Infrared remote sensing frozen lake detection method, device and storage medium

By using local entropy and census transformation in infrared remote sensing image processing, the missed detection and false detection problems in icy lake detection are solved, and more accurate icy lake detection is achieved .

CN116051975BActive Publication Date: 2024-11-29UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202211383113.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-11-07
Publication Date
2024-11-29
Estimated Expiration
2042-11-07

AI Technical Summary

Technical Problem

The prior art has problems of missed and missed detection in the detection of frozen lakes, making it difficult to accurately separate frozen lakes from backgrounds.

Method used

Using the detection method based on infrared remote sensing images, the image texture features are extracted through local entropy, and the weak texture features are extracted through census transformation. Combined with gradient vector flow and morphological processing, the frozen lake and background are segmented, and finally determined through area threshold determination. Ice lake.

Benefits of technology

It effectively improves the accuracy of icy lake detection, reduces missed and missed detection, and can more accurately separate icy lakes from backgrounds.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116051975B_ABST
    Figure CN116051975B_ABST
Patent Text Reader

Abstract

The invention discloses an infrared remote sensing frozen lake detection method, device and storage medium, which belong to the field of remote sensing image processing. The main purpose is to solve the existing problems of missed detection of frozen lakes and false detection of non-frozen lakes. The main scheme includes A. reading infrared remote sensing frozen lake images and extracting image texture features by using local entropy; B. extracting parts of the read image as left and right view images respectively and performing census transformation to extract weak texture features of the image; C. subtracting the images processed by steps A and B respectively to obtain images without texture features; D. then performing gradient vector flow processing to obtain the segmentation results of the frozen lake and the background; E. performing hole filling and sequential edge connection on the image processed by step D, processing discontinuous edges and obtaining multiple closed areas; F. calculating the area of ​​each irregular closed area, judging whether it is a frozen lake according to a threshold value, and marking the result on the original image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of remote sensing image processing and relates to an infrared remote sensing frozen lake detection method, equipment and storage medium. Background Art

[0002] Existing false alarm source detection for earth observation is based on the detection of complex backgrounds such as cirrus clouds, snow-capped mountains, frozen lakes, and rivers. Due to the difficulty in data acquisition, the research on false alarm source detection is relatively limited. With the development of hardware equipment, more and more information can be extracted from remote sensing images. In order to more accurately detect false alarm sources and eliminate them, it is necessary to model and analyze the characteristics of false alarm sources, so as to design a more general false alarm source detection method.

[0003] The current method for detecting frozen lakes is to process edge information through threshold method, region growing method, texture features, supervised classification, etc., to achieve the purpose of separating frozen lakes from the background. Machine learning and deep learning methods are mostly trained and tested on various features of frozen lakes, and the detection results have also achieved certain results. Summary of the invention

[0004] The purpose of the present invention is to provide a frozen lake detection method based on infrared remote sensing images to solve the problems of missed detection of frozen lakes and false detection of non-frozen lakes.

[0005] The technical solution adopted by the present invention is as follows:

[0006] A method for detecting frozen lakes based on infrared remote sensing images comprises the following steps:

[0007] Step A. read the infrared remote sensing frozen lake image, extract the image texture features using local entropy, and record the infrared remote sensing frozen lake image as image;

[0008] Step B. Weak texture feature processing: for the intercepted part of the infrared remote sensing frozen lake image read, respectively as the left and right view images, the left and right view images are subjected to census transformation to extract the weak texture features of the infrared remote sensing frozen lake image;

[0009] Step C. Texture-free feature processing: subtract the images processed by steps A and B respectively, remove the texture features of the frozen lake to facilitate subsequent gradient calculation, and obtain an infrared remote sensing frozen lake image without texture features;

[0010] Step D. Separation of frozen lake and background: performing gradient vector flow processing on the image processed in step C to obtain the segmentation result of the frozen lake and the background, that is, to obtain the final segmentation map;

[0011] Step E. Morphological and edge connection processing: performing hole filling and sequential edge connection on the image processed in step D, processing discontinuous edges and obtaining multiple edge closed areas;

[0012] Step F. Obtain the closed area and calculate the area to mark the result: Calculate the area of ​​each irregular closed area, determine whether it is a frozen lake based on the threshold, and mark the result on the original image.

[0013] Furthermore, the specific steps of step B are as follows:

[0014] Step B1: For the infrared remote sensing frozen lake image read, the image width range is intercepted as [0: width-maxDisparity] as the left view image, [maxDisparity: width] as the right view image, and an odd number (2N+1)×(2N+1) sliding window is designed, where width is the width of the image and maxDisparity is the maximum disparity.

[0015] Step B2: For any image, in the sliding window, all pixels except the center are compared with the center pixel. If the pixel is greater than the center pixel, it is recorded as 1, otherwise it is recorded as 0. The final comparison results are accumulated and can be expressed as follows:

[0016]

[0017] Where: C s (p) means to perform census transformation on the central pixel p and sum the results, q∈N p is the neighborhood of p, and ξ is the XOR operation of the central pixel p and the adjacent pixel q.

[0018] Step B3: The left view image and the right view image are traversed simultaneously, and neighborhood differences are compared for the positions of the same pixels.

[0019] C(p, q, maxDisparity) = |C s1 (p, q)-C s2 (p-maxDisparity,q)|

[0020] Where: C(p, q, maxDisparity) represents the left and right view images C with a disparity of maxDisparity s1 With C s2 The difference value of .

[0021] Furthermore, the specific steps of step E are as follows:

[0022] Step E1: For the hole-filled image, define the initial pixel point in the edge connection process to start from the upper left corner of the image, and set the point on the edge, define three movement directions, and obtain three movement paths. The path measurement for determining the quality of the path is expressed as follows:

[0023]

[0024] Where: Γ n is the last path measure, Γ n+1 is the current path measure, p1 is the probability that the pixel is on the edge, and p0 is the probability that the point is any pixel. is the pixel value of the edge point, is the pixel value of any point, P r is the path transition probability, s n+1 The status of the current path.

[0025] Step 2: For step 1, three possible motion paths are obtained. By comparing the path measures, only the largest measure is selected for extension, and three possible motion paths are obtained again. They are compared with the previous paths and the largest path measure is still selected for the next extension until it extends to the boundary of the image or forms a loop, obtaining the final clear edge of the image.

[0026] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0027] 1. A frozen lake detection method based on infrared remote sensing images, which extracts all texture features of the image through local entropy, and extracts the frozen lake features with weak texture through census transformation. The two are subtracted to obtain the frozen lake without texture features, and the frozen lake is segmented from the background through gradient vector flow. Finally, the morphological and sequential edge connection methods are used, and the area threshold is determined to obtain the actual frozen lake.

[0028] 2. In step B of the present invention, census transformation is performed on the infrared remote sensing frozen lake image, and the bit string of each image is directly accumulated. By comparing the difference of the accumulated values ​​of the left and right views, the texture features of the frozen lake can be effectively extracted.

[0029] 3. In step D of the present invention, the gradient vector flow calculation is performed on the textureless feature map, which can well segment the frozen lake and the background.

[0030] 4. In step E of the present invention, the final segmentation image is subjected to morphological hole filling processing and edge connection, so that all obvious edges can be obtained and connected, while false edges are eliminated. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0032] Figure 1 It is a flow chart of a method for detecting frozen lakes using infrared remote sensing;

[0033] Figure 2 The infrared remote sensing frozen lake image to be detected in the first embodiment of the present invention;

[0034] Figure 3 is the local entropy texture image extracted in the first embodiment of the present invention;

[0035] Figure 4 is the census transformation image of the first embodiment of the present invention;

[0036] Figure 5 The frozen lake image without texture features according to the first embodiment of the present invention;

[0037] Figure 6 is a segmentation map processed by a gradient vector flow according to the first embodiment of the present invention;

[0038] Figure 7 is a morphological processing and edge connection diagram of the first embodiment of the present invention;

[0039] Figure 8 This is a frozen lake detection result diagram according to the first embodiment of the present invention;

[0040] Fig. 9 It is a schematic diagram of the hardware structure of an icy lake detection device provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the embodiments described are only part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0042] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0043] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0044] A method for detecting frozen lakes based on infrared remote sensing images comprises the following steps:

[0045] Step A. read infrared remote sensing frozen lake images, and extract image texture features using local entropy. Hereinafter, the infrared remote sensing frozen lake images are referred to as images.

[0046] Step B. Weak texture feature processing: for the intercepted part of the infrared remote sensing frozen lake image read, respectively as the left and right view images, the left and right view images are subjected to census transformation to extract the weak texture features of the infrared remote sensing frozen lake image;

[0047] Step C. Texture-free feature processing: subtract the images processed by steps A and B respectively, remove the texture features of the frozen lake to facilitate subsequent gradient calculation, and obtain an infrared remote sensing frozen lake image without texture features;

[0048] Step D. Separation of frozen lake and background: performing gradient vector flow processing on the image processed in step C to obtain the segmentation result of the frozen lake and the background, that is, to obtain the final segmentation map;

[0049] Step E. Morphological and edge connection processing: performing hole filling and sequential edge connection on the image processed in step D, processing discontinuous edges and obtaining multiple edge closed areas;

[0050] Step F. Obtain the closed area and calculate the area to mark the result: Calculate the area of ​​each irregular closed area, determine whether it is a frozen lake based on the threshold, and mark the result on the original image.

[0051] A frozen lake detection method based on infrared remote sensing images extracts all texture features of the image through local entropy, and extracts the frozen lake features with weak texture through census transformation. The two are subtracted to obtain the frozen lake without texture features, and the frozen lake is segmented from the background through gradient vector flow. Finally, the morphology and sequential edge connection method are used, and the area threshold is determined to obtain the actual frozen lake.

[0052] Furthermore, the specific steps of step B are as follows:

[0053] Step B1: For the infrared remote sensing frozen lake image read, the image width range is intercepted as [0: width-maxDisparity] as the left view image, [maxDisparity: width] as the right view image, and an odd number (2N+1)×(2N+1) sliding window is designed, where width is the width of the image and maxDisparity is the maximum disparity.

[0054] Step B2: For any image, in the sliding window, all pixels except the center are compared with the center pixel. If the pixel is greater than the center pixel, it is recorded as 1, otherwise it is recorded as 0. The final comparison results are accumulated and can be expressed as follows:

[0055]

[0056] Where: C s (p) means to perform census transformation on the central pixel p and sum the results, q∈N p is the neighborhood of p, and ξ is the XOR operation of the central pixel p and the adjacent pixel q.

[0057] Step R3: The left view image and the right view image are traversed simultaneously, and neighborhood differences are compared for the positions of the same pixels.

[0058] C(p, q, maxDisparity) = |C s1 (p, q)-C s2 (p-maxDisparity,q)|

[0059] Where: C(p, q, maxDisparity) represents the left and right view images C with a disparity of maxDisparity s1 With C s2 The difference value of .

[0060] Furthermore, the specific steps of step E are as follows:

[0061] Step E1: For the hole-filled image, define the initial pixel point in the edge connection process to start from the upper left corner of the image, and set the point on the edge, define three movement directions, and the path measurement for determining the quality of the path is expressed as follows:

[0062]

[0063] Where: Γ n is the last path measure, Γ n+1 is the current path measure, p1 is the probability that the pixel is on the edge, and p0 is the probability that the point is any pixel. is the pixel value of the edge point, is the pixel value of any point, P r is the path transition probability, s n+1 The status of the current path.

[0064] Step E2: For step 1, three possible motion paths are obtained. By comparing the path measurements, only the largest measurement is selected for extension, and three possible motion paths are obtained again. After comparison with the previous path, the largest path measurement is still selected for the next extension until it extends to the boundary of the image or forms a loop, and the final obvious edge of the image is obtained.

[0065] The features and performance of the present invention are further described in detail below in conjunction with the embodiments.

[0066] Embodiment 1

[0067] A preferred embodiment of the present invention provides a frozen lake detection method based on infrared remote sensing images, comprising the following steps:

[0068] A. Read infrared frozen lake images, such as Figure 2 As shown in, local entropy is used to extract image texture features, such as Figure 3 As shown;

[0069] B. Weak texture feature processing: The intercepted parts of the read infrared frozen lake image are used as the left and right view images respectively, and the census transformation is performed on the two to extract the weak texture features of the frozen lake, such as Figure 4 As shown;

[0070] Step 1: For the infrared remote sensing frozen lake image read, the image width range is intercepted as [0: width-maxDisparity] as the left view image, [maxDisparity: width] as the right view image, and an odd number (2N+1)×(2N+1) sliding window is designed, where width is the width of the image and maxDisparity is the maximum disparity.

[0071] Step 2: For any image, in the window, all pixels except the center are compared with the pixel at the center. If it is greater than the center, it is recorded as 1, otherwise it is 0, and the final comparison results are accumulated, which can be expressed by the following formula:

[0072]

[0073] Where: C s (p) means to perform census transformation on the central pixel p and sum the results, qeN p is the neighborhood of p, and ξ is the XOR operation of the central pixel p and the adjacent pixel q.

[0074] Step 3: For the left view image and the right view image, traverse them simultaneously and compare the differences in the neighborhood for the same pixel positions.

[0075] C(p, q, maxDisparity) = |C s1 (p, q)-C s2 (p-maxDisparity,q)|

[0076] Where: C(p, q, maxDisparity) represents the left and right view images C with a disparity of maxDisparity s1 With C s2 The difference value of .

[0077] C. Texture-free feature processing: Subtract the images processed by steps A and B to remove the texture features of the frozen lake for subsequent gradient calculation, such as Figure 5 As shown;

[0078] D. Separation of frozen lake and background: Gradient vector flow processing is performed on the image processed in step C to obtain the segmentation result of the frozen lake and the background, which can be obtained by solving the following formula: Figure 6 Results shown;

[0079]

[0080] Where: ε is the energy function, u x , v x ,u y , v y is the gradient value corresponding to two directions at position (x, y) obtained by gradient vector flow, is the gradient of the edge function, and V(x, y) = [u(x, y), v(x, y)] is the vector field function.

[0081] Solve the vector field function V(x, y) that minimizes the energy function ε, which is the gradient to be calculated. For each pixel point, calculate the square of the amplitude of the corresponding vector field position to obtain the energy size of each pixel point position, that is, the final segmentation map.

[0082] E. Morphological and edge connection processing: Fill holes and sequentially connect edges of the image processed in step D, process discontinuous edges and obtain multiple closed areas, such as Figure 7 As shown;

[0083] Step 1: For the hole-filled image, define the initial pixel point in the edge connection process as starting from the upper left corner of the image, and set the point on the edge. Define three movement directions, and the path measurement for determining the quality of the path is expressed as follows:

[0084]

[0085] Where: Γ n is the last path measure, Γ n+1 is the current path measure, p1 is the probability that the pixel is on the edge, and p0 is the probability that the point is any pixel. is the pixel value of the edge point, is the pixel value of any point, P r is the path transition probability, s n+1 The status of the current path.

[0086] Step 2: For step 1, three possible motion paths are obtained. By comparing the path measures, only the largest measure is selected for extension, and three possible motion paths are obtained again. They are compared with the previous paths and the largest path measure is still selected for the next extension until it extends to the boundary of the image or forms a loop, obtaining the final clear edge of the image.

[0087] F. Obtain closed areas and calculate the area for result annotation: Calculate the area of ​​each irregular closed area, determine whether it is a frozen lake based on the threshold, and mark the result on the original image, such as Figure 8 shown.

[0088] A frozen lake detection method based on infrared remote sensing images extracts all texture features of the image through local entropy, and extracts the frozen lake features with weak texture through census transformation. The two are subtracted to obtain the frozen lake without texture features, and the frozen lake is segmented from the background through gradient vector flow. Finally, the morphology and sequential edge connection method are used, and the area threshold is determined to obtain the actual frozen lake.

[0089] An embodiment of the present invention also provides a device for executing a frozen lake detection method based on multi-texture feature fusion, including: a communication bus, a processor and a memory, and when the processor executes computer program instructions, the infrared remote sensing frozen lake detection method as in the above-mentioned embodiment is implemented.

[0090] An embodiment of the present invention provides a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the infrared remote sensing frozen lake detection method in the above embodiment is implemented.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An infrared remote sensing frozen lake detection method, characterized in that: The following steps are involved: Step A. read the infrared remote sensing frozen lake image, extract the image texture features using local entropy, and record the infrared remote sensing frozen lake image as image; Step B. Weak texture feature processing: for the intercepted part of the infrared remote sensing frozen lake image read, respectively as the left and right view images, the left and right view images are subjected to census transformation to extract the weak texture features of the infrared remote sensing frozen lake image; Step C. Texture-free feature processing: subtract the images processed by steps A and B respectively, remove the texture features of the frozen lake to facilitate subsequent gradient calculation, and obtain an infrared remote sensing frozen lake image without texture features; Step D. Separation of frozen lake and background: performing gradient vector flow processing on the image processed in step C to obtain the segmentation result of the frozen lake and the background, that is, to obtain the final segmentation map; Step E. Morphological and edge connection processing: performing hole filling and sequential edge connection on the image processed in step D, processing discontinuous edges and obtaining multiple edge closed areas; Step F. Obtain the closed area and calculate the area to mark the result: Calculate the area of ​​each irregular closed area, determine whether it is a frozen lake based on the threshold, and mark the result on the original image.

2. According to claim 1, a method for detecting frozen lakes by infrared remote sensing, characterized in that: In step A, all texture features in the image are extracted using local entropy.

3. According to claim 1, a method for detecting frozen lakes by infrared remote sensing, characterized in that: In step B, census transformation is used to extract weak texture features of frozen lakes, which is done through the following steps: Step B1: For the infrared remote sensing frozen lake image read, the image width is intercepted as As the left view image , As the right view image , and design odd numbers A sliding window, where is the width of the image, is the maximum parallax; Step 2: For any image, in the sliding window, all pixels except the center are compared with the center pixel. If it is greater than the center, it is recorded as 1, otherwise it is 0, and the final comparison results are accumulated, which can be expressed by the following formula: Where: Represents the center pixel conduct Transform and sum the results, for Neighborhood of Center pixel With adjacent pixels XOR operation; Step 3: For the left view image and the right view image, traverse them simultaneously and compare the differences in the neighborhood for the same pixel position: Where: The parallax is Left and right view images of and The difference value of .

4. The infrared remote sensing frozen lake detection method according to claim 1, characterized in that: In step D, the image is subjected to gradient vector flow calculation to segment the frozen lake from the background. This process can be performed by solving the following equation: Where: is the energy function, is the position obtained by gradient vector flow The corresponding gradient values ​​in two directions are is the gradient of the edge function, is a vector field function; Solve the energy function Minimum vector field function , which is the gradient to be calculated. For each pixel point, the square of the amplitude of the corresponding vector field position is calculated to obtain the energy of each pixel point position, that is, the final segmentation map.

5. The infrared remote sensing frozen lake detection method according to claim 1, characterized in that: In step E, the segmented infrared image is hole-filled and edge-connected to obtain multiple closed areas. This process can be performed through the following steps: Step E1: For the hole-filled image, define the initial pixel point in the edge connection process to start from the upper left corner of the image, and set the pixel point on the edge, define three movement directions, and obtain three movement paths. The path measurement for determining the quality of the path is expressed as follows: Where: is the last path measure, is the current path measure, is the probability that the pixel is on the edge, is the probability that the pixel is any pixel, is the pixel value of the edge pixel, is the pixel value of any point, is the path transition probability, The status of the current path; Step E2: For step E1, three possible motion paths are obtained. By comparing the path measurements, only the largest measurement is selected for extension, and three possible motion paths are obtained again. After comparing with the previous path, the largest path measurement is still selected for the next extension until it extends to the boundary of the image or forms a loop, and the final obvious edge of the image is obtained.

6. An icy lake detection device, characterized in that: include: Communications bus, memory, and processor; Wherein, the communication bus is used to realize the communication line connection between the memory and the processor; The memory is used to store the frozen lake detection program and related processes and results on the processor; The processor is configured to: Input infrared remote sensing frozen lake image; Obtain the local entropy feature map of infrared remote sensing frozen lake images; Acquire weak texture feature map of infrared remote sensing icing image based on census transformation; The local entropy feature map is subtracted from the weak texture feature map to obtain a texture-free feature map; Calculate the gradient vector flow of the texture-free feature map to obtain the segmentation map of the frozen lake and the background; Morphological processing and edge connection are performed on the segmentation map to obtain multiple closed areas, and area calculation and threshold judgment are performed to obtain the final detection results of frozen lakes.

7. A storage medium, characterized in that: The storage medium stores an ice lake detection program, and the program is executed by a processor to implement an infrared remote sensing ice lake detection method as described in any one of claims 1 to 5.

Citation Information

Patent Citations

  • Detection method and system for frozen lake in remote sensing image

    CN107229910A

  • A weighted local entropy infrared small target detection method based on multi-scale morphological fusion

    CN109816641A