CN105574855B - Infrared small target detection method under cloud background based on template convolution and false alarm rejection - Google Patents
Infrared small target detection method under cloud background based on template convolution and false alarm rejection Download PDFInfo
- Publication number
- CN105574855B CN105574855B CN201510917030.3A CN201510917030A CN105574855B CN 105574855 B CN105574855 B CN 105574855B CN 201510917030 A CN201510917030 A CN 201510917030A CN 105574855 B CN105574855 B CN 105574855B
- Authority
- CN
- China
- Prior art keywords
- template
- cloud
- carried out
- image
- target
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 48
- 238000012545 processing Methods 0.000 claims abstract description 23
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000006243 chemical reaction Methods 0.000 claims abstract description 5
- 238000000034 method Methods 0.000 claims description 30
- 238000001914 filtration Methods 0.000 claims description 19
- 230000011218 segmentation Effects 0.000 claims description 10
- 238000002955 isolation Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000013459 approach Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 230000004069 differentiation Effects 0.000 claims description 2
- 238000007689 inspection Methods 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 6
- 230000000694 effects Effects 0.000 description 8
- 230000000007 visual effect Effects 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 229910000906 Bronze Inorganic materials 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 239000010974 bronze Substances 0.000 description 1
- KUNSUQLRTQLHQQ-UHFFFAOYSA-N copper tin Chemical compound [Cu].[Sn] KUNSUQLRTQLHQQ-UHFFFAOYSA-N 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses the infrared small target detection methods based on template convolution and false alarm rejection under a kind of cloud background, max-medium filter is carried out to image first and removes significant noise completion image preprocessing, secondly inhibit background with Robinson's template convolution, prominent target, then cloud sector division is carried out to original image, binary conversion treatment is carried out using Low threshold to the filtered result of Robinson in cloud sector part, rather than cloud sector part then uses high threshold to handle, multiple " the pseudo- target points " that same target generates further finally are rejected to the result after binaryzation, to complete " rough detection "；Consecutive frame image for having carried out spatial processing continues that time domain operation is taken to handle, and " essence detection " is completed, to realize the detection of infrared small target.Constant false alarm restrainable algorithms are added in the present invention in interframe track association, greatly reduce detection false alarm rate.
Description
Technical field
The invention belongs to infrared image detection and processing technology field, under especially a kind of cloud background based on template convolution and
The infrared small target detection method of false alarm rejection.
Background technique
Infrared small target detection is that Infra-Red Search & Track System, big visual field object detection system, satellite remote sensing, disaster are pre-
A core technology in the systems such as police, fire control and disaster rescue.Due to infrared sensor by atmosphere, radiation from sea surface, operating distance with
And the factors such as noise of detector influence, so that target size on infrared image is smaller at a distance, or even present dotted；This
Outside, the noise of image is relatively low, in addition background is more complicated under normal conditions, target is easy to be flooded by noise and background clutter
Not yet, so that the detection of infrared small target becomes more difficult.
The research of infrared small target detection method is directed generally to how to improve the detection probability of target and be reduced empty
Alert rate.Traditional detection method is divided into two major classes: Time Domain Processing and spatial processing.Time Domain Processing is acquired more based on different moments
Frame image sequence fully takes into account the correlation of image interframe, can inhibit in adjacent interframe to false-alarm, keeps higher
Detection accuracy.Common time-domain processing method has entropy difference method, and (Wang Guangjun, field inscription on ancient bronze objects, infrared image of the Liu Jian based on local entropy are small
Target detection [J] is infrared and laser engineering, 2000,04:26-29.), (grandson is infrared after rigid sequence image for sequence image detection method
Detecting and tracking dim small target algorithm research [D] Postgraduate School, Chinese Academy of Sciences (Changchun optical precision optical machinery and physics Institute),
2014.) etc..But registration operation must be carried out between this method consecutive frame, processing difficulty is larger, and algorithm is complex, and uncomfortable
Close the detection of single-frame images.And then using object pixel, there are significant differences with neighborhood territory pixel in gray scale for spatial processing, and with
The characteristics of correlation is not present in background, Pixel-level processing is carried out directly in its spatial domain to single-frame images, due to usually utilizing
Template operation, therefore algorithm is easy to be transplanted in hardware.Common method has morphology filter method (to spend the profit autumn, Zhang Ying, Lin Xiao
Infrared small target detection method [J] laser of the spring based on shape filtering and infrared, 2005,06:451-453.), high-pass filtering
Method (small target deteection [J] system engineering of the such as Dong Hongyan, Li Jicheng, Shen Zhenkang based on high-pass filtering and Order Filtering and electricity
Sub- technology, 2004,26 (5)) etc..But the false alarm rate of this method is higher, and robustness is poor, and can not be applied to video sequence
In processing.
Summary of the invention
The present invention provides the infrared small target detection method based on template convolution and false alarm rejection under a kind of cloud background, can
Infrared small target under cloud background accurately detected.
The present invention is to solve the technical solution of prior art problem to be: being pressed down under a kind of cloud background based on template convolution and false-alarm
The infrared small target detection method of system using the airspace operation of template convolution, carries out maximum single-frame images to image first
Median filtering removes significant noise and completes image preprocessing, secondly inhibits background, prominent target with Robinson's template convolution, so
Cloud sector division is carried out to original image afterwards, binary conversion treatment is carried out using Low threshold to the filtered result of Robinson in cloud sector part,
Rather than cloud sector part then uses high threshold to handle, and finally further rejects the multiple of same target generation to the result after binaryzation
" pseudo- target point ", to complete " rough detection "；Consecutive frame image for having carried out spatial processing continues to take at time domain operation
Reason carries out track association in interframe, and is directed to the difference of real goal and false-alarm point in terms of gamma characteristic and kinetic characteristic
It carries out constant false alarm and inhibits operation, " essence detection " is completed, to realize the detection of infrared small target.
Compared with prior art, the present invention its remarkable advantage are as follows: (1) combine spatial processing and Time Domain Processing, simultaneously
It has that spatial processing algorithm complexity is low, is easy to Hardware concurrently and realizes and Time Domain Processing detection accuracy is high, false alarm rate is low feature.
(2) preprocessing part proposes a kind of max-medium filter method on the basis of median filtering, to consider image all directions
On gamma characteristic.(3) it target rough detection part: introduces Robinson's filter method and inhibits background, prominent target；Divide cloud sector with it is non-
Cloud sector carries out binary conversion treatment to different piece using dual threshold, guarantees that the Small object in cloud sector is not filtered out blindly.(4)
Target essence detection part, using the difference between real goal and false-alarm point in terms of gamma characteristic and kinetic characteristic, in interframe
Constant false alarm restrainable algorithms are added in track association, greatly reduce detection false alarm rate.
The present invention is described in further detail with reference to the accompanying drawing.
Detailed description of the invention
Fig. 1 is the process of the infrared small target detection method based on template convolution and false alarm rejection under cloud background of the present invention
Figure.
Fig. 2 is the infrared image under single frames complexity cloud background comprising Small object.
Fig. 3 is max-medium filter treated effect picture.
Fig. 4 is the effect picture after Robinson's filtering processing.
Fig. 5 is the effect picture after cloud sector differentiates.
Fig. 6 is the effect picture after binarization operation.
Fig. 7 is the effect picture after adjacent point target combination.
Fig. 8 is the effect picture after interframe track association.
Fig. 9 is the effect picture after false alarm rejection.
Figure 10 is false alarm rate statistical chart.
Figure 11 is detectivity statistical chart.
Specific embodiment
In conjunction with Fig. 1, infrared small target detection method under cloud background of the present invention based on template convolution and false alarm rejection, for
Single-frame images is carried out max-medium filter to image first and is removed significant noise completion using the airspace operation of template convolution
Secondly image preprocessing inhibits background, prominent target with Robinson's template convolution, then cloud sector division is carried out to original image, in cloud
Area part carries out binary conversion treatment using Low threshold to the filtered result of Robinson, rather than cloud sector part then uses at high threshold
Reason finally utilizes " neighborhood non-maxima suppression " principle further to reject the multiple of same target generation to the result after binaryzation
" pseudo- target point ", to complete " rough detection "；Consecutive frame image for having carried out spatial processing continues to take at time domain operation
Reason carries out track association in interframe, and is directed to the difference of real goal and false-alarm point in terms of gamma characteristic and kinetic characteristic
It carries out constant false alarm and inhibits operation, " essence detection " is completed, to realize the detection of infrared small target.Its specific implementation step is as follows:
1. image preprocessing.Max-medium filter is carried out to input picture as shown in Figure 2, removes significantly making an uproar in image
Sound, i.e. max-medium filter take horizontal, vertical, 45 degree left, right 45 degree of four filtering directions, take in all directions in pixel grey scale
The maximum value of value gives center pixel.Max-medium filter has fully taken into account the distribution of the pixel grey scale in multiple directions, and more
The energy for retaining target in image well realizes efficiently denoising, processing knot on the basis of not destroying target original gray feature
Fruit is as shown in Figure 3.
For the max-medium filter template of (2N+1) * (2N+1), its calculation formula is:
fmax-med(i, j)=max (z1,z2,z3,z4) (1)
In formula (1),
Wherein, (i, j) is center pixel coordinate, and med is to take median operation, and max is to be maximized operation, and N characterizes mould
Board size size, in the method for the present invention, by taking N=2 as an example, i.e., template size is 5*5.
2. rough detection, process are as follows:
(1) Robinson's template convolution is carried out to the image after max-medium filter, inhibits cloud background, and prominent target.It should
Template is using this larger characteristic of gray difference between real goal and cloud background, by central pixel point gray value and surrounding picture
The maximum gradation value of element is compared and arithmetical operation.It if the gray scale of center pixel is stronger, can be retained, otherwise will be pressed down
System.Meanwhile template-setup has isolation strip, it is ensured that the gamma characteristic of real goal is not destroyed.Its processing result such as Fig. 4 institute
Show.
The method of Robinson's template convolution are as follows: for Robinson's Filtering Template of (2N+1) * (2N+1), calculation formula
Are as follows:
In formula (3),
Wherein, (i, j) is center pixel coordinate, and max is to be maximized operation, and N characterizes the template size size present invention
In method, by taking N=4 as an example, i.e., template size is 9*9.
(2) cloud sector differentiation is carried out to original image.Since cloud sector typically exhibits out linear or stratiform in infrared image, utilize
This feature, the method that template matching can be taken to filter in cloud, cloud exterior domain is distinguish.On the one hand the template is equipped with sky
White isolation strip to protect the gamma characteristic of real goal not to be damaged, on the other hand by by the gray scale of central row all pixels it
It carries out doing difference operation with the sum of the pixel grey scale with the top a line or bottom a line and normalizes to [0,255] section, and
Cloud sector is completed using binarization operation to divide.Specifically, which is divided into following 3 steps:
(i) matched filtering operation is carried out to original image.For the matched filtering template of (2M+1) * (2N+1) size, meter
Calculate formula are as follows:
In formula (5),
z1=sum (f (i-N, j-M:j+M))
z2=sum (f (i, j-M:j+M)) (6)
z3=sum (f (i+N, j-M:j+M))
Wherein, (i, j) is center pixel coordinate, and sum is sum operation, and M, N characterize template size size, in the present invention
M takes 5, N to take 1, i.e. template size is 11*3.
(ii) gray scale normalization processing is carried out to the result after matched filtering.Normalize formula are as follows:
(iii) binarization segmentation is carried out to the result after normalization.The formula of binarization operation are as follows:
F in formula (8)thresholdFor binarization threshold.Herein, the pixel expression cloud sector that gray value is 255, and gray scale
Value indicates non-cloud sector for 0 pixel.
Through aforesaid operations, cloud sector as shown in Figure 5 can be obtained and differentiate result.
(3) in the filtered image of Robinson, binaryzation point is carried out using different threshold values to cloud sector and non-cloud sector part
It cuts.Since the contrast difference between cloud sector target and cloud background is smaller, rather than cloud sector part is then larger, therefore takes in cloud sector
Low threshold binarization segmentation takes high threshold binarization segmentation in non-cloud sector.Specific formula are as follows:
Cloud sector part:
Non- cloud sector part:
F in formula (9) (10)LAnd fHRespectively indicate two different size of binarization thresholds.Wherein the former is Low threshold,
The latter is high threshold.As shown in fig. 6, more complete extraction effect can be obtained after Double Thresholding Segmentation.
(4) continuous point target combination is carried out to the result after binarization segmentation, it is ensured that a target only leaves a pixel
Size " isolated point ".Concrete operations are that the pixel for being 255 to gray value after binaryzation records its coordinate position, and map
Into the filtered image of Robinson, using it as center pixel, the template (L refers to template length) of a L*L is opened up, is such as opened up
The template of one 5*5.If the gray value of center pixel is the maximum value of all pixels in template, by it in binary image
Retain, is otherwise just rejected and (" neighborhood non-maxima suppression " principle is utilized further to reject to the result after binaryzation together
Multiple " the pseudo- target points " that one target generates).Its processing result is as shown in fig. 7, each continuous point target is successfully merged into one
Isolated point.
3. the step of essence detection are as follows:
(1) image that continuous multiple frames are passed through with above-mentioned airspace operation, carries out interframe track association.It is built first with list structure
Vertical track manager successively compares the point coordinate in present frame with the point coordinate in previous frame since the second frame image
Compared with, if fall in previous frame some point institute opened round Bo Mennei, be considered as and be successfully associated.The wave door of traditional track association is chosen
The method dynamic generation using prediction of speed is usually required, but the number of pixels that infrared small target occupies in the picture is seldom, frame
Between amount of movement also very little, therefore wave door radius can be taken as a definite value (generally can use 1-2 pixel).As shown in figure 8, every
The suspected target obtained after a rough detection has been respectively formed stable track in interframe (track is red mark part).
(2) constant false alarm is carried out to the result after track association and inhibits operation.The method of the present invention utilizes real goal and false-alarm
There are difference of both kinetic characteristic and gamma characteristic between point, the two is distinguished.Assuming that current time is k, object
It is v in the speed at k momentk, it is in the sum of the speed at preceding k-1 momentIt enables(wherein, alpha+beta=1).
Since noise movement meets standardized normal distribution, so its Δ v levels off to 0 whithin a period of time, and the Δ v of real goal will become
It is bordering on a positive number.Setting speed threshold value VT, as Δ v > VTWhen, it is real goal by the target-recognition, conversely, then making an uproar for false-alarm
Sound (kinetic characteristic)；
In terms of gamma characteristic, since real goal is gradually to approach ours in visual field, in this mistake from the distant to the near
Cheng Zhong, gray scale is necessarily also ascending transformation, and the gray scale of noise is held essentially constant.For " shade of gray " this object
Reason amount investigates gamma characteristic: assuming that current time is k, gray scale of the object at the k moment is Ik, in the gray scale at k-1 moment
For Ik-1, then the shade of gray J of current time objectk=Ik-Ik-1, it is in the sum of the gradient at k-1 moment before thisIt enables(wherein, alpha+beta=1).For noise, whithin a period of time, Δ J levels off to 0, and real goal
Δ J approach is a positive number.Set Grads threshold JT, as Δ J > JTWhen, it is real goal by the target-recognition, conversely, then regarding
For false-alarm noise.Based on two above criterion, the constant false alarm inhibition operation during track association is just completed.Specific processing knot
Fruit is not as shown in figure 9, be labeled as real goal by yellow box by the track that false-alarm generates.
Advantage in infrared small target detection reliability to illustrate the invention, using the method for the present invention to 90 ° of big visual fields
Mobile Small object video under the cloud background of infrared thermovision system acquisition carries out simulation process, and is directed to false alarm rate and detectivity two
Index carries out statistics calculating.Wherein, false alarm rate is defined as the detection number of false target in per hour；Verification and measurement ratio is defined as correctly
Detect percentage of the destination number relative to realistic objective quantity.
As shown in Figure 10, abscissa is hourage, and ordinate is wrong report number, establishes false alarm rate statistical chart.It is computed,
The method of the present invention can control false alarm rate in 1.7 times/hour under this scene.
As shown in figure 11, abscissa is hourage, and ordinate is detectivity, establishes detectivity statistical chart.It is computed, herein
The method of the present invention can control detectivity in 96% under scene.
Claims (5)
1. the infrared small target detection method under a kind of cloud background based on template convolution and false alarm rejection, it is characterised in that: for
Single-frame images is carried out max-medium filter to image first and is removed significant noise completion using the airspace operation of template convolution
Secondly image preprocessing inhibits background, prominent target with Robinson's template convolution, then cloud sector division is carried out to original image, in cloud
Area part carries out binary conversion treatment using Low threshold to the filtered result of Robinson, rather than cloud sector part then uses at high threshold
Reason finally further rejects multiple " the pseudo- target points " that same target generates to the result after binaryzation, to complete " Rough Inspection
It surveys "；Consecutive frame image for having carried out spatial processing continues that time domain operation is taken to handle, i.e., carries out track association in interframe,
And carry out constant false alarm for the difference of real goal and false-alarm point in terms of gamma characteristic and kinetic characteristic and inhibit operation, it completes
" essence detection ", to realize the detection of infrared small target；
Described image preprocessing process are as follows: max-medium filter is carried out to input picture, removes the significant noise in image, i.e., most
Big median filtering takes horizontal, vertical, 45 degree left, 45 degree of four filtering directions in the right side, takes in all directions pixel grey scale intermediate value most
Big value gives center pixel；
The step of essence detection are as follows:
(1) image that continuous multiple frames are passed through with airspace operation, carries out interframe track association, establishes track pipe first with list structure
Device is managed, since the second frame image, is successively compared the point coordinate in present frame with the point coordinate in previous frame, if falling in
Some point institute opened round Bo Mennei in previous frame, then be considered as and be successfully associated；Wave door radius is taken as a definite value；
(2) constant false alarm is carried out to the result after track association and inhibits operation, moved using existing between real goal and false-alarm point
Difference of both characteristic and gamma characteristic, distinguishes the two, i.e. hypothesis current time is k, speed of the object at the k moment
For vk, it is in the sum of the speed at preceding k-1 momentIt enablesWherein, alpha+beta=1；Since noise movement is full
Sufficient standardized normal distribution, so its Δ v levels off to 0 whithin a period of time, and the Δ v of real goal will level off to a positive number；
Setting speed threshold value VT, as Δ v > VTWhen, it is real goal by the target-recognition, conversely, being then false-alarm point；
For in terms of gamma characteristic, for " shade of gray ", this physical quantity investigates gamma characteristic: assuming that current time
For k, gray scale of the object at the k moment is Ik, it is I in the gray scale at k-1 momentk-1, then the shade of gray J of current time objectk=Ik-
Ik-1, it is in the sum of the gradient at k-1 moment before thisIt enablesWherein, alpha+beta=1, for noise,
Whithin a period of time, Δ J levels off to 0, and the Δ J of real goal approach is a positive number；Set Grads threshold JT, as Δ J
> JTWhen, it is real goal by the target-recognition, conversely, being then considered as false-alarm point；Based on two above characteristic, track is just completed
False alarm rejection operation in association process.
2. the infrared small target detection side based on template convolution and false alarm rejection under cloud background according to claim 1
Method, it is characterised in that for the max-medium filter template of (2N+1) * (2N+1), its calculation formula is:
fmax-med(i, j)=max (z1,z2,z3,z4)(1)
In formula (1),
Wherein, (i, j) is center pixel coordinate, and med is to take median operation, and max is to be maximized operation, and N characterizes template size
Size.
3. the infrared small target detection side based on template convolution and false alarm rejection under cloud background according to claim 1
Method, it is characterised in that the process of rough detection are as follows:
(1) Robinson's template convolution is carried out to the image after max-medium filter, inhibits cloud background, and prominent target, the template
Using this larger characteristic of the gray difference between real goal and cloud background, most by center pixel gray value and surrounding pixel
High-gray level value is compared and arithmetical operation, if the gray scale of center pixel is stronger, can be retained, otherwise will be suppressed；Together
When, template-setup has isolation strip, guarantees that the gamma characteristic of real goal is not destroyed；
(2) to original image carry out cloud sector differentiation, take template matching filter method in cloud, cloud exterior domain be distinguish, should
On the one hand template is equipped with blank isolation strip to protect the gamma characteristic of real goal not to be damaged, on the other hand by by central row
The sum of the sum of gray scale of all pixels and the pixel grey scale of the top a line or bottom a line are carried out doing difference operation and be normalized
To [0,255] section, and cloud sector is completed using binarization operation and is divided；
(3) in the filtered image of Robinson, binarization segmentation is carried out using different threshold values to cloud sector and non-cloud sector part,
Low threshold binarization segmentation is taken in cloud sector, takes high threshold binarization segmentation in non-cloud sector, specifically:
Cloud sector part:
Non- cloud sector part:
F in formula (9) (10)LAnd fHTwo different size of binarization thresholds are respectively indicated, wherein the former is Low threshold, the latter
For high threshold；
(4) continuous point target combination is carried out to the result after binarization segmentation, it is ensured that a target only leaves a pixel size
" isolated point ", i.e., to gray value after binaryzation be 255 pixel, record its coordinate position, and be mapped to Robinson filtering
In image afterwards, using it as center pixel, the template of a L*L is opened up, if the gray value of center pixel is all pictures in template
The maximum value of element, then retained in binary image, otherwise just rejected.
4. the infrared small target detection side based on template convolution and false alarm rejection under cloud background according to claim 3
Method, it is characterised in that the method for Robinson's template convolution are as follows: for Robinson's Filtering Template of (2N+1) * (2N+1), calculate
Formula are as follows:
In formula (3),
Wherein, (i, j) is center pixel coordinate, and max is to be maximized operation, and N characterizes template size size.
5. the infrared small target detection side based on template convolution and false alarm rejection under cloud background according to claim 3
Method, it is characterised in that the step of cloud sector differentiates are as follows:
(i) matched filtering operation is carried out to original image, for the matched filtering template of (2M+1) * (2N+1) size, calculated public
Formula are as follows:
In formula (5),
Wherein, (i, j) is center pixel coordinate, and sum is sum operation, and M, N characterize template size size；
(ii) gray scale normalization processing is carried out to the result after matched filtering, normalizes formula are as follows:
(iii) binarization segmentation, the formula of binarization operation are carried out to the result after normalization are as follows:
F in formula (8)thresholdFor binarization threshold, herein, the pixel that gray value is 255 indicates cloud sector, and gray value is 0
Pixel indicate non-cloud sector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510917030.3A CN105574855B (en) | 2015-12-10 | 2015-12-10 | Infrared small target detection method under cloud background based on template convolution and false alarm rejection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510917030.3A CN105574855B (en) | 2015-12-10 | 2015-12-10 | Infrared small target detection method under cloud background based on template convolution and false alarm rejection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105574855A CN105574855A (en) | 2016-05-11 |
| CN105574855B true CN105574855B (en) | 2019-01-22 |
Family
ID=55884946
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510917030.3A Expired - Fee Related CN105574855B (en) | 2015-12-10 | 2015-12-10 | Infrared small target detection method under cloud background based on template convolution and false alarm rejection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105574855B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106056115B (en) * | 2016-05-25 | 2019-01-22 | 西安科技大学 | A kind of infrared small target detection method under non-homogeneous background |
| CN106803260B (en) * | 2016-12-28 | 2019-08-09 | 辽宁师范大学 | Infrared ship activity of imagination contours segmentation method based on the convex optimization of local entropy |
| CN107633505A (en) * | 2017-08-24 | 2018-01-26 | 南京理工大学 | A kind of undercarriage detection method based on target gray distribution character |
| CN107644414A (en) * | 2017-08-27 | 2018-01-30 | 南京理工大学 | A kind of sea wake detection method based on constant statistics and Radon transform |
| CN109410137B (en) * | 2018-10-11 | 2021-10-01 | 中国科学院上海技术物理研究所 | Method for detecting dim and weak target |
| CN109544535B (en) * | 2018-11-26 | 2022-06-24 | 马杰 | Peeping camera detection method and system based on optical filtering characteristics of infrared cut-off filter |
| CN110084231A (en) * | 2019-04-29 | 2019-08-02 | 北京富吉瑞光电科技有限公司 | A kind of method and apparatus of infrared panoramic system multi-target detection |
| CN111493853A (en) * | 2020-04-24 | 2020-08-07 | 天津恒宇医疗科技有限公司 | Blood vessel parameter evaluation method and system for angiodermic diseases |
| CN113111883B (en) * | 2021-03-23 | 2023-06-06 | 浙江大华技术股份有限公司 | License plate detection method, electronic device and storage medium |
| CN112967305B (en) * | 2021-03-24 | 2023-10-13 | 南京莱斯电子设备有限公司 | Image cloud background detection method under complex sky scene |
| CN113240664B (en) * | 2021-06-03 | 2023-06-09 | 郑州航空工业管理学院 | Infrared detection false alarm detection method based on scene space-time significance and application thereof |
| CN113776676A (en) * | 2021-09-23 | 2021-12-10 | 中国航天科工集团八五一一研究所 | Infrared small target detection method based on image curvature and gradient |
| CN115311460B (en) * | 2022-08-16 | 2023-05-02 | 哈尔滨工业大学 | Infrared small target detection method fusing time-space domain information under slow motion background |
| CN115631114B (en) * | 2022-12-06 | 2023-03-14 | 北京九章星图科技有限公司 | Dark and weak moving target indication enhanced on-track processing method based on time domain profile analysis |
| CN117315498B (en) * | 2023-10-10 | 2024-05-24 | 中国人民解放军战略支援部队航天工程大学 | False alarm discrimination method based on space target detection result |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104125372A (en) * | 2014-07-29 | 2014-10-29 | 北京机械设备研究所 | Target photoelectric search and detection method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7742620B2 (en) * | 2003-03-21 | 2010-06-22 | Lockhead Martin Corporation | Target detection improvements using temporal integrations and spatial fusion |
-
2015
- 2015-12-10 CN CN201510917030.3A patent/CN105574855B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104125372A (en) * | 2014-07-29 | 2014-10-29 | 北京机械设备研究所 | Target photoelectric search and detection method |
Non-Patent Citations (2)
| Title |
|---|
| 基于光流估计和自适应背景抑制的弱小目标检测;秦剑等;《光子学报》;20110331;第40卷(第3期);476-482 |
| 基于面阵探测器的红外搜索预警系统关键技术研究;陆恺立;《万方学位论文数据库》;20151203;正文第17-31,36-39,42-43页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105574855A (en) | 2016-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105574855B (en) | Infrared small target detection method under cloud background based on template convolution and false alarm rejection | |
| EP3869459B1 (en) | Target object identification method and apparatus, storage medium and electronic apparatus | |
| CN106910203B (en) | The quick determination method of moving target in a kind of video surveillance | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| CN104899866B (en) | A kind of intelligentized infrared small target detection method | |
| Shahbaz et al. | Evaluation of background subtraction algorithms for video surveillance | |
| CN106022345B (en) | A kind of high voltage isolator state identification method based on Hough forest | |
| CN106096604A (en) | Multi-spectrum fusion detection method based on unmanned platform | |
| CN104657945A (en) | Infrared small target detection method for multi-scale spatio-temporal union filtering under complex background | |
| CN103761731A (en) | Small infrared aerial target detection method based on non-downsampling contourlet transformation | |
| CN104182992B (en) | Method for detecting small targets on the sea on the basis of panoramic vision | |
| CN111507235B (en) | Railway perimeter foreign matter intrusion detection method based on video | |
| CN109711256B (en) | Low-altitude complex background unmanned aerial vehicle target detection method | |
| CN113743260B (en) | Pedestrian tracking method under condition of dense pedestrian flow of subway platform | |
| CN102663385B (en) | Detection method for spot target on satellite | |
| CN103593679A (en) | Visual human-hand tracking method based on online machine learning | |
| Lian et al. | A novel method on moving-objects detection based on background subtraction and three frames differencing | |
| CN110503092A (en) | The improvement SSD monitor video object detection method adapted to based on field | |
| CN110084833A (en) | A kind of infrared motion target detection method based on adaptive neighborhood Technology of Judgment | |
| CN114663795A (en) | Target detection method for obtaining rear image of glass curtain wall by range gating imaging equipment | |
| CN111950518B (en) | Video image enhancement method for violent behavior recognition | |
| CN116012768B (en) | Crowd density detection method and device, electronic equipment and computer storage medium | |
| CN108198422A (en) | A kind of road ponding extraction system and method based on video image | |
| Han et al. | Object Detection based on Combination of Visible and Thermal Videos using A Joint Sample Consensus Background Model. | |
| Ho et al. | Automated detection of people distribution by A 3D camera |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190122 Termination date: 20201210 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |