximum is defined as motion principal direction.Calculate the mean value of the motion excursion amount of corresponding all image blocks in principal direction as main motion shift value.
In described step S1.2, adopt based on squared difference and template matching method, the method for searching for the accurate location of template image block in next frame image in the first two field picture comprises:
Template matches is exactly find the region the most similar with template image block in piece image.The grey scale pixel value packets of information of image contains all information of image record, and the coupling based on image pixel gray level value is the most basic matching algorithm thought.Usually directly utilize the half-tone information of image to set up similarity measurement between template image block and image to be matched, then adopt certain searching method to find the parameter value of the transformation model making similarity measure values maximum or minimum.Conventional similarity measurement is the similarity between calculating two image blocks, and making similarity maximum is exactly current best match position.
How measuring the similarity between two width images, is the parameter value adopting certain searching method to find the transformation model making similarity measure values maximum or minimum.Adopt quadratic sum (the Scharstein D of difference, Szeliski R. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms [J]. International journal of computer vision, 2002,47(1-3): 7-42.) represent this similarity:
（1）
Above formula is launched, obtains:
（2）
After abbreviation, obtain:
（3）
Wherein, the correlativity of each image block in template image block and image to be matched is:
（4）
Wherein,
represent template image block,
it is the coordinate position of template image;
the sampled images block in image to be matched,
be step-size in search, refer to and to search at the neighborhood position of template image, move x and y pixel respectively in X-direction and Y-direction and obtain new image block
, general selection x=1 and y=1.
Can be seen by above-mentioned formula, the Section 1 (namely the energy of template image block T) of above formula is a constant, the Section 3 energy of image block I (in the image to be matched) also can be similar to a constant, so can see, remaining Section 2 is exactly the correlativity between two image blocks, namely cross-correlation item.And cross correlation between two image blocks is larger, the similarity between them is less, otherwise correlation values is larger, and its similarity is larger.
Obtain the similarity measurements spirogram between template image and benchmark image according to said method, then find the image block that similarity is maximum.
In described step S3, do conspicuousness to foreground image and detect, the method obtaining conspicuousness confidence map comprises:
Due to the inconsistency of the target travel under dynamic scene and background motion, the foreground image obtained only comprises the Small object of motion, conspicuousness is done to foreground image and detects (Hou X, Zhang L. Saliency detection:A spectral residual approach [C], IEEE Conference on Computer Vision and Pattern Recognition, 2007.) extract the Small object moved.
The method step detected based on conspicuousness is: first to image
carry out Fourier transform, obtain its spectral amplitude
for:
（5）
In above formula,
ask image
fourier transform, obtain the phase spectrum of image further:
（6）
The spectral amplitude based on logarithm is asked to be:
（7）
According to the spectrum residual error of following formulae discovery image
for:
（8）
Then carry out Fourier inversion to spectrum residual error, then carry out a Gaussian blur filter and just obtain so-called salient region, final conspicuousness confidence map is:
（9）
In above formula,
gaussian blurring function,
for inverse Fourier transform.
In described step S3, method conspicuousness confidence map being done to thresholding process extraction moving small target comprises:
A Threshold segmentation is done to conspicuousness confidence map, obtains real Small object region:
（10）
In above formula, T is segmentation threshold, and its size is relevant with the gray scale of target with the contrast of object and background, usually selects T to be between 0.3-0.6.
By above-mentioned thresholding process obtain value be 1 region think the region at target place, finally according to the actual position of regional determination target, namely to extract moving small target.
Fig. 2 is the template matching method schematic diagram based on image block.Template matches is exactly find the region the most similar with template image block in piece image, and Fig. 2 mainly shows the flow process adopting image block to carry out template matches.First in the first two field picture, a selected image block, as template image, finds the position of mating most with this template image block in the neighborhood of then current location in the next frame.
Fig. 3 is the optimum position figure based on template matches.After selecting template image block in the first frame, Local Search is done in the neighborhood of next frame, find the image block of maximum correlation, the present invention adopts the gentle method of difference to calculate degree of similarity between two image blocks, just as shown in Figure 3 in search neighborhood, finds best matching image block position.
Fig. 4 is background motion model parameters estimation method schematic diagram.Add up direction of motion and the moving displacement size h1 of all image blocks in the first two field picture, h2, h3, hN, first remove some direction of motion and the larger singular point of offset deviation, then set up the remaining institute of the coordinate axis statistics direction of motion a little in four directions, obtain principal direction and the main motion shift value of background motion according to statistics, formation background motion model parameters estimation.
Fig. 5 is for determining motion principal direction and principal direction shift value.Set up two-dimensional coordinate system, the direction of motion of image is divided into four direction.First determine that the moving displacement amount of each image block is in X-direction, the positive negativity of Y-direction, wherein X-direction side-play amount dx is greater than 0, and Y-direction side-play amount dy is greater than 0, is designated as direction 1; X-direction side-play amount dx is less than 0, and Y-direction side-play amount dy is greater than 0, is designated as direction 2; X-direction side-play amount dx is less than 0, and Y-direction side-play amount dy is less than 0, is designated as direction 3; X-direction side-play amount dx is greater than 0, and Y-direction side-play amount dy is less than 0, is designated as direction 4.Then which direction each image block of statistical computation belongs to, and adds up the number of each direction epigraph block, and that direction that image block number is maximum is defined as motion principal direction.Calculate the mean value of the motion excursion amount of corresponding all image blocks in principal direction as main motion shift value.
Fig. 6 is based on background motion model parameters estimation experiment effect figure.Fig. 6 mainly shows the foreground image that can be obtained by background motion model parameters estimation under dynamic scene, first two field picture is added the motion vector of estimation, motion compensated image can be obtained, obtain error image with this compensation image of next frame figure image subtraction, this error image is exactly the foreground image after removing background.
Fig. 7 is the process schematic from prospect extracting target from images.The foreground image under dynamic scene can be obtained by aforesaid background estimating method, due to the inconsistency of the target travel under dynamic scene and background motion, the foreground image obtained only comprises the Small object of motion, foreground image is done to the Small object of conspicuousness Detection and Extraction campaign.Detected by conspicuousness and obtain conspicuousness confidence map, carry out thresholding process obtain value be 1 region think the region at target place, finally according to the actual position of regional determination target, namely to extract moving small target.
The present invention is by the estimation to context parameter, obtain the compensation image of preceding frame image, then foreground image is obtained in conjunction with two inter-frame relation, according to the moving target characteristic under complex scene, moving small target is comprised in foreground image, according to conspicuousness Detection and Extraction moving small target, realize detecting moving small target from complicated Infrared Scene with without any priori condition, automatically can detect infrared moving small target fast and accurately.
Although content of the present invention has done detailed introduction by above preferred embodiment, will be appreciated that above-mentioned description should not be considered to limitation of the present invention.After those skilled in the art have read foregoing, for multiple amendment of the present invention and substitute will be all apparent.Therefore, protection scope of the present invention should be limited to the appended claims.
Claims (8)
1., based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, comprise following steps:
Step S1, the background motion vector parameter utilizing template matching method to estimate between two two field pictures;
Step S2, the first two field picture is added background motion vector parameter, obtain motion compensated image, by next frame figure image subtraction motion compensated image, obtain the foreground image after removing background;
Step S3, foreground image done to conspicuousness and detect, obtain conspicuousness confidence map, thresholding process is done to conspicuousness confidence map and extracts moving small target.
2., as claimed in claim 1 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, described step S1 comprises following steps:
Step S1.1, from arbitrary continuation two two field picture, first from the first two field picture, choose N number of image block;
Step S1.2, using any one image block in the first frame as template image block, adopt based on squared difference and template matching method, search for the accurate location of this template image block in next frame image;
Step S1.3, obtain the motion vector of this template image block between two two field pictures by image block matching primitives, this motion vector comprises direction of motion and moving displacement value;
Step S1.4, all image blocks in the first two field picture are carried out to the operation of step S1.2 and step S1.3, obtain the direction of motion of all image blocks between two two field pictures and moving displacement value;
Step S1.5, moving displacement value to be sorted from big to small, remove the image block that largest motion shift value is corresponding, statistical study is carried out to the direction of motion of residual image block and shift value, obtains direction of primary motion and the main motion shift value of background motion, form background motion vector parameter.
3. as claimed in claim 2 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, in described step S1.1, the N number of image block chosen can all regions of overlay image.
4. as claimed in claim 2 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, in described step S1.2, adopt based on squared difference and template matching method, the method for searching for the accurate location of template image block in next frame image in the first two field picture comprises:
The quadratic sum of employing difference represents the similarity between two two field pictures:
（1）
Above formula is launched, obtains:
（2）
After abbreviation, obtain:
（3）
Wherein, the correlativity of each image block in template image block and image to be matched is:
（4）
Wherein,
represent template image block,
it is the coordinate position of template image;
the sampled images block in image to be matched,
be step-size in search, refer to and to search at the neighborhood position of template image, move x and y pixel respectively in X-direction and Y-direction and obtain new image block
;
The image block that in image to be matched, correlation values is maximum is exactly the accurate location of template image block in next frame image.
5. as claimed in claim 2 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, in described step S1.5, the described direction of motion to residual image block and shift value carry out statistical study and comprise: set up two-dimensional coordinate system, the direction of motion of image block is divided into four direction, the X-direction moving displacement amount of the image block on first direction is greater than 0, and Y-direction moving displacement amount is greater than 0; The X-direction moving displacement amount of the image block in second direction is less than 0, and Y-direction moving displacement amount is greater than 0; The X-direction moving displacement amount of the image block on the 3rd direction is less than 0, and Y-direction moving displacement amount is less than 0; The X-direction moving displacement amount of the image block on four direction is greater than 0, and Y-direction moving displacement amount is less than 0; Add up total number of each direction epigraph block, directions maximum for image block number is defined as motion principal direction, using the mean value of the motion excursion amount of image blocks all on direction of primary motion as main motion shift value.
6. as claimed in claim 1 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, in described step S3, do conspicuousness to foreground image and detect, the method obtaining conspicuousness confidence map comprises:
To image
carry out Fourier transform, obtain its spectral amplitude
for:
（5）
Wherein,
ask image
fourier transform;
Obtain the phase spectrum of image further:
（6）
The spectral amplitude based on logarithm is asked to be:
（7）
The spectrum residual error of computed image
for:
（8）
To spectrum residual error
carry out Fourier inversion, then carry out a Gaussian blur filter and just obtain salient region, the final conspicuousness confidence map obtained is:
（9）
Wherein,
gaussian blurring function,
for inverse Fourier transform.
7. as claimed in claim 6 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, in described step S3, method conspicuousness confidence map being done to thresholding process extraction moving small target comprises:
To conspicuousness confidence map
do Threshold segmentation, obtain real Small object region:
（10）
Wherein, T is segmentation threshold;
In Small object region value be 1 region think in this region, to judge the region at target place the actual position of target, complete the extraction of moving small target.
8., as claimed in claim 7 based on the infrared moving small target detection method that complex background is estimated, it is characterized in that, the span of described segmentation threshold T is 0.3≤T≤0.6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410744338.8A CN104463911A (en) | 2014-12-09 | 2014-12-09 | Small infrared moving target detection method based on complicated background estimation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410744338.8A CN104463911A (en) | 2014-12-09 | 2014-12-09 | Small infrared moving target detection method based on complicated background estimation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104463911A true CN104463911A (en) | 2015-03-25 |
Family
ID=52909893
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410744338.8A Pending CN104463911A (en) | 2014-12-09 | 2014-12-09 | Small infrared moving target detection method based on complicated background estimation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104463911A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069781A (en) * | 2015-07-17 | 2015-11-18 | 安徽大学 | Salient object spatial three-dimensional positioning method |
| CN105096333A (en) * | 2015-09-06 | 2015-11-25 | 河海大学常州校区 | Segmentation method for infrared thermal imaging image of forest fire |
| CN105869156A (en) * | 2016-03-25 | 2016-08-17 | 中国科学院武汉物理与数学研究所 | Infrared small target detection method based on fuzzy distance |
| CN106097317A (en) * | 2016-06-02 | 2016-11-09 | 南京康尼机电股份有限公司 | A kind of many spot detection based on discrete cosine phase information and localization method |
| CN106875415A (en) * | 2016-12-29 | 2017-06-20 | 北京理工雷科电子信息技术有限公司 | The continuous-stable tracking of small and weak moving-target in a kind of dynamic background |
| CN108198198A (en) * | 2017-12-22 | 2018-06-22 | 湖南源信光电科技股份有限公司 | Single frames infrared small target detection method based on wavelet transformation and Steerable filter |
| CN108805878A (en) * | 2018-05-22 | 2018-11-13 | 深圳腾视科技有限公司 | A kind of foreground object detection solution based on computer vision |
| CN109410243A (en) * | 2018-10-16 | 2019-03-01 | 中电科新型智慧城市研究院有限公司 | Based on the overlapping multi-object tracking method with template matching in region |
| CN109839115A (en) * | 2019-03-21 | 2019-06-04 | 北京中飞艾维航空科技有限公司 | A kind of UAV Landing method, apparatus and storage medium |
| CN110689554A (en) * | 2019-09-25 | 2020-01-14 | 深圳大学 | Background motion estimation method and device for infrared image sequence and storage medium |
| CN111489330A (en) * | 2020-03-24 | 2020-08-04 | 中国科学院大学 | Weak and small target detection method based on multi-source information fusion |
| CN112802020A (en) * | 2021-04-06 | 2021-05-14 | 中国空气动力研究与发展中心计算空气动力研究所 | Infrared dim target detection method based on image inpainting and background estimation |
| CN117218448A (en) * | 2023-09-28 | 2023-12-12 | 西安科技大学 | Coal and rock identification method based on Bayesian model fusion geological information |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103336947A (en) * | 2013-06-21 | 2013-10-02 | 上海交通大学 | Method for identifying infrared movement small target based on significance and structure |
-
2014
- 2014-12-09 CN CN201410744338.8A patent/CN104463911A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103336947A (en) * | 2013-06-21 | 2013-10-02 | 上海交通大学 | Method for identifying infrared movement small target based on significance and structure |
Non-Patent Citations (4)
| Title |
|---|
| HAIBIN DUAN ET AL.: "Small and Dim Target Detection via Lateral Inhibition Filtering and Artificial Bee Colony Based Selective Visual Attention", 《PLOS ONE》 * |
| 徐功益 等: "复杂背景下运动目标检测方法", 《红外与激光工程》 * |
| 胡暾 等: "基于显著性及主成分分析的红外小目标检测", 《红外与毫米波学报》 * |
| 赖作镁 等: "背景运动补偿和假设检验的目标检测算法", 《光学精密工程》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069781A (en) * | 2015-07-17 | 2015-11-18 | 安徽大学 | Salient object spatial three-dimensional positioning method |
| CN105069781B (en) * | 2015-07-17 | 2017-11-28 | 安徽大学 | A kind of spatial three-dimensional positioning method of well-marked target |
| CN105096333A (en) * | 2015-09-06 | 2015-11-25 | 河海大学常州校区 | Segmentation method for infrared thermal imaging image of forest fire |
| CN105869156B (en) * | 2016-03-25 | 2018-07-17 | 中国科学院武汉物理与数学研究所 | A kind of infrared small target detection method based on fuzzy distance |
| CN105869156A (en) * | 2016-03-25 | 2016-08-17 | 中国科学院武汉物理与数学研究所 | Infrared small target detection method based on fuzzy distance |
| CN106097317A (en) * | 2016-06-02 | 2016-11-09 | 南京康尼机电股份有限公司 | A kind of many spot detection based on discrete cosine phase information and localization method |
| CN106875415B (en) * | 2016-12-29 | 2020-06-02 | 北京理工雷科电子信息技术有限公司 | Continuous and stable tracking method for small and weak moving targets in dynamic background |
| CN106875415A (en) * | 2016-12-29 | 2017-06-20 | 北京理工雷科电子信息技术有限公司 | The continuous-stable tracking of small and weak moving-target in a kind of dynamic background |
| CN108198198A (en) * | 2017-12-22 | 2018-06-22 | 湖南源信光电科技股份有限公司 | Single frames infrared small target detection method based on wavelet transformation and Steerable filter |
| CN108805878A (en) * | 2018-05-22 | 2018-11-13 | 深圳腾视科技有限公司 | A kind of foreground object detection solution based on computer vision |
| CN109410243A (en) * | 2018-10-16 | 2019-03-01 | 中电科新型智慧城市研究院有限公司 | Based on the overlapping multi-object tracking method with template matching in region |
| CN109839115A (en) * | 2019-03-21 | 2019-06-04 | 北京中飞艾维航空科技有限公司 | A kind of UAV Landing method, apparatus and storage medium |
| CN110689554A (en) * | 2019-09-25 | 2020-01-14 | 深圳大学 | Background motion estimation method and device for infrared image sequence and storage medium |
| CN110689554B (en) * | 2019-09-25 | 2022-04-12 | 深圳大学 | Background motion estimation method and device for infrared image sequence and storage medium |
| CN111489330A (en) * | 2020-03-24 | 2020-08-04 | 中国科学院大学 | Weak and small target detection method based on multi-source information fusion |
| CN111489330B (en) * | 2020-03-24 | 2021-06-22 | 中国科学院大学 | Weak and small target detection method based on multi-source information fusion |
| CN112802020A (en) * | 2021-04-06 | 2021-05-14 | 中国空气动力研究与发展中心计算空气动力研究所 | Infrared dim target detection method based on image inpainting and background estimation |
| CN112802020B (en) * | 2021-04-06 | 2021-06-25 | 中国空气动力研究与发展中心计算空气动力研究所 | Infrared dim target detection method based on image inpainting and background estimation |
| CN117218448A (en) * | 2023-09-28 | 2023-12-12 | 西安科技大学 | Coal and rock identification method based on Bayesian model fusion geological information |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104463911A (en) | Small infrared moving target detection method based on complicated background estimation | |
| CN103077521B (en) | A kind of area-of-interest exacting method for video monitoring | |
| CN109086724B (en) | Accelerated human face detection method and storage medium | |
| CN102156995A (en) | Video movement foreground dividing method in moving camera | |
| CN110782442A (en) | Image artificial fuzzy detection method based on multi-domain coupling | |
| KR20110021500A (en) | Method for real-time moving object tracking and distance measurement and apparatus thereof | |
| CN111161308A (en) | Dual-band fusion target extraction method based on key point matching | |
| Fu et al. | Infrared small dim target detection under maritime near sea-sky line based on regional-division local contrast measure | |
| Yu et al. | Accurate motion detection in dynamic scenes based on ego-motion estimation and optical flow segmentation combined method | |
| Si-ming et al. | Moving shadow detection based on Susan algorithm | |
| CN114429593A (en) | Infrared small target detection method based on rapid guided filtering and application thereof | |
| Mei et al. | An Algorithm for Automatic Extraction of Moving Object in the Image Guidance | |
| KR100540889B1 (en) | Tracking Method for The Moving Object using Basic Points | |
| Wang et al. | An efficient image-registration method based on probability density and global parallax | |
| CN111476821B (en) | Target tracking method based on online learning | |
| CN101685538A (en) | Method and device for tracking object | |
| CN115205327B (en) | Infrared small target tracking method fusing historical library information | |
| Hu et al. | Anti-occlusion tracking algorithm of video target based on prediction and re-matching strategy | |
| Qu et al. | A high-precision registration algorithm for heterologous image based on effective sub-graph extraction and feature points bidirectional matching | |
| Prahara et al. | Depth inpainting scheme based on edge guided non local means | |
| CN108389219A (en) | A kind of Dim target tracking loss weight detecting method based on multi-peak judgement | |
| Xiangsuo et al. | Dim and small target detection based on local energy aggregation degree of sequence images | |
| Fan et al. | Infrared Dim and Small Target detection based on sample data | |
| Zhang et al. | Scale adaptive infrared small target detection with patch contrast measure | |
| Mito et al. | An object detection and extraction method using stereo camera |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150325 |