A dual-band based infrared image blind pixel compensation method

Abstract

The application relates to a kind of double-band based infrared image blind cell compensation methods, belong to double-band infrared image processing technical field.The method includes: the double-band infrared image of input X i ( i , j )、 Y i ( i , j ) respectively carry out blind cell detection;Edge extraction is carried out to the double-band infrared image, and edge image Ex i , j 、 Ey i , j ;According to the detection condition and extraction condition, the pixel condition of the double-band infrared image is classified, and the image is output after the pixel is compensated according to the classification condition.The application method can improve the accuracy of blind cell detection, improve the accuracy of blind cell compensation, and effectively improve the compensation effect of large blind cell area in the image by using the difference of double-band background radiation information.​​

Description

Technical Field

[0001] This invention belongs to the field of dual-band infrared image processing technology, specifically relating to a method for blind pixel compensation in infrared images based on dual-band technology. Background Technology

[0002] Infrared focal plane arrays often exhibit blind pixels due to manufacturing processes and other factors. These blind pixels include fixed and random blind pixels, which typically appear as dark or bright spots in the image. Blind pixels affect the image quality of the detector, thus requiring blind pixel detection and compensation for the focal plane array.

[0003] Dual-band infrared focal plane arrays, including mid-to-long wavelength, mid-to-short wavelength, and long-to-short wavelength bands, offer significant advantages over traditional single-band infrared focal plane arrays. In the extremely complex and variable environments of modern battlefields, dual-band infrared focal plane arrays can acquire more target information, improving the probability of target detection and identification. However, dual-band infrared focal plane arrays still face the challenge of blind pixels.

[0004] Previously, domestic research on blind pixel correction was usually at the single-band level, with virtually no research on blind pixel compensation algorithms based on dual-band. Due to the lack of real scene information, blind pixel compensation algorithms based on single-band algorithms compensate blind pixels by using scene information from the neighborhood. This approach is not very effective for large blind pixel clusters (such as blind pixel clusters larger than 7×7) and blind pixels located at the edge.

[0005] Therefore, overcoming the shortcomings of existing technologies is a pressing issue that needs to be addressed in the field of dual-band infrared image processing technology. (Invention Content)

[0006] The purpose of this invention is to address the shortcomings of existing technologies and provide a method for blind pixel compensation in infrared images based on dual-band technology. By utilizing the difference in background radiation information between the two bands, blind pixels in any single band can be better compensated, thereby improving the blind pixel phenomenon in existing dual-band infrared images.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] A method for blind pixel compensation in dual-band infrared images includes the following steps:

[0009] Step S1, process the input dual-band infrared image X i (i, j), Y i Blind pixel detection is performed on (i, j) respectively;

[0010] Step S2: Extract the edges of the dual-band infrared image and extract the edge images Ex(i,j) and Ey(i,j);

[0011] Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

[0012] Furthermore, preferably, the dual-band infrared image includes mid-to-long band, mid-to-short band, short-to-long band, and mid-to-medium band images.

[0013] Furthermore, preferably, the infrared image blind pixel compensation method based on dual-band infrared image includes the following steps:

[0014] Step S1, process the input dual-band infrared image X i (i, j), Y i Blind pixel detection is performed on (i, j) respectively. A threshold h is set and the initial blind pixel table is set to 1. If a certain band pixel (i, j) is judged as a blind pixel for h consecutive frames, the corresponding band blind pixel table bpx(i, j) or bpy(i, j) is set to 0.

[0015] Step S2: Extract the edges of the dual-band infrared image, extract the edge images Ex(i,j) and Ey(i,j), set the edge parts to 0 and the non-edge parts to 1;

[0016] Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

[0017] 0 and 1 are simply labels for blind cells and edge states. 0 represents a blind cell or edge, and 1 represents a non-blind cell or edge, which facilitates the corresponding identification calculations during compensation.

[0018] Furthermore, preferably, in step S2, the Sobel operator is used for edge extraction.

[0019] Furthermore, preferably, in step S3, the pixel situation of the dual-band infrared image is classified, specifically as follows:

[0020] (1) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge;

[0021] (2) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge;

[0022] (3) The pixel is a blind pixel in both bands and the pixel is not located at the edge;

[0023] (4) The pixel is a blind pixel in both bands and the pixel is located at the edge.

[0024] Furthermore, preferably, in step S3, different strategies are adopted to compensate the pixels according to the classification situation. When the pixel is a blind pixel in both bands, a single-band compensation strategy is adopted to compensate the blind pixel, as shown in (1) and (2); when the pixel is a blind pixel only in one band and not in the other band, the neighborhood similarity of the blind pixel in the two bands is used as the compensation coefficient, and the information of the two bands is used to compensate the blind pixel, as shown in (3) and (4); the specific compensation process is as follows:

[0025] (1) Pixel (i, j) is a blind pixel in both bands and is not located at the edge:

[0026]

[0027] (2) Pixel (i, j) is a blind pixel in both bands and is located at the edge:

[0028]

[0029] (3) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge:

[0030]

[0031] At this time, X o (i, j) represents the pixel value at position (i, j) in the output image of the band containing the blind pixel after blind pixel compensation; the other band does not require compensation.

[0032] (4) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge:

[0033]

[0034] At this time, X o (i, j) represents the pixel value at position (i, j) in the output image of the band containing the blind pixel after blind pixel compensation; the other band does not require compensation.

[0035] Among them, X i (i, j), Y i (i, j) represents the pixel value at position (i, j) in the dual-band input image; X o (i, j), Y o (i, j) represents the pixel value at position (i, j) in the dual-band output image after blind pixel compensation. m and n represent the m×n neighborhood around pixel (i, j). Ex(i, j) and Ey(i, j) are the pixel values ​​at position (i, j) in the edge image. There are no restrictions on the range of values ​​for m and n, as long as they are smaller than the size of the infrared image.

[0036] Blind neighborhood similarity is achieved through cosine similarity dist(X). imn Y imn To measure, dist(X) imn Y imn The calculation formula is as follows:

[0037]

[0038] When the similarity is high, the proportion of information from the other band that can be used for blind pixel compensation is large; when the similarity is low, the proportion of information from the other band is small.

[0039] The blind pixel compensation method used in this invention leverages the advantages of dual-band infrared focal plane arrays, simultaneously utilizing information from both bands to compensate for blind pixels. When a pixel is a blind pixel in one band and a non-blind pixel in the other, neighborhood information from the blind pixel band and real-world scene information from the non-blind pixel band can be used simultaneously for blind pixel compensation. This helps improve the compensation effect for large blind pixel clusters (such as clusters larger than 7×7) and blind pixels at edge locations, and compared to single-band blind pixel compensation algorithms, it can better recover the true location of blind pixels.

[0040] Compared with the prior art, the beneficial effects of this invention are as follows:

[0041] The blind pixel compensation method used in this invention retains the advantages of single-band blind pixel compensation algorithms in terms of speed and convenience, while also leveraging the advantages of dual-band infrared focal plane arrays. It can simultaneously utilize information from both bands to compensate for blind pixels. This invention first classifies the distribution of blind pixels in both bands and then uses different compensation strategies accordingly. When a pixel is a blind pixel in one band and not in the other, it can simultaneously use neighborhood information from the blind pixel band and real-world scene information from the non-blind pixel band for compensation. This helps improve the compensation effect for large blind pixel clusters (such as clusters larger than 7×7) and blind pixels at edge positions, and compared to single-band blind pixel compensation algorithms, it can better recover the true location of blind pixels. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart of the dual-band infrared image blind pixel compensation method of the present invention;

[0044] Figure 2The image shows a mid-to-long wavelength image with blind elements, based on the dual-band infrared image blind element compensation method of the present invention; where (a) is a long-wave image and (b) is a mid-wave image.

[0045] Figure 3 The diagram shows the compensation effect; where (a) represents the method used in the existing method; and (b) represents the method used in this invention. Detailed Implementation

[0046] The present invention will now be described in further detail with reference to the embodiments.

[0047] Those skilled in the art will understand that the following embodiments are for illustrative purposes only and should not be construed as limiting the scope of the invention. Where specific techniques or conditions are not specified in the embodiments, they are performed in accordance with the techniques or conditions described in the literature in the field or according to the product instructions. Materials or equipment whose manufacturers are not specified are all conventional products that can be obtained by purchase.

[0048] Example 1

[0049] A method for blind pixel compensation in dual-band infrared images includes the following steps:

[0050] Step S1, process the input dual-band infrared image X i (i, j), Y i Blind pixel detection is performed on (i, j) respectively;

[0051] Step S2: Extract the edges of the dual-band infrared image and extract the edge images Ex(i,j) and Ey(i,j);

[0052] Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

[0053] Example 2

[0054] A method for blind pixel compensation in dual-band infrared images includes the following steps:

[0055] Step S1, process the input dual-band infrared image X i (i, j), Y i Blind pixel detection is performed on (i, j) respectively. A threshold h is set and the initial blind pixel table is set to 1. If a certain band pixel (i, j) is judged as a blind pixel for h consecutive frames, the corresponding band blind pixel table bpx(i, j) or bpy(i, j) is set to 0.

[0056] Step S2: Extract the edges of the dual-band infrared image, extract the edge images Ex(i,j) and Ey(i,j), set the edge parts to 0 and the non-edge parts to 1;

[0057] Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

[0058] The dual-band infrared images include medium-long band, medium-short band, short-long band, and medium-medium band images.

[0059] Example 3

[0060] A method for blind pixel compensation in dual-band infrared images includes the following steps:

[0061] Step S1, process the input dual-band infrared image X i (i, j), Y i Blind pixel detection is performed on (i, j) respectively. A threshold h is set and the initial blind pixel table is set to 1. If a certain band pixel (i, j) is judged as a blind pixel for h consecutive frames, the corresponding band blind pixel table bpx(i, j) or bpy(i, j) is set to 0.

[0062] Step S2: Extract the edges of the dual-band infrared image, extract the edge images Ex(i,j) and Ey(i,j), set the edge parts to 0 and the non-edge parts to 1;

[0063] Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

[0064] The dual-band infrared images include medium-long band, medium-short band, short-long band, and medium-medium band images.

[0065] In step S2, the Sobel operator is used for edge extraction.

[0066] In step S3, the pixel situation of the dual-band infrared image is classified, specifically as follows:

[0067] (1) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge;

[0068] (2) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge;

[0069] (3) The pixel is a blind pixel in both bands and the pixel is not located at the edge;

[0070] (4) The pixel is a blind pixel in both bands and the pixel is located at the edge.

[0071] Example 4

[0072] A method for blind pixel compensation in dual-band infrared images includes the following steps:

[0073] Step S1, process the input dual-band infrared image X i (i, j), Y i Blind pixel detection is performed on pixels (i, j) respectively. A threshold h is set, and the initial blind pixel table is set to 1. If a pixel (i, j) in a certain band is identified as a blind pixel for h consecutive frames, then the corresponding band blind pixel table bpx(i, j) or bpy(i, j) is set to 0.

[0074] Step S2: Extract the edges of the dual-band infrared image, extract the edge images Ex(i,j) and Ey(i,j), set the edge parts to 0 and the non-edge parts to 1;

[0075] Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

[0076] The dual-band infrared images include medium-long band, medium-short band, short-long band, and medium-medium band images.

[0077] In step S2, the Sobel operator is used for edge extraction.

[0078] In step S3, the pixel situation of the dual-band infrared image is classified, specifically as follows:

[0079] (1) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge;

[0080] (2) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge;

[0081] (3) The pixel is a blind pixel in both bands and the pixel is not located at the edge;

[0082] (4) The pixel is a blind pixel in both bands and the pixel is located at the edge.

[0083] In step S3, different strategies are adopted to compensate the pixels according to the classification situation. When the pixel is a blind pixel in both bands, a single-band compensation strategy is adopted to compensate the blind pixel, as shown in (1) and (2). When the pixel is a blind pixel only in one band and not in the other band, the neighborhood similarity of the blind pixel in the two bands is used as the compensation coefficient, and the information of the two bands is used to compensate the blind pixel, as shown in (3) and (4). The specific compensation process is as follows:

[0084] (1) Pixel (i, j) is a blind pixel in both bands and is not located at the edge:

[0085]

[0086] (2) Pixel (i, j) is a blind pixel in both bands and is located at the edge:

[0087]

[0088] (3) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge:

[0089]

[0090] (4) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge:

[0091]

[0092] Among them, X i (i, j), Y i (i, j) represents the pixel value at position (i, j) in the dual-band input image; X o (i, j), Y o (i,j) is the pixel value at position (i,j) in the dual-band output image after blind pixel compensation, m and n represent the m×n neighborhood around pixel (i,j), and Ex(i,j) and Ey(i,j) are the pixel values ​​at position (i,j) in the edge image.

[0093] Blind neighborhood similarity is achieved through cosine similarity dist(X). imn Y imn To measure, dist(X) imn Y imn The calculation formula is as follows:

[0094]

[0095] Application Examples

[0096] Combination Figure 1 Taking a 384×288 pixel mid / long band infrared image as an example, the method includes:

[0097] Step S1: Input a 384×288 pixel mid / long band infrared image, set the threshold h to 10, and perform mid-wave blind pixel detection and long-wave blind pixel detection on it respectively. If a certain band pixel (i, j) is judged as a blind pixel for 10 consecutive frames, then the corresponding band blind pixel table bpx(i, j) or bpy(i, j) is set to 0.

[0098] Step S2: Use the Sobel operator to extract edges from the mid-wave and long-wave images respectively, and set the edge portion of the extracted edge image to 0 and the non-edge portion to 1.

[0099] Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

[0100] Where m = 3, n = 3.

[0101] In step S3, the pixel situation of the dual-band infrared image is classified, specifically as follows:

[0102] (1) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge;

[0103] (2) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge;

[0104] (3) The pixel is a blind pixel in both bands and the pixel is not located at the edge;

[0105] (4) The pixel is a blind pixel in both bands and the pixel is located at the edge.

[0106] In step S3, different strategies are adopted to compensate the pixels according to the classification situation. When the pixel is a blind pixel in both bands, a single-band compensation strategy is adopted to compensate the blind pixel, as shown in (1) and (2). When the pixel is a blind pixel only in one band and not in the other band, the neighborhood similarity of the blind pixel in the two bands is used as the compensation coefficient, and the information of the two bands is used to compensate the blind pixel, as shown in (3) and (4). The specific compensation process is as follows:

[0107] (1) Pixel (i, j) is a blind pixel in both bands and is not located at the edge:

[0108]

[0109] (2) Pixel (i, j) is a blind pixel in both bands and is located at the edge:

[0110]

[0111] (3) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge:

[0112]

[0113] (4) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge:

[0114]

[0115] Among them, X i (i, j), Y i (i, j) represents the pixel value at position (i, j) in the dual-band input image; X o (i, j), Y o (i,j) is the pixel value at position (i,j) in the dual-band output image after blind pixel compensation, m and n represent the m×n neighborhood around pixel (i,j), and Ex(i,j) and Ey(i,j) are the pixel values ​​at position (i,j) in the edge image.

[0116] Blind neighborhood similarity is achieved through cosine similarity dist(X). imn Y imn To measure, dist(X) imn Y imn The calculation formula is as follows:

[0117]

[0118] Example images Figure 2 As shown, the mid-wave image has no blind elements, while the long-wave image contains scattered small blind elements and a 7×10 cluster of large blind elements.

[0119] Renderings Figure 3 As shown. The compensation effect in this example is as follows. Figure 3 As shown in (b), compared to the results of the usual single-band compensation algorithm, Figure 3 As shown in (a), the compensation effect of this invention for scattered small blind elements is similar to that of single-band compensation, but the compensation effect for 7×10 large blind element clusters is significantly better than that of single-band compensation algorithm.

[0120] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for blind pixel compensation in dual-band infrared images, characterized in that, Includes the following steps: Step S1, process the input dual-band infrared image X i (i,j), Y i Blind cell detection is performed on (i,j) respectively; Step S2: Extract the edges of the dual-band infrared image and extract the edge images Ex(i,j) and Ey(i,j); Step S3: Based on the detection results of Step S1 and the extraction results of Step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification results, and then output the image. The classification of pixel information in dual-band infrared images is as follows: (1) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge; (2) A pixel is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge; (3) The pixel is a blind pixel in both bands and the pixel is not located at the edge; (4) The pixel is a blind pixel in both bands and the pixel is located at the edge; Depending on the classification, different strategies are adopted to compensate for the pixels. When a pixel is a blind pixel in both bands, a single-band compensation strategy is adopted to compensate for the blind pixel, as shown in (1) and (2). When a pixel is a blind pixel only in one band and not in the other band, the neighborhood similarity of the blind pixel in the two bands is used as the compensation coefficient, and the information of the two bands is used to compensate for the blind pixel, as shown in (3) and (4). The specific compensation process is as follows: (1) Pixel (i, j) is a blind pixel in both bands and is not located at the edge: (2) Pixel (i, j) is a blind pixel in both bands and is located at the edge: (3) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is not located at the edge: (4) Pixel (i, j) is a blind pixel in one band and a non-blind pixel in another band, and this pixel is located at the edge: Among them, X i (i,j), Y i (i,j) represents the pixel value at position (i,j) in the dual-band input image; X o (i,j), Y o (i,j) is the pixel value at position (i,j) in the dual-band output image after blind pixel compensation, m and n represent the m×n neighborhood around pixel (i,j), and Ex(i,j) and Ey(i,j) are the pixel values ​​at position (i,j) in the edge image. Blind neighborhood similarity is achieved through cosine similarity dist(X). imn ,Y imn To measure, dist(X) imn ,Y imn The calculation formula is as follows:

2. The infrared image blind pixel compensation method based on dual-band according to claim 1, characterized in that, The dual-band infrared images include medium-long band, medium-short band, short-long band, and medium-medium band images.

3. The infrared image blind pixel compensation method based on dual-band according to claim 1, characterized in that, Includes the following steps: Step S1, process the input dual-band infrared image X i (i,j), Y i Blind pixel detection is performed on (i,j) respectively. A threshold h is set and the initial blind pixel table is set to 1. If a certain band pixel (i,j) is judged as a blind pixel for h consecutive frames, the corresponding band blind pixel table bpx(i,j) or bpy(i,j) is set to 0. Step S2: Extract the edges of the dual-band infrared image, extract the edge images Ex(i,j) and Ey(i,j), set the edge parts to 0 and the non-edge parts to 1; Step S3: Based on the detection results of step S1 and the extraction results of step S2, classify the pixel situation of the dual-band infrared image, compensate the pixels according to the classification, and output the image.

4. The infrared image blind pixel compensation method based on dual-band according to claim 1, characterized in that, In step S2, the Sobel operator is used for edge extraction.

Images