CN116912193A - Infrared weak and small target detection method and device based on local contrast of annular structure - Google Patents

Infrared weak and small target detection method and device based on local contrast of annular structure Download PDF

Info

Publication number
CN116912193A
CN116912193A CN202310847514.XA CN202310847514A CN116912193A CN 116912193 A CN116912193 A CN 116912193A CN 202310847514 A CN202310847514 A CN 202310847514A CN 116912193 A CN116912193 A CN 116912193A
Authority
CN
China
Prior art keywords
image
target
local contrast
annular
background
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
Application number
CN202310847514.XA
Other languages
Chinese (zh)
Inventor
杨德贵
胡亮
王存易
王行
白正阳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Central South University
Original Assignee
Central South University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Central South University filed Critical Central South University
Priority to CN202310847514.XA priority Critical patent/CN116912193A/en
Publication of CN116912193A publication Critical patent/CN116912193A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/194—Segmentation; Edge detection involving foreground-background 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
    • G06T2207/10048—Infrared image

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Quality & Reliability (AREA)
  • Image Processing (AREA)

Abstract

The invention provides an infrared weak and small target detection method and device based on local contrast of a ring structure, and the method comprises the following steps: s1, acquiring an original image to be detected, performing edge suppression processing, and outputting an edge suppression weighting coefficient; s2, constructing an annular structure according to an original image to be detected; s3, estimating an image background according to the gray value discrete degree of the peripheral annular background area, pre-enhancing the target according to the pixel value in the central area, obtaining the annular structure local contrast according to the ratio between the image background estimation result and the pre-enhancing processing result, and calculating a target enhanced image by using the annular structure local contrast; s4, obtaining a final target detection image according to the edge suppression weighting coefficient and the target enhancement image. The method can effectively estimate the image background and enhance the difference between the weak and small targets and the background, thereby rapidly and accurately realizing the detection of the weak and small targets.

Description

Infrared weak and small target detection method and device based on local contrast of annular structure
Technical Field
The invention relates to the technical field of image processing, in particular to an infrared weak and small target detection method and device based on local contrast of a ring structure
Background
Unmanned aerial vehicles, aerostats and other 'low-low' aircrafts bring potential threats to public safety, and effective detection and stable tracking of the aircrafts are necessary; also, accurate monitoring is required for non-cooperative targets with high threat such as ballistic missiles and high-altitude unmanned reconnaissance aircraft.
At present, the detection of the target mainly comprises visible light detection, radar detection, acoustic detection and infrared detection means. The audio detection mainly comprises the step of analyzing the sound emitted by a specific part of a sound signal target (such as an engine, an unmanned aerial vehicle wing and the like), when the target is far away from a monitoring platform or encounters a target with smaller sound (such as an aerostat and the like), the target is difficult to detect by the audio detection, and the propagation loss of the sound in a medium is larger, so that the audio detection is only suitable for short-distance detection and is easily influenced by environmental noise. The radar detection relies on the receiver to receive the signal that the transmitter that the target was reflected transmitted to detect the target, can long-range empty detection, and real-time is better, but to low altitude target, ground clutter can greatly reduced target detected's probability, and the radar belongs to the initiative detection device simultaneously, easily exposes the position in the course of the work. The visible light images through receiving the light energy reflected by the target, the imaging information is rich and has better comprehensibility, but the imaging distance of the target is short, and the problem of missed detection is easy to occur when the target is shielded, so that the application is often less when the remote early warning is performed. The infrared sensor performs imaging by capturing the radiant energy of a target, has the advantages of long imaging distance, all-weather work and the like, and is widely applied to the fields of early warning systems, space situation awareness and the like.
The current infrared target detection means can be roughly divided into a detection algorithm based on filtering, a detection algorithm based on local contrast, a detection algorithm based on data optimization and a detection algorithm based on depth science. The detection algorithm based on filtering mainly comprises three types of spatial filtering, frequency domain filtering and morphological filtering, and the detection effect of the algorithm is greatly influenced by the shapes of filter windows and structural elements, and has poor adaptability to complex backgrounds with a large number of edges and random highlight noise points. The infrared weak and small target detection algorithm based on data optimization is to realize detection of a target by establishing a data optimization model of the target and a background, however, for an infrared image with random highlight noise and strong and short edges, the false alarm rate of the algorithm is high, the time consumption of the algorithm is long, and the algorithm cannot be used for real-time detection. The detection algorithm based on deep learning is based on a constructed network model, and trains the existing data to generate a detection network, but the current infrared small target data set is less, and the generalization capability of the algorithm is poor.
In summary, the infrared target detection method in the prior art has low detection performance on the weak and small targets, so that the detection omission ratio is high, the time consumption is long, the detection difficulty is high, the adaptability to complex backgrounds with a large number of edges and random highlight noise points is poor, the false alarm rate is high, and the real-time detection on the weak and small targets is difficult to realize quickly and accurately in practice.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: aiming at the technical problems in the prior art, the invention provides an infrared dim target detection method and device based on the local contrast of a ring structure, which are used for enhancing the difference between a dim target and a background and eliminating the influence of random highlight noise and image edges on the dim target detection result.
In order to solve the technical problems, the technical scheme provided by the invention is as follows: an infrared weak and small target detection method based on local contrast of a ring structure comprises the following steps:
s1, acquiring an original image to be detected, performing edge suppression processing, and obtaining an edge suppression weighting coefficient output according to an edge suppression processing result;
s2, constructing an annular structure according to an original image to be detected, wherein a central area of the annular structure is a rectangular pixel block, and an annular area at the periphery of the central area is a peripheral annular background area;
s3, estimating an image background according to the gray value discrete degree of the annular background area, pre-enhancing the target according to the pixel value in the central area, obtaining the local contrast of the annular structure according to the ratio between the image background estimation and the pre-enhancing processing result, and calculating a target enhanced image by using the local contrast of the annular structure;
s4, obtaining a final target detection image according to the edge suppression weighting coefficient and the target enhancement image.
Further, in the step S2, the central area Ω C Pixel blocks set to n x n, the central region Ω C The peripheral annular background area isWhere (x, y) is the center pixel coordinate of the local structure, (p, q) is the pixel coordinate of the annular region, L represents the annular background region maximum number, and n represents the height and width (unit: pixel) of the center region.
Further, in the step S3, estimating the image background according to the gray value discrete degree of the peripheral annular background area includes:
s301, according to pixel values of peripheral annular background areasCalculating the degree of gray value dispersion +.>The computational expression is: :
wherein mean () represents the mean function;
s302, according to the gray value obtained by calculationDegree of dispersionThe image background BE is estimated as follows:
where k=1, 2, …, L represents the annular background area maximum number,the gray value dispersion degree of the jth peripheral annular background area is represented.
Further, in the step S3, the pre-enhancing the target according to the pixel value in the central area according to the following formula includes:
wherein G is n Representing the nth largest pixel value in the central region, mc representing the target pre-emphasis processing result, and K representing the total number of selected pixel values in the central region. .
Further, in the step S3, the local contrast of the annular structure is used to calculate the target enhanced image ALDM according to the following formula:
wherein Mc is the target pre-enhancement processing result, BE is the background estimation result,gray value degree of dispersion representing jth peripheral annular background region, +.>Representing the local contrast of the ring structure.
Further, in the step S1, filtering the original image to be detected by using a multi-wavelength multi-directional Gabor filter bank, calculating an image gradient strength of the original image to be detected, and calculating the edge suppression weighting coefficient according to a filtering result of the Gabor filter bank and the image gradient strength, wherein a wavelength set of the multi-wavelength filter bank is [ lambda ] 1 ,λ 2 ,…,λ N ]，λ N Represents the Nth wavelength, N represents the total number of wavelengths, and the direction group of the multidirectional filter bank is [ theta ] 1 ,θ 2 ,…,θ M ]，θ M Represents the M-th direction angle, M represents the total number of angles, and the direction group satisfiesAnd selecting the minimum response value of each wavelength as the image filtering response of the corresponding wavelength, and selecting the maximum response value of the wavelength set as the response of the whole Gabor filter set.
Further, when the image gradient intensity of the original image to be detected is calculated, selecting the gradient value of the pixel point in the neighborhood as the gradient value of the current pixel to obtain an updated gradient map Grad 2 And for the updated gradient map Grad 2 And performing corrosion operation to obtain final image gradient strength.
Further, the image gradient intensity is calculated according to the following formula: :
Grad 2 ＝max(Grad 1 ∩SE 1 ),SE 1 ＝[0,1,0；1,1,1；0,1,0]
Grad＝Grad 2 ΘSE 2 ,SE 2 ＝[1,1；1,1]
wherein Grad 2 Representing updated gradient map, grad 1 Representing the gradient strength of the original image, Θ representing the erosion operation, and n representing the intersection of the structural element with the image.
Further, the step S4 includes: the response F of the multi-wavelength multidirectional Gabor filter bank is multiplied by an image gradient map item by item to obtain the edge suppression weighting coefficient, the target enhanced image is multiplied by the edge suppression weighting coefficient item by item to obtain a target detection map, and the calculation formula of the target detection map is as follows:
W＝F⊙Grad
DSALCM＝ALDM⊙W
wherein W represents an edge suppression weighting coefficient, grad represents a target enhancement map, DSALCM represents a target detection map, and almm represents a target enhancement image.
The invention also provides an infrared dim target detection device based on the local contrast of the annular structure, which comprises a processor and a memory, wherein the memory is used for storing a computer program, and the processor is used for executing the computer program to execute the method
Compared with the prior art, the invention has the advantages that:
1. according to the invention, an annular structure is constructed according to the diffusion characteristic of a small target, the image background is estimated according to the gray value discrete degree of an annular background area, the image background is effectively estimated, meanwhile, the target is pre-enhanced according to the pixel value in a central area, then the local contrast of the annular structure is obtained according to the ratio between the image background estimation result and the pre-enhancement processing result, the local contrast of the annular structure is used for calculating the target enhanced image, the difference between the small target and the background can be enhanced, and the influence of random highlight noise on the detection result can be effectively eliminated while the target is enhanced.
2. The invention further adopts the multi-wavelength multi-directional Gabor filter group to carry out edge suppression processing on the image, and the edge suppression weighting coefficient is constructed by combining the image gradient through the multi-wavelength multi-directional Gabor filter group response selection method, so that the edge suppression can be effectively realized, and the influence of the edge on target detection is reduced.
3. The invention further uses the response of the wavelength group of the Gabor filter group as the response of the whole filter group, and multiplies the image gradient strength item by item to construct the edge suppression weighting coefficient, thereby further improving the edge suppression effect and further reducing the influence of the edge on target detection.
4. The method is further aimed at the problem of the internal cavity of the target caused by the traditional gradient calculation method, on the basis of the traditional gradient intensity method, the gradient value of the pixel point in the neighborhood is selected as the gradient value of the current pixel to generate the gradient map of the image, and then the obtained gradient map is subjected to corrosion operation, so that the internal cavity of the target can be effectively eliminated, and the problem of the internal cavity of the traditional target is solved.
Drawings
Fig. 1 is a schematic implementation flow chart of an infrared dim target detection method based on local contrast of a ring structure according to an embodiment of the invention.
Fig. 2 is a schematic diagram of an implementation flow of an infrared dim target detection algorithm according to an embodiment of the present invention.
Fig. 3 is a schematic view of a ring structure constructed according to an embodiment of the present invention.
Detailed Description
The invention is further described below in connection with the drawings and the specific preferred embodiments, but the scope of protection of the invention is not limited thereby.
As shown in fig. 1 to 3, the method for detecting the infrared weak and small target based on the local contrast of the annular structure in the embodiment comprises the following steps:
s1, acquiring an original image to be detected, performing edge suppression processing, and outputting an edge suppression weighting coefficient according to an edge suppression processing result;
s2, constructing an annular structure according to an original image to be detected, wherein a central area of the annular structure is a rectangular pixel block, and an annular area at the periphery of the central area is a peripheral annular background area;
s3, estimating an image background according to the gray value discrete degree of the annular background area, pre-enhancing the target according to the pixel value in the central area, obtaining the annular structure local contrast according to the ratio between the image background estimation result and the pre-enhancing processing result, and calculating a target enhanced image by using the annular structure local contrast;
s4, obtaining a final target detection image according to the edge suppression weighting coefficient and the target enhancement image.
In the embodiment, the edge suppression processing is performed by acquiring the original image to be detected, so that the output of the edge suppression weighting coefficient is obtained, the edge suppression can be effectively realized, the influence of the edge on the detection of the weak and small target is reduced, and the adaptability of the complex environment with a large number of edges is improved; meanwhile, according to the diffusion characteristics of a small target, an annular structure is constructed on an original image to be detected, wherein a central area of the annular structure is a rectangular pixel block, an annular area at the periphery of the central area is a peripheral annular background area, and the target is subjected to pre-enhancement treatment according to pixel values in the central area, so that the influence of random highlight noise points on detection can be effectively removed; the local contrast of the annular structure is obtained according to the ratio between the image background estimation result and the pre-enhancement processing result, the image background can be effectively estimated, the difference between the weak and small targets and the background is enhanced, and therefore the effect of random highlight noise on the detection result is eliminated while the enhancement of the targets is achieved.
As shown in fig. 3, in step S2 in the present embodiment, the center region Ω C Pixel block set as n×n, central region Ω C The peripheral annular background area isWhere (x, y) is the center pixel coordinate of the local structure, (p, q) is the pixel coordinate of the annular region, L represents the annular background region maximum number, and n represents the height and width (unit: pixel) of the center region.
Preferably, the central region Ω C The pixel block can be set as 3*3, and can be specifically configured according to actual requirements.
In this embodiment, in step S3, estimating the image background according to the gray value discrete degree of the peripheral annular background area includes:
s301, according to pixel values of peripheral annular background areasCalculating the degree of gray value dispersion +.>The computational expression is:
wherein mean () represents the mean function;
s302, obtaining the gray value discrete degree according to calculationThe image background BE is estimated as follows:
where k=1, 2, …, L represents the annular background area maximum number,the gray value dispersion degree of the jth peripheral annular background area is represented.
In this embodiment, the pixel values of the peripheral annular background area are calculatedGray value dispersion degree +.>When, first, the pixel value +.>The difference between the maximum value and the average value thereof, and calculating the pixel value +.>And finally selecting the maximum value of the two differences as the pixel value of the peripheral annular background area +.>Gray value dispersion degree +.>According to the calculated gray value dispersion degree +.>Selecting gray value dispersion degree->The maximum value in (3) estimates the image background BE, and the background estimation can BE effectively realized.
In this embodiment, the pre-enhancing the target according to the pixel value in the central area in step S3 includes:
wherein G is n Representing the nth largest pixel value in the central region, mc representing the target pre-emphasis processing result, and K representing the total number of selected pixel values in the central region.
According to the embodiment, the average value of the K gray values in the front of the central area is obtained, a pre-enhanced image of the target is generated, the difference between the weak and small target and the background can be effectively enhanced, and the influence of random highlight noise points on detection is removed.
In this embodiment, in step S3, the local contrast of the annular structure is used to obtain the target enhanced image ALDM according to the following formula:
wherein Mc is the target pre-enhancement processing result, BE is the background estimation result,gray value degree of dispersion representing jth peripheral annular background region, +.>Representing the local contrast of the ring structure.
According to the method and the device, the local contrast of the annular structure is obtained according to the ratio between the image background estimation result and the pre-enhancement processing result, target enhancement in the image is effectively achieved by utilizing the local contrast of the annular structure, and the difference between the weak and small targets and the background is fully highlighted, so that the weak and small targets can be effectively detected.
As shown in fig. 2, in step S1 of this embodiment, a multi-wavelength multi-directional Gabor filter bank is specifically used to perform filtering processing on an original image to be detected and calculate an image gradient strength of the original image to be detected, and an edge suppression weighting coefficient is calculated according to a filtering result of the Gabor filter bank and the image gradient strength, where a wavelength set of the multi-wavelength filter bank is [ lambda ] 1 ,λ 2 ,…,λ N ]，λ N N represents the N-th wavelength, N represents the total number of wavelengths, and the direction group of the multidirectional filter bank is [ theta ] 1 ,θ 2 ,…,θ M ]，θ M Represents the M-th direction angle, M represents the total number of angles, and the direction group satisfiesAnd selecting the minimum response value of each wavelength as the image filtering response of the corresponding wavelength, and selecting the maximum response value of the wavelength set as the response of the whole Gabor filter set.
Small objects typically appear as two-dimensional gaussian spots in the image, i.e. the gray value fluctuations in the central region in various directions are large, so that the response is relatively sharp for a multidirectional Gabor filter bank, which has small gray value fluctuations in the background region, and thus small for a multidirectional Gabor filter bank, but the edges of the image tend to have large gray fluctuations in a certain direction, and thus have large values for a multidirectional Gabor filter bank only in a certain direction or directions. Therefore, edge suppression can be achieved according to the response of the multidirectional Gabor filter bank at different wavelengths. The present embodiment uses the above characteristics of the multi-wavelength multidirectional Gabor filter bank to implement edge suppression by performing filter processing on the original image using the multidirectional Gabor filter bank.
In this embodiment, for a small target with unknown size, filter sets with different wavelengths are selected to filter the image, and the selected filter wavelength set is [ lambda ] 1 ,λ 2 ,…,λ N ](Unit: pixels), the directional group of the filter is [ θ ] 1 ,θ 2 ,…,θ M ](unit: °), and the direction group satisfiesWherein lambda is N The nth wavelength is represented, N represents the total number of wavelengths, θm represents the mth direction angle, and M represents the total number of angles. Let the wavelength be lambda j (j=1, …, N), direction θ i The response of the filter of (i=1, …, M) to the image is noted +.>For a wavelength lambda j (j=1, …, N) selecting the minimum value of the filter response to the image as the image filter response for the current wavelength, namely:
the maximum response value of the wavelength set is selected as the response of the whole filter set, namely:
F＝max(F j ),(j＝1,…,N) (6)
the gradient of the image can well describe the gray value fluctuation of the pixel points, and the gradient of the image is calculated as follows:
g in x (x,y)、G y (x, y) represents gradient values of the point (x, y) in the horizontal and vertical directions, respectively, whereby gradient intensities at the point (x, y) can be derived, respectively:
in this embodiment, when calculating the image gradient intensity of the original image to be detected, specifically selecting the gradient value of the pixel point in the neighborhood as the gradient value of the current pixel to obtain the updated gradient map Grad 2 And for the updated gradient map Grad 2 Performing corrosion operation to obtain final image gradient strength, wherein the image gradient strength is obtained by calculating according to the following formula:
Grad 2 ＝max(Grad 1 ∩SE 1 ),SE 1 ＝[0,1,0；1,1,1；0,1,0] (9)
Grad＝Grad 2 ΘSE 2 ,SE 2 ＝[1,1；1,1] (10)
wherein Grad 2 Representing updated gradient map, grad 1 Representing the gradient strength of the original image, Θ representing the erosion operation, and n representing the intersection of the structural element with the image.
For some small targets with uniform temperature distribution and larger pixel areas, the conventional gradient strength calculation method can cause a void phenomenon inside the target. On the basis of the traditional gradient intensity method, the embodiment generates an updated gradient map by selecting the gradient value of the pixel points in the neighborhood as the gradient value of the current pixel, and then carries out corrosion operation on the obtained gradient map, thereby effectively removing the target cavity phenomenon caused by the traditional gradient intensity calculation mode.
In this embodiment, step S4 includes: the response F of the multi-wavelength multidirectional Gabor filter bank is multiplied by the image gradient intensity item by item to obtain an edge suppression weighting coefficient W based on multidirectional filtering and image gradient, the target enhancement map is multiplied by the edge suppression weighting coefficient W item by item to obtain a target detection map, and a calculation formula of the target detection map is as follows:
W＝F⊙Grad (11)
DSALCM＝ALDM⊙W (12)
wherein W represents an edge suppression weighting coefficient, grad represents a target enhancement map, DSALCM represents a target detection map, and almm represents a target enhancement image.
In this embodiment, the maximum response value of the wavelength set is specifically selected as the response of the whole filter set, and the maximum response value is Hadamard product with the gradient map after the corrosion operation, so as to obtain the edge suppression weighting coefficient W based on multi-directional filtering and image gradient, which can effectively solve the problem of high false alarm rate caused by the image edge in the small target detection, and reduce the influence of the edge on the weak small target detection. Further, the target detection graph is obtained by carrying out Hadamard product on the target enhancement graph and the edge suppression weighting coefficient W, so that the influence of the image edge on target detection can be reduced, the difference between a weak and small target and a background can be enhanced, the influence of random highlight noise, edges and the like on the detection result can be eliminated, and the detection of the small target can be rapidly and accurately realized.
The embodiment provides an infrared weak small target detection device based on local contrast of a ring structure, which comprises a processor and a memory, wherein the memory is used for storing a computer program, and the processor is used for executing the computer program to execute the method.
The foregoing is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. While the invention has been described with reference to preferred embodiments, it is not intended to be limiting. Therefore, any simple modification, equivalent variation and modification of the above embodiments according to the technical substance of the present invention shall fall within the scope of the technical solution of the present invention.

Claims (10)

1. The method for detecting the infrared weak and small target based on the local contrast of the annular structure is characterized by comprising the following steps of:
s1, acquiring an original image to be detected, performing edge suppression processing, and obtaining an edge suppression weighting coefficient output according to an edge suppression processing result;
s2, constructing an annular structure according to an original image to be detected, wherein a central area of the annular structure is a rectangular pixel block, and an annular area at the periphery of the central area is a peripheral annular background area;
s3, estimating an image background according to the gray value discrete degree of the peripheral annular background area, pre-enhancing the target according to the pixel value in the central area, obtaining the annular structure local contrast according to the ratio between the image background estimation result and the pre-enhancing processing result, and calculating a target enhanced image by using the annular structure local contrast;
s4, obtaining a final target detection image according to the edge suppression weighting coefficient and the target enhancement image.
2. The method for detecting infrared small targets based on local contrast of annular structures according to claim 1, wherein in the step S2, the central region Ω is set C A pixel block set as n x n, the peripheral annular background area isWherein (x, y) is the center pixel coordinate of the center region, (p, q) is the pixel coordinate of the annular region, L represents the annular background region maximum number, and n represents the height and width of the center region.
3. The method for detecting infrared small objects based on the local contrast of the annular structure according to claim 1, wherein estimating the image background according to the gray value dispersion degree of the peripheral annular background region in the step S3 comprises:
s301, according to pixel values of peripheral annular background areasCalculating the degree of gray value dispersion +.>The computational expression is:
wherein mean () represents the mean function;
s302, according to the calculated gray value discrete degreeThe image background BE is estimated as follows:
where k=1, 2, …, L represents the annular background area maximum number,the gray value dispersion degree of the jth peripheral annular background area is represented.
4. The method for detecting infrared small targets based on the local contrast of the annular structure according to claim 1, wherein in the step S3, pre-enhancement processing is performed on the targets according to the following formula according to the pixel values in the central area:
wherein G is n Representing the nth largest pixel value in the central region, mc representing the target pre-emphasis processing result, and K representing the total number of selected pixel values in the central region.
5. The method for detecting infrared small target based on local contrast of annular structure according to claim 1, wherein in step S3, the target enhanced image ALDM is obtained by using the local contrast of annular structure according to the following formula:
wherein Mc is the target pre-enhancement processing result, BE is the background estimation result,representing the degree of gray value dispersion of the jth peripheral annular background region, < >>Representing the local contrast of the ring structure.
6. The method for detecting infrared small targets based on local contrast of annular structure according to any one of claims 1-5, wherein in the step S1, filtering an original image to be detected by using a multi-wavelength multidirectional Gabor filter bank, calculating an image gradient intensity of the original image to be detected, and calculating the edge suppression weighting coefficient according to a filtering result of the Gabor filter bank and the image gradient intensity, wherein a wavelength set of the multi-wavelength filter bank is [ lambda ] 1 ,λ 2 ,…,λ N ]，λ N Represents the Nth wavelength, N represents the total number of wavelengths, and the direction group of the multidirectional filter bank is [ theta ] 1 ,θ 2 ,…,θ M ]，θ M Represents the M-th direction angle, M represents the total number of angles, and the direction group satisfiesAnd selecting the minimum response value of each wavelength as the image filtering response of the corresponding wavelength, and selecting the maximum response value of the wavelength set as the response of the whole Gabor filter set.
7. The method for detecting infrared small targets based on local contrast of annular structures according to claim 6, wherein when calculating the image gradient intensity of the original image to be detected, selecting pixel points in the neighborhoodIs used as the gradient value of the current pixel to obtain an updated gradient map Grad 2 And for the updated gradient map Grad 2 And performing corrosion operation to obtain final image gradient strength.
8. The method for detecting an infrared small target based on the local contrast of the annular structure according to claim 6, wherein the image gradient intensity is calculated according to the following formula:
Grad 2 ＝max(Grad 1 ∩SE 1 ),SE 1 ＝[0,1,0；1,1,1；0,1,0]
Grad＝Grad 2 ΘSE 2 ,SE 2 ＝[1,1；1,1]
wherein Grad 2 Representing updated gradient map, grad 1 Representing the gradient strength of the original image, Θ representing the erosion operation, and n representing the intersection of the structural element with the image.
9. The method for detecting infrared small objects based on the local contrast of the annular structure according to any one of claims 1 to 5, wherein the step S4 includes: multiplying the response F of the multi-wavelength multidirectional Gabor filter bank with the image gradient intensity item by item to obtain the edge suppression weighting coefficient, multiplying the target enhanced image with the edge suppression weighting coefficient item by item to obtain a target detection diagram, wherein the calculation formula of the target detection diagram is as follows:
W＝F⊙Grad
DSALCM＝ALDM⊙W
wherein W represents an edge suppression weighting coefficient, grad represents a target enhancement map, DSALCM represents a target detection map, and almm represents a target enhancement image.
10. An infrared small target detection device based on the local contrast of a ring structure, comprising a processor and a memory for storing a computer program, characterized in that the processor is adapted to execute the computer program for performing the method according to any one of claims 1-9.
CN202310847514.XA 2023-07-11 2023-07-11 Infrared weak and small target detection method and device based on local contrast of annular structure Pending CN116912193A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310847514.XA CN116912193A (en) 2023-07-11 2023-07-11 Infrared weak and small target detection method and device based on local contrast of annular structure

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310847514.XA CN116912193A (en) 2023-07-11 2023-07-11 Infrared weak and small target detection method and device based on local contrast of annular structure

Publications (1)

Publication Number Publication Date
CN116912193A true CN116912193A (en) 2023-10-20

Family

ID=88367658

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310847514.XA Pending CN116912193A (en) 2023-07-11 2023-07-11 Infrared weak and small target detection method and device based on local contrast of annular structure

Country Status (1)

Country Link
CN (1) CN116912193A (en)

Similar Documents

Publication Publication Date Title
Nasiri et al. Infrared small target enhancement based on variance difference
CN110780271B (en) Spatial target multi-mode radar classification method based on convolutional neural network
CN112558066B (en) Dual polarized SAR image system
Sadjadi Improved target classification using optimum polarimetric SAR signatures
CN109597065B (en) False alarm suppression method and device for through-wall radar detection
CN110992298A (en) Genetic algorithm-based radiation source target identification and information analysis method
CN112731307B (en) RATM-CFAR detector based on distance-angle joint estimation and detection method
CN111580099A (en) Wall clutter suppression method of through-wall imaging radar based on joint entropy
CN116188944A (en) Infrared dim target detection method based on Swin-transducer and multi-scale feature fusion
CN114549642B (en) Low-contrast infrared dim target detection method
CN112669332A (en) Method for judging sea and sky conditions and detecting infrared target based on bidirectional local maximum and peak local singularity
CN103969647A (en) Sparse array architectural layout imaging method based on image domain filtering
CN116912193A (en) Infrared weak and small target detection method and device based on local contrast of annular structure
CN111311640A (en) Unmanned aerial vehicle identification and tracking method based on motion estimation
Lee et al. Background adaptive division filtering for hand-held ground penetrating radar
Wen et al. Multi-radar data fusion for maritime moving target detection based on three-dimensional sliding window
Yan et al. Clustering statistic Hough transform based estimation method for motion elements of multiple underwater targets
AU2020279716B2 (en) Multi-timescale doppler processing and associated systems and methods
CN110827257B (en) Visual navigation positioning method for embedded airborne infrared image
Lyons et al. Comparison of prescreening algorithms for target detection in synthetic aperture sonar imagery
Ma et al. Robust power line detection with particle-filter-based tracking in radar video
Hożyń et al. Tracking of unmanned aerial vehicles using computer vision methods: A comparative analysis
Faramarzi et al. Infrared dim and small targets detection based on multi-scale local contrast measure utilizing efficient spatial filters
Zhang et al. Target detection in sar images based on sub-aperture coherence and phase congruency
Sun et al. Infrared Small-Target Detection Based on Multi-level Local Contrast Measure

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination