A method for scene correction of infrared images based on random difference

Abstract

The present invention relates to an infrared image scene correction method based on random differences, belonging to the field of infrared image processing. The method can simultaneously perform uniform scene recognition calculation and non-uniformity correction processing in the infrared image scene correction processing, thereby making up for the shortcomings of traditional infrared image scene correction methods. A method for searching a uniform scene pixel column set based on random differences and taking minimization of result image non-uniformity as an objective function is proposed, thereby forming a more stable and effective infrared image scene correction method, which can effectively identify uniform scene areas in infrared images, maximize the preservation of original image information, and obtain stable and high-quality scene correction result images.

Description

Technical Field

[0001] The invention belongs to the field of infrared image processing and relates to an infrared image scene correction method based on random difference. Background Art

[0002] Due to the influence of manufacturing processes and materials, the response transfer function of each pixel in an infrared detector varies, resulting in striped noise in the acquired infrared image, known as non-uniformity noise. This noise severely degrades image quality and significantly hinders the widespread application of infrared imaging in fields such as medicine, surveillance, agriculture, and forestry. Therefore, non-uniformity correction is essential for improving infrared image quality in infrared remote sensing applications.

[0003] There are two main types of infrared image non-uniformity correction methods: calibration-based methods and scene-based methods. Calibration-based methods require a standard radiation source, such as a blackbody, which has significant limitations in practical applications. Scene-based methods extract parameters based on the characteristics of the image in the scene to achieve non-uniformity correction. They have better applicability and are widely used. Scene-based correction methods generally assume that the first-order and second-order statistics of the pixel response in the time domain are the same over a sufficiently long period of time. However, in practical applications, this assumption is no longer met due to the influence of the captured image content and detector temperature drift. This makes the scene-based correction method unstable and unsuitable for real-time infrared image processing applications. Summary of the Invention

[0004] The technical problem solved by the present invention is: to overcome the shortcomings of the existing technology and propose an infrared image scene correction method based on random difference, which can effectively identify uniform scene areas in infrared images, maximize the preservation of original image information, and obtain stable and high-quality scene correction result images.

[0005] The solution of the present invention is:

[0006] A method for infrared image scene correction based on random difference includes the following steps:

[0007] Step 1: Input an infrared image X with m rows and n columns, and set the initial correction map Y, let Y = X; establish an initial uniform scene column set C = [1, 2, ..., n], and the optimal uniform scene column set Set iteration count b = 1;

[0008] Step 2: Set the initial column number k = 2 and calculate the random difference and D of the kth column k ;

[0009] Step 3: judge k. When k < n, set k = k + 1 and return to step 2. When k ≥ n, go to step 4.

[0010] Step 4: Calculate random differences and D k The mean

[0011] Step 5. Find the random difference and D k All values ​​less than the mean Subsequence of

[0012] Step 6: Sort from small to large to get the updated subsequence S is the updated subsequence The element number in ;

[0013] Step 7: Update the uniform scene column set C to the random difference sum D k All less than The labeled subsequence [C1, C2, ..., C q ];

[0014] Step 8: Columnize the optimal uniform scene Updated to C with The intersection of

[0015] Step 9: Update the correction image Y based on the uniform scene column set C;

[0016] Step 10: Calculate the optimal uniform scene sequence set The corresponding result graph Z;

[0017] Step 11: Calculate the non-uniformity value e of the current result graph Z b ;

[0018] Step 12: Judge b. When b>1, and |e b-1 -e b |<1.0e -5 When , output the result graph Z; otherwise, let b=b+1 and judge the updated b; when When , the output result graph Z; when , return to step 2.

[0019] In the above infrared image scene correction method based on random difference, in step 2, random difference and D k The calculation method is:

[0020]

[0021] Where T is the candidate sequence; the elements in T are 1, 2, ..., m; and the elements in T are randomly arranged.

[0022] In the above-mentioned infrared image scene correction method based on random difference, in step 4, the mean The calculation method is:

[0023]

[0024] In the above-mentioned infrared image scene correction method based on random difference, in step seven, For Round down.

[0025] In the above-mentioned infrared image scene correction method based on random difference, in step nine, the updating method of the correction map Y is:

[0026]

[0027] Where q is the number of elements in the labeled subsequence;

[0028] 1≤i≤m;

[0029] 1≤j≤n.

[0030] In the above-mentioned infrared image scene correction method based on random difference, in step 10, the calculation method of the result image Z is:

[0031]

[0032] Where h is the optimal uniform scene column set The number of elements in

[0033] 1≤i≤m;

[0034] 1≤j≤n.

[0035] In the above-mentioned infrared image scene correction method based on random difference, in the step 11, the non-uniformity value e of the current result image Z is b The calculation method is:

[0036]

[0037] The beneficial effects of the present invention compared with the prior art are:

[0038] (1) The present invention can simultaneously perform uniform scene recognition calculation and non-uniformity correction processing in the infrared image scene correction process, thereby making up for the shortcomings of the traditional infrared image scene correction method;

[0039] (2) The present invention proposes a method for searching uniform scene pixel column sets based on random differences and minimizing the non-uniformity of the result image as the objective function, forming a more stable and effective infrared image scene correction method, which can effectively identify uniform scene areas in infrared images, maximize the preservation of original image information, and obtain stable and high-quality scene correction result images. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flow chart of infrared image scene correction of the present invention;

[0041] Figure 2 For the random difference and curve graph of the present invention;

[0042] Figure 3 This is a graph showing the convergence of the uniform number of columns of the present invention;

[0043] Figure 4 This is the infrared image scene correction result diagram of the present invention. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the embodiments.

[0045] The present invention provides an infrared image scene correction method based on random differences, which can simultaneously perform uniform scene recognition calculation and non-uniformity correction processing in the infrared image scene correction processing, making up for the shortcomings of traditional infrared image scene correction methods. A uniform scene pixel column set search method based on random differences and minimizing the non-uniformity of the result image is proposed, forming a more stable and effective infrared image scene correction method, which can effectively identify uniform scene areas in infrared images, maximize the preservation of original image information, and obtain stable and high-quality scene correction result images.

[0046] A method for scene correction of infrared images based on random difference, such as Figure 1 As shown, the specific steps include:

[0047] Step 1: Input an infrared image X with m rows and n columns, and set the initial correction map Y, let Y = X; establish an initial uniform scene column set C = [1, 2, ..., n], and the optimal uniform scene column set Set the iteration count b=1.

[0048] Step 2: Set the initial column number k = 2 and calculate the random difference and D of the kth column k ; Random Difference and D k The calculation method is:

[0049]

[0050] Where T is the candidate sequence; the elements in T are 1, 2, ..., m; and the elements in T are randomly arranged.

[0051] Step 3: Judge k. When k < n, set k = k + 1 and return to step 2. When k ≥ n, go to step 4.

[0052] Step 4: Calculate random differences and D k The mean mean The calculation method is:

[0053]

[0054] Step 5. Find the random difference and D k All values ​​less than the mean Subsequence of

[0055] Step 6: Sort from small to large to get the updated subsequence S is the updated subsequence The element number in .

[0056] Step 7: Update the uniform scene column set C to the random difference sum D k All less than The labeled subsequence [C1, C2, ..., C q ]; For Round down.

[0057] Step 8: Columnize the optimal uniform scene Updated to C with The intersection of

[0058] Step 9: Update the correction image Y based on the uniform scene sequence set C. The updating method of the correction image Y is:

[0059]

[0060] Where q is the number of elements in the labeled subsequence;

[0061] 1≤i≤m;

[0062] 1≤j≤n.

[0063] Step 10: Calculate the optimal uniform scene sequence set The corresponding result graph Z; the calculation method of the result graph Z is:

[0064]

[0065] Where h is the optimal uniform scene column set The number of elements in

[0066] 1≤i≤m;

[0067] 1≤j≤n.

[0068] Step 11: Calculate the non-uniformity value e of the current result graph Z b ; The non-uniformity value e of the current result map Z b The calculation method is:

[0069]

[0070] Step 12: Judge b. When b>1, and |e b-1 -e b |<1.0e -5 When , output the result graph Z; otherwise, let b=b+1 and judge the updated b; when When , the output result graph Z; when , return to step 2.

[0071] Example

[0072] Taking the 256×2800 12-bit infrared image X to be scene corrected as an example, the process of the infrared image scene correction method based on random difference is as follows: Figure 1 As shown, the specific process is as follows:

[0073] (1) According to the input m-row n-column infrared odd pixel image X, set the initial correction image Y = X, the initial uniform scene column set C = [1, 2, ..., n], and the optimal uniform scene column set Iteration count b = 1; in this example, m = 256, n = 2800,

[0074] (2) Set the initial column number k = 2;

[0075] (3) Generate a random integer sequence T from 1 to m; T is a sequence of m numbers, which can be generated by randomly selecting a number from the alternative sequence {1, 2, ..., n}, and deleting a number from the alternative sequence as it is selected, until there are no numbers left in the alternative sequence.

[0076] (4) Calculate the random difference sum of the kth column

[0077] Random difference and curve diagram as shown Figure 2 As shown;

[0078] (5) If k < n, set k = k + 1 and return to step (3); otherwise, go to step (6);

[0079] (6) Calculate the mean of random differences and D

[0080] (7) Find all random differences and all values ​​in D that are less than the mean Subsequence of

[0081] (8) Yes Sort to get subsequences

[0082] (9) Update the uniform scene column set C to the random difference and all values ​​in D that are less than the mean The labeled subsequence [C1, C2, ..., C q ];

[0083] (10) Updated to C and intersection of The uniform column number convergence curve is as follows Figure 3 shown.

[0084] (11) Update the correction map Y based on the uniform scene column set C, Where q is the number of elements in set C, 1≤i≤m, 1≤j≤n;

[0085] (12) Calculate the optimal uniform scene sequence set Corresponding result graph Z, Where h is the set The number of elements in , 1≤i≤m, 1≤j≤n;

[0086] (13) Calculate the non-uniformity value of the current result graph Z

[0087] The non-uniform convergence curve is as follows Figure 3 shown.

[0088] (14) If b>1 and |e b-1 -e b |<1.0e-5, then output the result graph Z;

[0089] (15) Let b = b + 1, if Output the result graph Z, otherwise, return to step (2).

[0090] The result of infrared image scene correction is shown in the figure below: Figure 4 shown.

[0091] Although the present invention has been disclosed above in terms of preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solutions of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention shall fall within the scope of protection of the technical solutions of the present invention.

Claims

1. A method for scene correction of infrared images based on random differences, characterized by: The steps include: Step 1: Input an infrared image X with m rows and n columns, and set the initial correction map Y, let Y = X; establish an initial uniform scene column set C = [1, 2, ..., n], and set the optimal uniform scene column set Set iteration count b = 1; Step 2: Set the initial column number k = 2 and calculate the random difference and D of the kth column k ; Step 3: judge k. When k < n, set k = k + 1 and return to step 2. When k ≥ n, go to step 4. Step 4: Calculate random differences and D k The mean Step 5. Find the random difference and D k All values ​​less than the mean Subsequence of Step 6: Sort from small to large to get the updated subsequence S is the updated subsequence The element number in ; Step 7: Update the uniform scene column set C to the random difference sum D k All less than The labeled subsequence [C1, C2, ..., C q ]; q is the number of elements in the labeled subsequence; Step 8: Columnize the optimal uniform scene Updated to C with The intersection of Step 9: Update the correction image Y based on the uniform scene column set C; Step 10: Calculate the optimal uniform scene sequence set The corresponding result graph Z; Step 11: Calculate the non-uniformity value e of the current result graph Z b ; Step 12: Judge b. When b>1, and |e b-1 -e b |<1.0e -5 When , output the result graph Z; otherwise, let b=b+1 and judge the updated b; when When , the output result graph Z; when , return to step 2.

2. The infrared image scene correction method based on random difference according to claim 1, characterized in that: In the step 2, random difference and D k The calculation method is: Where T is the candidate sequence; the elements in T are 1, 2, ..., m; and the elements in T are randomly arranged.

3. The infrared image scene correction method based on random difference according to claim 2, characterized in that: In the step 4, the mean The calculation method is:

4. The infrared image scene correction method based on random difference according to claim 3, characterized in that: In the step seven, For Round down.

5. The infrared image scene correction method based on random difference according to claim 4, characterized in that: In step nine, the updating method of the correction map Y is: Where q is the number of elements in the labeled subsequence; 1≤i≤m; 1≤j≤n.

6. The infrared image scene correction method based on random difference according to claim 5, characterized in that: In step 10, the calculation method of the result graph Z is: Where h is the optimal uniform scene column set The number of elements in 1≤i≤m; 1≤j≤n.

7. The infrared image scene correction method based on random difference according to claim 6, characterized in that: In the step 11, the non-uniformity value e of the current result graph Z is b The calculation method is:

Images