CN115731221A - Self-adaptive infrared small target detection method considering neighborhood anisotropy - Google Patents
Self-adaptive infrared small target detection method considering neighborhood anisotropy Download PDFInfo
- Publication number
- CN115731221A CN115731221A CN202211525288.5A CN202211525288A CN115731221A CN 115731221 A CN115731221 A CN 115731221A CN 202211525288 A CN202211525288 A CN 202211525288A CN 115731221 A CN115731221 A CN 115731221A
- Authority
- CN
- China
- Prior art keywords
- neighborhood
- response
- target
- representing
- gray
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of image processing and target detection, and particularly relates to a self-adaptive infrared small target detection method considering neighborhood anisotropy. The invention provides a self-adaptive infrared small target detection method considering neighborhood anisotropy, which is characterized in that gray level responses in different directions around a central region are calculated through a template designed by the method, the gray level responses are different from the gray level value of the central region to obtain target significance responses in different directions, the minimum value of the target significance responses is taken as the significance response of a central point, two types of templates of 7 x 7 (a first template) and 9 x 9 (a second template) are designed to be respectively applied to the steps in order to adapt to targets with different sizes, the value of the significance responses under two different scales is taken as the final significance correspondence, the neighborhood gray level distribution characteristics of the small target in an infrared image are fully utilized, and the target significance in each direction can be well represented.
Description
Technical Field
The invention belongs to the technical field of image processing and target detection, and particularly relates to a self-adaptive infrared small target detection method considering neighborhood anisotropy.
Background
The infrared detection technology has been widely applied to various military and civil fields because of its excellent all-day and anti-interference capabilities. However, due to the problems of long imaging distance, complex imaging background, noise and clutter interference, the infrared detection system usually shows very weak targets on the imaging plane. The common visible light waveband target detection method is often failed in the field of infrared small targets, and a method developed for detecting the infrared small targets is difficult to balance time overhead and accuracy, namely the existing high-precision detection method is complex in theory, such as tensor low-rank decomposition and the like, and large in calculation consumption.
Therefore, an accurate and efficient method for detecting small infrared targets is needed.
Disclosure of Invention
Based on the problem that the existing infrared small target detection method is difficult to consider both high accuracy and low time consumption, the invention provides the infrared small target detection method considering neighborhood anisotropy, which is high in efficiency and good in detection performance.
The technical scheme adopted by the invention is as follows:
a self-adaptive infrared small target detection method considering neighborhood anisotropy comprises the following steps:
a: taking an original infrared image containing a small target, and calculating 8-direction gray response images with 7 multiplied by 7 as the size of a neighborhood according to a designed first template;
b: respectively subtracting the original infrared image containing the small target from the gray level response images in 8 directions obtained in the step A, and taking the absolute value to obtain target significance response images in 8 directions in a 7 multiplied by 7 neighborhood;
c: calculating the minimum value of the 8 absolute difference values obtained in the step B pixel by pixel to obtain a target significance response image of a 7 multiplied by 7 neighborhood;
d: taking an original infrared image containing a small target, and calculating a gray response image in 8 directions of the size of a 9 multiplied by 9 neighborhood according to a designed second template;
e: d, calculating a gray level average value image of the original infrared image containing the small target with the size of 3 x 3 as the neighborhood, respectively making a difference between the gray level average value image and the gray level response images in 8 directions obtained in the step D, and taking an absolute value to obtain target significance response images in 8 directions in a 9 x 9 neighborhood;
f: calculating the minimum value of the 8 absolute difference values obtained in the step E pixel by pixel to obtain a target significance response image of a 9 multiplied by 9 neighborhood;
g: taking a larger value in the result graphs obtained in the step C and the step F pixel by pixel to obtain a final target significance response graph;
h: and carrying out normalization and threshold segmentation on the final target significance response graph to obtain a target segmentation result graph.
After the technical scheme is adopted, gray level responses in different directions around the central area are calculated through the template designed by the invention, the difference is made between the gray level responses and the gray level value of the central area, target significance responses in different directions are obtained, and the minimum value is taken as the significance response of the central point. It should be noted that, in order to adapt to the targets with different sizes, the present invention designs two types of templates, namely 7 × 7 (first template) and 9 × 9 (second template), to be applied to the above steps respectively, and takes the larger value of the significance responses at two different scales as the final significance response.
Further, the gray response map in 8 directions with the neighborhood size of 7 × 7 in step a may be performed through 8 image-dependent filtering operations, which are specifically as follows:
in formula (1):representing a two-dimensional correlation filtering operation; f represents an original infrared image containing a small target; h 1,i Represents the ith template in the first templates; m 1,i Represents the ith twoThe result of the dimension dependent filtering operation, i.e. the gray response map in the ith direction in the 7 x 7 neighborhood.
Further, the step B of taking the difference between the original infrared image containing the small target and the gray-scale response maps in 8 directions obtained in step a and taking the absolute value to obtain the saliency response maps of the target in 8 directions in the 7 × 7 neighborhood is specifically performed by the following operations:
D 1,i =|F-M 1,i | i=1,2,…,8 (2)
in formula (2): | represents an absolute value, F represents an original infrared image containing a small target, M 1,i Representing the gray response map of the ith direction in a 7 × 7 neighborhood, D 1,i Representing the target saliency response plot for the ith direction within the 7 x 7 neighborhood.
Further, the minimum value of the 8 absolute differences obtained in the step B is calculated pixel by pixel in the step C to be used as a target saliency response of a 7 × 7 neighborhood of the center pixel point, and specifically, the following operations are performed:
D 1 =min 1≤i≤8 (D 1,i ) (3)
in formula (3): d 1,i Representing the saliency response of objects in the ith direction in a 7 × 7 neighborhood, D 1 A target significance response map representing a 7 x 7 neighborhood.
Further, the step D of taking the original infrared image containing the small target, calculating a gray response map in 8 directions with a neighborhood size of 9 × 9 according to the designed second template, and may be performed through 8 image-dependent filtering operations:
in formula (4):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small object, H 2,i Represents the ith template in the second template, M 2,i Represents the firstThe result of i two-dimensional correlation filtering operations, i.e. the gray response map in the ith direction in the 9 x 9 neighborhood.
Further, the step E calculates a gray-scale mean value map of the original infrared image including the small target with a size of 3 × 3 as a neighborhood, and makes a difference between the gray-scale mean value map and the gray-scale response maps in 8 directions obtained in the step D, and takes an absolute value to obtain a target saliency response map in 8 directions in a 9 × 9 neighborhood, specifically by the following operations:
D 2,i =|T-M 2,i | i=1,2,…,8 (6)
in formula (5):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small target, H mean Representing a 3 × 3 mean filtering template, and T represents a gray level mean image;
in formula (6): | represents absolute value, T represents gray-scale mean value graph, M 2,i Representing the gray response map of the ith direction in a 9 × 9 neighborhood, D 2,i And representing the saliency response map of the target in the ith direction in the 9 x 9 neighborhood, and T representing the gray mean map.
Further, the step F of calculating the minimum value of the 8 absolute difference values obtained in the step E pixel by pixel to be used as a target saliency response of the 9 × 9 neighborhood of the central pixel point is specifically performed by the following operations:
D 2 =min 1≤i≤8 (D 2,i ) (7)
in formula (7): d 2,i Representing the saliency response of objects in the ith direction in the 9X 9 neighborhood, D 2 A target saliency response map representing a 9 x 9 neighborhood.
Further, the step G of taking a larger value in the result graphs obtained in the step C and the step F pixel by pixel to obtain a final target significance response graph is specifically performed by the following operations:
S=max(D 1 ,D 2 ) (8)
in formula (8): d 1 Target saliency response map, D, representing a 7 × 7 neighborhood 2 Representing the target saliency response map of the 9 x 9 neighborhood, and S representing the final target saliency response map.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. the invention relates to an infrared small target detection method considering neighborhood anisotropy. It should be noted that, in order to adapt to the targets with different sizes, the present invention designs two types of templates, i.e., 7 × 7 (first template) and 9 × 9 (second template), to be applied to the above steps, respectively, and takes the larger value of the significance responses at two different scales as the final significance response.
2. According to the method, the gray level responses of 8 directions around the central area are calculated in the step A and the step D according to the special template, the gray level distribution characteristics of the small target in the infrared image are fully utilized, and the target significance of each direction can be well represented after the step B and the step E are combined respectively.
3. In the invention, the step C and the step F respectively take the minimum value of the 8 absolute difference values obtained in the step B and the step E, so that a larger response can still be kept at a real target with significance in each direction, and the response is greatly reduced at a non-target without significance or with significance in a few directions.
4. In the invention, the target saliency map is normalized and subjected to threshold segmentation in the step H, so that false alarms with low response can be removed, and only real targets with high response are reserved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
FIG. 1 is a first template designed according to the present invention;
FIG. 2 is a second template of the present invention design;
FIG. 3 is a flow chart of a method of infrared small target detection that accounts for neighborhood anisotropy;
fig. 4 is an original infrared image containing a small target to be detected according to the first embodiment of the present invention;
FIG. 5 is a gray scale response plot of 8 directions around in a 7 × 7 neighborhood of the first embodiment of the present invention;
FIG. 6 is a graph of the saliency response of an object for 8 directions within a 7 × 7 neighborhood of the first embodiment of the present invention;
FIG. 7 is a graph of the target saliency response for a 7 × 7 neighborhood of embodiment one of the present invention;
FIG. 8 is a gray scale response plot for 8 directions around the 9 × 9 neighborhood of the first embodiment of the present invention;
FIG. 9 is a graph of the saliency response of objects for 8 directions within a 9 × 9 neighborhood of the first embodiment of the present invention;
FIG. 10 is a graph of the target saliency response for the 9 × 9 neighborhood of the first embodiment of the present invention;
FIG. 11 is a final target saliency response plot of an embodiment one of the present invention;
FIG. 12 is a graph of the result of the target segmentation according to the first embodiment of the present invention;
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
It is noted that relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
A self-adaptive infrared small target detection method considering neighborhood anisotropy comprises the following steps:
a: taking an original infrared image containing a small target, and calculating a gray response image in 8 directions with 7 multiplied by 7 as the size of a neighborhood according to a designed first template;
b: respectively subtracting the original infrared image containing the small target from the gray level response images in 8 directions obtained in the step A, and taking the absolute value to obtain target significance response images in 8 directions in a 7 multiplied by 7 neighborhood;
c: calculating the minimum value of the 8 absolute difference values obtained in the step B pixel by pixel to obtain a target significance response map of a 7 multiplied by 7 neighborhood;
d: taking an original infrared image containing a small target, and calculating a gray response image in 8 directions of the size of a 9 multiplied by 9 neighborhood according to a designed second template;
e: d, calculating a gray level average value image of the original infrared image containing the small target with the size of 3 x 3 as the neighborhood, respectively making a difference between the gray level average value image and the gray level response images in 8 directions obtained in the step D, and taking an absolute value to obtain target significance response images in 8 directions in a 9 x 9 neighborhood;
f: calculating the minimum value of the 8 absolute difference values obtained in the step E pixel by pixel to obtain a target significance response image of a 9 multiplied by 9 neighborhood;
g: taking a larger value in the result graphs obtained in the step C and the step F pixel by pixel to obtain a final target significance response graph;
h: and carrying out normalization and threshold segmentation on the final target significance response graph to obtain a target segmentation result graph.
In this embodiment, the gray response map in 8 directions with a neighborhood size of 7 × 7 in step a may be performed through 8 image-dependent filtering operations, which are specifically as follows:
in formula (1):representing a two-dimensional correlation filtering operation; f represents an original infrared image containing a small target; h 1,i Represents the ith template in the first templates; m is a group of 1,i Representing the result of the ith two-dimensional correlation filtering operation, i.e., the gray response map for the ith direction in a 7 x 7 neighborhood.
In this embodiment, the difference between the original infrared image containing the small target in step B and the gray-scale response maps in 8 directions obtained in step a is obtained, and the absolute value is obtained to obtain the saliency response maps of the target in 8 directions in a 7 × 7 neighborhood, which is specifically performed by the following operations:
D 1,i =|F-M 1,i | i=1,2,…,8 (2)
in the formula (2): | represents an absolute value, F represents an original infrared image containing a small target, M 1,i Representing within a 7 x 7 neighborhoodGray scale response map of ith direction, D 1,i Representing the target saliency response plot for the ith direction within the 7 x 7 neighborhood.
In this embodiment, the minimum value of the 8 absolute difference values obtained in the step B is calculated pixel by pixel in the step C, and is used as the target saliency response of the 7 × 7 neighborhood of the central pixel, specifically performed by the following operations:
D 1 =min 1≤i≤8 (D 1,i ) (3)
in formula (3): d 1,i Representing the saliency response of objects in the ith direction in the 7 × 7 neighborhood, D 1 A target significance response map representing a 7 x 7 neighborhood.
In this embodiment, the step D of taking the original infrared image containing the small target, and calculating a gray response map in 8 directions of the size of the 9 × 9 neighborhood according to the designed second template may be performed through 8 image-related filtering operations:
in formula (4):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small object, H 2,i Represents the ith template in the second template, M 2,i Representing the result of the ith two-dimensional correlation filtering operation, i.e., the gray response map of the ith direction in a 9 x 9 neighborhood.
In this embodiment, the step E calculates a gray scale average value map of the original infrared image including the small target with 3 × 3 as a neighborhood size, and performs a difference between the gray scale average value map and the gray scale response maps in 8 directions obtained in step D, and takes an absolute value to obtain a target saliency response map in 8 directions in a 9 × 9 neighborhood, specifically by the following operations:
D 2,i =|T-M 2,i | i=1,2,…,8 (6)
in formula (5):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small target, H mean Representing a 3 × 3 mean filtering template, and T represents a gray level mean value graph;
in formula (6): | represents the absolute value, T represents the gray-level mean value map, M 2,i Representing the gray response map of the ith direction in a 9 × 9 neighborhood, D 2,i And representing the saliency response map of the target in the ith direction in the 9 x 9 neighborhood, and T representing the gray mean map.
In this embodiment, the step F of calculating the minimum value of the 8 absolute difference values obtained in the step E pixel by pixel is used as a target saliency response of a 9 × 9 neighborhood of the central pixel, and specifically performed by the following operations:
D 2 =min 1≤i≤8 (D 2,i ) (7)
in formula (7): d 2,i Representing the saliency response of objects in the ith direction in the 9X 9 neighborhood, D 2 A target saliency response map representing a 9 x 9 neighborhood.
In this embodiment, the larger value in the result graphs obtained in step C and step F is obtained pixel by pixel in step G to obtain a final target saliency response graph, which is specifically performed by the following operations:
S=max(D 1 ,D 2 ) (8)
in formula (8): d 1 Target saliency response map, D, representing a 7 × 7 neighborhood 2 Representing the target saliency response map of the 9 x 9 neighborhood, and S representing the final target saliency response map. .
The features and properties of the present invention are described in further detail below with reference to examples.
Referring to fig. 3, a method for detecting a small infrared target considering neighborhood anisotropy according to a preferred embodiment of the present invention includes the following steps:
A. taking an original infrared image containing small targets, as shown in fig. 4 (the small infrared targets are marked by white boxes), calculating a gray response graph in 8 directions with 7 × 7 as the neighborhood size according to a first template (see fig. 1) designed by the invention, and obtaining a result as shown in fig. 5;
in formula (1):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small object, H 1,i Represents the ith template (i.e., the correlation kernel), M, in template 1 1,i Representing the result of the ith two-dimensional correlation filtering operation (i.e., the gray response map for the ith direction in a 7 x 7 neighborhood).
B. Taking the original infrared image containing the small target and the gray level response images in 8 directions obtained in the step A to respectively perform difference, and taking the absolute value to obtain target significance response images in 8 directions in a 7 × 7 neighborhood, wherein the result is shown in FIG. 6;
in formula (2): | represents the absolute value, F represents the original infrared image containing the small object, M 1,i Representing the gray response map of the ith direction in a 7 × 7 neighborhood, D 1,i Representing the target significance response map of the ith direction in the 7 x 7 neighborhood.
C. Calculating the minimum value of the 8 absolute difference values obtained in the step B pixel by pixel to obtain a target significance response diagram of a 7 multiplied by 7 neighborhood, wherein the result is shown in FIG. 7;
D 1 =min 1≤i≤8 (D 1,i ) (3)
in formula (3): d 1,i Representing the saliency response of objects in the ith direction in the 7 × 7 neighborhood, D 1 A target saliency response map representing a 7 x 7 neighborhood.
D. Taking an original infrared image containing a small target, calculating a gray response image in 8 directions with the size of a 9 multiplied by 9 neighborhood according to a designed second template, and obtaining a result shown in fig. 8;
in formula (4):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small target, H 2,i Represents the ith template (i.e., the correlation kernel), M, in template 2 2,i Representing the result of the ith two-dimensional correlation filtering operation (i.e., the gray response map for the ith direction in a 9 x 9 neighborhood).
E. D, calculating a gray level average value image of the original infrared image containing the small target with the size of 3 x 3 as the neighborhood, respectively carrying out difference on the gray level average value image and the gray level response images in the 8 directions obtained in the step D, taking absolute values to obtain target significance response images in the 8 directions in the 9 x 9 neighborhood, and obtaining a result shown in fig. 9;
D 2,i =|T-M 2,i | i=1,2,…,8 (6)
in formula (5):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small target, H mean Represents a 3 × 3 mean filtering template;
in formula (6): | represents taking the absolute value, M 2,i Representing the gray response map of the ith direction in a 9 × 9 neighborhood, D 2,i Representing the target saliency response map in the ith direction within the 9 x 9 neighborhood.
F. Calculating the minimum value of the 8 absolute difference values obtained in the step E pixel by pixel to obtain a target significance response map of a 9 × 9 neighborhood, and the result is shown in fig. 10;
D 2 =min 1≤i≤8 (D 2,i ) (7)
in formula (7): d 2,i Representing the saliency response of objects in the ith direction in a 9 × 9 neighborhood, D 2 A target saliency response map representing a 9 x 9 neighborhood.
G. Taking a larger value in the result graphs obtained in the step C and the step F pixel by pixel to obtain a final target significance response graph, wherein the result is shown in FIG. 11;
S=max(D 1 ,D 2 ) (8)
in formula (8): d 1 Target saliency response map, D, representing a 7 × 7 neighborhood 2 Representing the target saliency response map of the 9 x 9 neighborhood, and S representing the final target saliency response map.
H. And (4) normalizing and threshold segmentation are carried out on the final target significance response graph to obtain a target segmentation result graph, and the result is shown in fig. 12.
The embodiment shows that the gray level responses in different directions around the central area are calculated through the target designed by the method, the difference is made between the gray level responses and the gray level value of the central area, the target significance responses in different directions are obtained, and the minimum value is taken as the significance response of the central point. It should be noted that, in order to adapt to the targets with different sizes, the present invention designs two types of templates, namely 7 × 7 (first template) and 9 × 9 (second template), to be applied to the above steps respectively, and takes the larger value of the significance responses at two different scales as the final significance response.
The above-mentioned embodiments only express the specific embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for those skilled in the art, without departing from the technical idea of the present application, several changes and modifications can be made, which are all within the protection scope of the present application.
Claims (8)
1. A self-adaptive infrared small target detection method considering neighborhood anisotropy is characterized by comprising the following steps: the method comprises the following steps:
a: taking an original infrared image containing a small target, and calculating a gray response image in 8 directions with 7 multiplied by 7 as the size of a neighborhood according to a designed first template;
b: respectively subtracting the original infrared image containing the small target from the gray level response images in 8 directions obtained in the step A, and taking the absolute value to obtain target significance response images in 8 directions in a 7 multiplied by 7 neighborhood;
c: calculating the minimum value of the 8 absolute difference values obtained in the step B pixel by pixel to obtain a target significance response image of a 7 multiplied by 7 neighborhood;
d: taking an original infrared image containing a small target, and calculating a gray response image in 8 directions of the size of a 9 multiplied by 9 neighborhood according to a designed second template;
e: d, calculating a gray level average value image of the original infrared image containing the small target with the size of 3 x 3 as the neighborhood, respectively carrying out difference on the gray level average value image and the gray level response images in the 8 directions obtained in the step D, and taking absolute values to obtain target significance response images in the 8 directions in the 9 x 9 neighborhood;
f: calculating the minimum value of the 8 absolute difference values obtained in the step E pixel by pixel to obtain a target significance response image of a 9 multiplied by 9 neighborhood;
g: taking a larger value in the result graphs obtained in the step C and the step F pixel by pixel to obtain a final target significance response graph;
h: and carrying out normalization and threshold segmentation on the final target significance response graph to obtain a target segmentation result graph.
2. The method for adaptive infrared small target detection considering neighborhood anisotropy according to claim 1, characterized in that: the gray response map in 8 directions with the neighborhood size of 7 × 7 in step a may be performed through 8 image-dependent filtering operations, which are specifically as follows:
in formula (1):representing a two-dimensional correlation filtering operation; f represents an original infrared image containing a small target; h 1,i Represents the ith template in the first templates; m 1,i Representing the result of the ith two-dimensional correlation filtering operation, i.e., the gray response map for the ith direction in a 7 x 7 neighborhood.
3. The method for adaptive infrared small target detection considering neighborhood anisotropy according to claim 2, characterized in that: and B, respectively subtracting the original infrared image containing the small target from the gray response images in 8 directions obtained in the step A, and taking absolute values to obtain the significance response images of the targets in 8 directions in a 7 multiplied by 7 neighborhood, wherein the method specifically comprises the following steps of:
D 1,i =|F-M 1,i | i=1,2,…,8 (2)
in the formula (2): | represents an absolute value, F represents an original infrared image containing a small target, M 1,i Representing the gray response map of the ith direction in the 7 x 7 neighborhood, D 1,i Representing the target saliency response plot for the ith direction within the 7 x 7 neighborhood.
4. The method for adaptive infrared small target detection considering neighborhood anisotropy according to claim 1, characterized in that: the minimum value of the 8 absolute differences obtained in the step B is calculated pixel by pixel in the step C, and is used as a target significance response of a 7 × 7 neighborhood of the central pixel point, and the method specifically includes the following operations:
D 1 =min 1≤i≤8 (D 1,i ) (3)
in formula (3): d 1,i Representing the saliency response of objects in the ith direction in the 7 × 7 neighborhood, D 1 A target significance response map representing a 7 x 7 neighborhood.
5. The method for adaptive infrared small target detection considering neighborhood anisotropy according to claim 1, characterized in that: the step D of taking the original infrared image containing the small target, calculating a gray response map in 8 directions of the size of the 9 × 9 neighborhood according to the designed second template, which may be performed by 8 image-dependent filtering operations:
in formula (4):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small object, H 2,i Represents the ith template in the second template, M 2,i Representing the result of the ith two-dimensional correlation filtering operation, i.e., the gray response map of the ith direction in a 9 x 9 neighborhood.
6. The method for adaptive infrared small target detection considering neighborhood anisotropy according to claim 1, characterized in that: step E, calculating a gray-scale mean value graph of the original infrared image containing the small target with 3 × 3 as the neighborhood size, respectively subtracting the gray-scale mean value graph from the gray-scale response graphs in 8 directions obtained in step D, and taking the absolute value to obtain target significance response graphs in 8 directions in a 9 × 9 neighborhood, specifically by the following operations:
D 2,i =|T-M 2,i | i=1,2,…,8 (6)
in formula (5):representing a two-dimensional correlation filtering operation, F representing an original infrared image containing a small target, H mean Representing a 3 × 3 mean filtering template, and T represents a gray level mean image;
in formula (6): | represents the absolute value, T represents the gray-level mean value map, M 2,i Representing the gray response map of the ith direction in a 9 × 9 neighborhood, D 2,i And representing the saliency response of the target in the ith direction in a 9 x 9 neighborhood, and T represents a gray-level mean value map.
7. The method for adaptive infrared small target detection considering neighborhood anisotropy according to claim 1, characterized in that: step F, calculating the minimum value of the 8 absolute difference values obtained in step E pixel by pixel to be used as the target significance response of the 9 × 9 neighborhood of the central pixel point, specifically by the following operations:
D 2 =min 1≤i≤8 (D 2,i ) (7)
in formula (7): d 2,i Representing the saliency response of objects in the ith direction in a 9 × 9 neighborhood, D 2 A target saliency response map representing a 9 x 9 neighborhood.
8. The method for adaptive infrared small target detection considering neighborhood anisotropy according to claim 1, characterized in that: and G, obtaining a final target significance response diagram by taking a larger value in the result diagrams obtained in the step C and the step F pixel by pixel, and specifically performing the following operations:
S=max(D 1 ,D 2 ) (8)
in formula (8): d 1 Target saliency response map, D, representing a 7 × 7 neighborhood 2 Representing the target saliency response map of the 9 x 9 neighborhood, and S representing the final target saliency response map.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211525288.5A CN115731221A (en) | 2022-11-30 | 2022-11-30 | Self-adaptive infrared small target detection method considering neighborhood anisotropy |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211525288.5A CN115731221A (en) | 2022-11-30 | 2022-11-30 | Self-adaptive infrared small target detection method considering neighborhood anisotropy |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115731221A true CN115731221A (en) | 2023-03-03 |
Family
ID=85299602
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211525288.5A Pending CN115731221A (en) | 2022-11-30 | 2022-11-30 | Self-adaptive infrared small target detection method considering neighborhood anisotropy |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115731221A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116363390A (en) * | 2023-05-25 | 2023-06-30 | 之江实验室 | Infrared dim target detection method and device, storage medium and electronic equipment |
-
2022
- 2022-11-30 CN CN202211525288.5A patent/CN115731221A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116363390A (en) * | 2023-05-25 | 2023-06-30 | 之江实验室 | Infrared dim target detection method and device, storage medium and electronic equipment |
CN116363390B (en) * | 2023-05-25 | 2023-09-19 | 之江实验室 | Infrared dim target detection method and device, storage medium and electronic equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110097093B (en) | Method for accurately matching heterogeneous images | |
Wang et al. | Unsupervised SAR image change detection based on SIFT keypoints and region information | |
CN111160120A (en) | Fast R-CNN article detection method based on transfer learning | |
CN107220962B (en) | Image detection method and device for tunnel cracks | |
CN113077392B (en) | High-accuracy automatic reading method for fuzzy photo of pointer table | |
CN109447036A (en) | A kind of segmentation of image digitization and recognition methods and system | |
CN115731221A (en) | Self-adaptive infrared small target detection method considering neighborhood anisotropy | |
CN110930425B (en) | Damaged target detection method based on neighborhood vector inner product local contrast image enhancement | |
CN117037132A (en) | Ship water gauge reading detection and identification method based on machine vision | |
CN108932468B (en) | Face recognition method suitable for psychology | |
CN107729863B (en) | Human finger vein recognition method | |
CN111950563A (en) | Image matching method and device and computer readable storage medium | |
CN115861669A (en) | Infrared dim target detection method based on clustering idea | |
CN108460773B (en) | Sonar image segmentation method based on offset field level set | |
CN113537397B (en) | Target detection and image definition joint learning method based on multi-scale feature fusion | |
CN114742849A (en) | Leveling instrument distance measuring method based on image enhancement | |
CN114429593A (en) | Infrared small target detection method based on rapid guided filtering and application thereof | |
Zhang et al. | Adaptive river segmentation in SAR images | |
CN113822361A (en) | SAR image similarity measurement method and system based on Hamming distance | |
CN113569731A (en) | Intelligent identification method, system, equipment and medium for satellite image tower | |
CN108364013B (en) | Image key point feature descriptor extraction method and system based on neighborhood Gaussian differential distribution | |
CN112329796A (en) | Infrared imaging cirrus cloud detection method and device based on visual saliency | |
CN112686222A (en) | Method and system for detecting ship target by satellite-borne visible light detector | |
Yu et al. | Improved threshold-based segmentation method for millimeter wave radiometric image | |
CN112884778B (en) | Robust machine vision target recognition segmentation method and system |
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 |