An infrared image optimization processing method based on spatial domain hybrid operation

Abstract

The present invention belongs to the field of infrared image processing technology, and more particularly discloses an infrared image optimization processing method based on spatial domain hybrid operation, comprising the following steps: (1) using a dynamic range compression method to enhance details to convert a large dynamic infrared image I wdr Compressed and mapped to 8-bit grayscale image I grey ;(2) to I grey Smoothing filter calculation basic image I bs ;(3)From I grey with I bs Difference extraction of image details I dt ;(4) to I grey Edge detection calculates the image depth factor depth; (5) According to the image details I dt , image depth factor depth, image detail gain G dt And grayscale image I grey Compute detail enhanced image I dt_enh ; (6) Detail enhanced image I dt_enh with I grey According to the enhancement factor R enh Fusion output fused image I fuse ; (7) For the fused image I fuse Edge-preserving and noise-filtering output optimized image I opt The present invention can increase the grayscale number of the main interval, reduce the grayscale number of the transition interval, suppress noise, enhance details and improve image contrast and clarity.

Description

Technical Field

[0001] The invention belongs to the technical field of infrared image processing and relates to a method for optimizing infrared image clarity by hybrid operation in space domain. Background Art

[0002] In infrared thermal imaging application scenarios, when low-temperature and high-temperature objects coexist, the overall dynamic range of their thermal radiation from the lowest temperature to the highest temperature is relatively large, but the temperature distribution within the local range of the object is relatively gentle, and the temperature difference is generally small. In order not to lose the slight changes in the temperature signal while taking into account the large dynamic range, high-performance infrared imaging systems usually use 14-bit or higher-bit A / D sampling and processing for the output signal of the infrared detector to retain the detail information in the scene to the greatest extent. Since display devices generally use 8-bit data for display, the collected 14-bit or higher-bit infrared image data must be dynamically compressed and compressed into 8-bit grayscale data for display. During the compression process, if the compression method is inappropriate, a large amount of detail information implicit in the infrared image data collected by the high-precision A / D will be lost.

[0003] For ordinary scenes with small temperature differences, the grayscale of infrared images will be concentrated in a small range. The linear mapping method can be used to compress the dynamic range to between 0 and 255. For scenes with both high-temperature and low-temperature objects and a large temperature difference between the two, the grayscale distribution characteristics of the infrared image pixels are as follows: (1) The temperature difference in the overall area of ​​the temperature distribution is large, and the grayscale value variation range of the pixel is much larger than the 256 grayscale levels of the 8-bit; (2) The temperature difference in the local area of ​​the temperature distribution is relatively small, and the grayscale value variation range of the pixel is small; (3) There will be a temperature fault in the high and low temperature transition area of ​​the temperature distribution, that is, the number of pixels with some grayscale values ​​in the histogram statistics is zero. If linear mapping is still used for infrared image data of scenes with a large dynamic range, the 256 grayscale levels cannot be effectively used, making the displayed image appear overall dark, low contrast, low signal-to-noise ratio, unclear details, and loss of details, which seriously affects the visual effect of the image.

[0004] In addition, the histogram equalization mapping method pulls the grayscale of pixels with high probability density in the image apart, and compresses the grayscale of pixels with low or zero probability density, so that each grayscale is effectively utilized and the image contrast is enhanced. However, histogram equalization will enhance both the image and the noise composed of pixels with high probability density, while the targets and details with low probability density will be suppressed or even lost.

[0005] Infrared images are temperature images generated by infrared detectors sensing the infrared thermal radiation distribution of the scene. Due to the limitations of detector performance and the influence of temperature spatial distribution characteristics, infrared images often exhibit low signal-to-noise ratio, low contrast, and unclear details.

[0006] The current common infrared image data processing flow is: first enhance the details of the large dynamic range data, then filter the data noise, and finally compress the dynamic range. The disadvantages of this infrared image processing flow are: the processing data bit width is large, the processing scale is large, which is not conducive to real-time circuit processing; in addition, the enhanced details will be lost in the dynamic range compression, the image is dark, and the contrast is low. Summary of the invention

[0007] In order to overcome the deficiencies of the prior art, the present invention provides an infrared image optimization processing method based on spatial domain hybrid operation, which can enhance the details of the infrared image and effectively improve the clarity of the infrared image.

[0008] In order to achieve the above object, the present invention adopts the following technical solution: an infrared image optimization processing method based on spatial domain hybrid operation, comprising the following steps:

[0009] (1) Using the dynamic range compression method to enhance the details, the large dynamic infrared image I wdr Compressed and mapped to 8-bit grayscale image I grey ; The calculation steps are as follows:

[0010] (A) Histogram of AD values ​​of input infrared image;

[0011] (B) Find the maximum AD value X in the histogram where the number of pixels is greater than 0 max With the minimum AD value X min ;

[0012] (C) Calculate the mapped minimum AD value X s0 ;

[0013] (D) Calculate the mapped maximum AD value X s5 ;

[0014] (F) Find the number of s0 , X s5 ] interval with the largest number of pixels AD value X p1 ;

[0015] (F) From X p1 To X s0 Search for the AD value X at the lower boundary of main area 1 s1 ;

[0016] (G) From X p1 To X s5 Search for the AD value X on the upper boundary of main area 1 s2 ;

[0017] (H) Find the number of s2 , X s5 ] The maximum number of pixels in the interval is AD value X p2 ;

[0018] (I) From X p2 To X s2 Search for the AD value X at the lower boundary of main area 2 s3 ;

[0019] (J) From X p2 To X s5 Search for the AD value X on the upper boundary of main area 2 s4 ;

[0020] (K) Calculate the number of main area AD X main1 , X main2 、Number of transition zone AD X trans1 , X trans2 , X trans3 ; In this step, X main1 , X main2 , X trans1 , X trans2 , X trans3 The calculation is performed as follows:

[0021] (i) Number of ADs in primary zone 1: X main1 =X s2 -X s1 ,

[0022] (ii) Number of ADs in primary zone 2: X main2 =X s4 -X s3 ,

[0023] (iii) Number of ADs in transition zone 1: X trans1 =X s1 –X s0 ,

[0024] (iv) Number of ADs in transition region 2: X trans2 =X s3 –X s2 ,

[0025] (v) Number of ADs in transition zone 3: X trans3 =X s5 –X s4 ;

[0026] (L) Calculate the number of grayscales Y in the transition area trans1 , Y trans2 , Y trans3 ;

[0027] (M) Calculate the grayscale number Y of the main area mapping main1 , Y main2 ;

[0028] (N) Calculate AD value X s0 , Xs1 , X s2 , X s3 , X s4 , X s5 The corresponding mapping grayscale Y s0 , Y s1 , Y s2 , Y s3 , Y s4 , Y s5 ;

[0029] (O) Map the AD value of the input infrared image to [Y s0 , Y s5 ] range of gray values, that is, complete I wdr Compressed and mapped to 8-bit grayscale image I grey ;

[0030] (2) to I grey Smoothing filter calculation basic image I bs ;

[0031] (3)From I grey with I bs Difference extraction of image details I dt ;

[0032] (4) to I grey Edge detection calculates the image depth factor depth;

[0033] (5) According to the image details I dt , image depth factor depth, image detail gain G dt And grayscale image I grey Compute detail enhanced image I dtenh ;

[0034] (6) Detail-enhanced image I dt_enh with I grey According to the enhancement ratio R enh Fusion output fused image I fuse ;

[0035] (7) For the fused image I fuse Edge-preserving and noise-filtering output optimized image I opt .

[0036] The present invention adopts a detail-enhanced dynamic range compression method to compress and map a high-bit infrared image to an 8-bit grayscale image, which can effectively enhance infrared image details and improve image clarity. The present invention performs partition processing on the AD value distribution interval in the infrared image histogram, extracts two main intervals and three transition intervals, increases the number of mapped grayscales for the main interval containing a large number of AD value pixels and rich image details, and reduces the number of mapped grayscales for the transition interval containing a small number of AD value pixels, thereby improving the display contrast of the main interval and ensuring that the main and rich image details will not be lost during the grayscale dynamic range compression process.

[0037] Compared with existing infrared image dynamic range compression technologies such as linear compression and histogram equalization compression, the method of the present invention can effectively enhance the detail information of the main temperature area, improve the grayscale utilization rate of the display and the overall clarity of the image. Moreover, compared with the method of optimizing the image processing using high dynamic infrared data, the calculation circuit scale of the real-time image processing system can be significantly reduced, and the calculation complexity can be reduced. There are many mature image processing algorithms for 8-bit grayscale data, which can be conveniently used.

[0038] Furthermore, in step (A), the number of pixels with the k-level grayscale of the input infrared image AD value is counted and recorded as H(X k );

[0039] In the step (F), X s1 The calculation is as follows: From the histogram X p1 Start to X s0 Search, when H(X k )<1 or X k =X s0 When the corresponding AD value is X s1 ;

[0040] In the step (G), X s2 The calculation is as follows: From the histogram X p1 Start to X s5 Search, when H(X k )<1 or X k =X s5 When the corresponding AD value is X s2 .

[0041] The step (H), X p2 The calculation is as follows: In the histogram (X s2 , X s5 ]Interval search pixel maximum value H(X p2 ), the corresponding AD value is X p2 .

[0042] The step (I) is calculated as follows: from the histogram X p2 Start to X s2 Search, when H(X k )<1 or X k =X s2 When AD value is X s3 ;

[0043] The step (J), X s4 The calculation is as follows: From the histogram X p2 Start to X s5 Search, when H(X k )<1 or X k =X s5 When AD value is X s4 .

[0044] Furthermore, in step (L), the maximum grayscale of the transition zone is recorded as Y trans_max , the value is 10, Y trans1 , Y trans1 , Y trans1 The calculation is performed as follows:

[0045] (a) When X trans1 >Y trans_max , Y trans1 =Y trans_max , otherwise Y trans1 =X trans1 ,

[0046] (b) When X trans2 >Y trans_max , Y trans2 =Y trans_max , otherwise Y trans2 =X trans2 ,

[0047] (c) When X trans3 >Y trans_max , Y trans3 =Y trans_max , otherwise Y trans3 =X trans3 .

[0048] Furthermore, in the step (M), the maximum value of the mapped grayscale is recorded as Y max , Y main1 , Y main2 The calculation is performed as follows:

[0049] (a)X main =X main1 +X main2 ;

[0050] (b)Ymain =Y max +1-Y trans1 -Y trans2 -Y trans3 ;

[0051] (c)

[0052] (d)

[0053] Furthermore, in step (C), the minimum AD value X is mapped s0 The calculation is as follows: From the histogram X min Start to accumulate the number of pixels at each gray level in sequence. When the accumulated value exceeds N, the AD value is recorded as X s0 ; Where N is height / 2, and height is the number of pixels in a column of the image.

[0054] Furthermore, the smoothing filter in step (2) includes Gaussian filtering, mean filtering, median filtering and bilateral filtering;

[0055] In the step (4), the edge detection method is Roberts operator detection, prewitt operator detection, sobel operator detection, Kirsch operator detection, Laplace operator detection, Log operator detection or Canny operator detection.

[0056] Furthermore, in step (3), I dt The calculation formula is:

[0057] I dt =I grey -I bs .

[0058] Furthermore, in step (5), the detail enhanced image I dt_enh Calculated using the following formula:

[0059] I dt_enh =I grey +I dt ×depth×G dt .

[0060] Furthermore, in step (6), the fusion image I fuse Calculated using the following formula:

[0061] I fuse =(1-R enh )×I grey +R enh ×I dt , R enh ∈[0, 1].

[0062] Furthermore, in step (7), the edge-preserving noise filtering process is performed using a surface blurring algorithm, and the algorithm formula is as follows:

[0063]

[0064] Among them, r is the surface blur radius, TH is the surface blur threshold, I fuse1 is the current pixel value, I fuse (i) is the i-th pixel value in the template, I opt Calculates the resulting value for the current pixel.

[0065] In summary, the present invention can increase the grayscale number of the main interval, reduce the grayscale number of the transition interval, suppress noise, enhance details and improve image contrast and clarity. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a schematic diagram of the histogram statistical partitioning of the present invention.

[0067] Figure 2 It is the dynamic range compression mapping curve of the present invention.

[0068] Figure 3 This is a block diagram of infrared image data processing of the present invention. DETAILED DESCRIPTION

[0069] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0070] like Figure 1-3 As shown, an infrared image optimization processing method based on spatial domain hybrid operation includes the following steps:

[0071] (1) Using the dynamic range compression method to enhance the details, the large dynamic infrared image I wdr Compressed and mapped to 8-bit grayscale image I grey ; The calculation steps are as follows:

[0072] (A) Statistical histogram of the AD value of the input infrared image; Statistical histogram of the AD value of the input infrared image at each gray level X k The number of occurrences is recorded as H(X k ). Where k is the index number, k = 0, 1, 2, ..., L-1, X k Indicates the kth gray level of the image to be processed. L is the total number of gray levels in the image. When the data width is 14 bits, L = 2 14 =16384.

[0073] (B) Find the maximum AD value X in the histogram where the number of pixels is greater than 0max With the minimum AD value X min ; Among them, X min , X max The calculation is as follows: Start from 0 and search upwards for the minimum AD value in the histogram where the number of pixels is greater than 0, record it as X min , taking 14 bits as an example, start from 16383 and search downwards for the maximum AD value in the histogram where the number of pixels is greater than 0, record it as X max .

[0074] (C) Calculate the mapped minimum AD value X s0 ; The mapped minimum AD value X s0 The calculation is as follows: From the histogram X min Start to accumulate the number of pixels of each gray level in sequence, and the number of pixels is the number of pixels of the AD value. When the accumulated value exceeds N, the AD value is recorded as X s0 ; Where N is height / 2, and height is the number of pixels in a column of the image.

[0075] (D) Calculate the mapped maximum AD value X s5 ; The mapped maximum AD value X s5 The calculation is as follows: From the histogram X max The number of pixels at each gray level is accumulated downwards. The number of pixels at each gray level is the number of pixels of the AD value. When the accumulated value exceeds N, the AD value is recorded as X. s5 ; Where N is height / 2, and height is the number of pixels in a column of the image.

[0076] (E) Find the number of s0 , X s5 ] interval with the largest number of pixels AD value X p1 ;X p1 The calculation is as follows: In the histogram [X s0 , X s5 ] interval search pixel number maximum value H(X p1 ), the corresponding AD value is X p1 .

[0077] (F) From X p1 To X s0 Search for the AD value X at the lower boundary of main area 1 s1 ; This step starts from the histogram X p1 Start to X s0 Search, when H(X k )<1 or X k =X s0 When the corresponding AD value is X s1 .

[0078] (G) From X p1 To X s5 Search for the AD value X on the upper boundary of main area 1 s2 ; From the histogram X p1 Start to X s5 Search, when H(X k )<1 or X k =X s5 When the corresponding AD value is X s2 .

[0079] (H) Find [X s2 , X s5 ] The maximum number of pixels in the interval is AD value X p2 ; In the histogram (X s2 , X s5 ]Interval search pixel maximum value H(X p2 ), the corresponding AD value is X p2 .

[0080] (I) From X p2 To X s2 Search for the AD value X at the lower boundary of main area 2 s3 ;X s3 The calculation is as follows: From the histogram X p2 Start to X s2 Search, when H(X k )<1 or X k =X s2 When AD value is X s3 ;

[0081] (J) From X p2 To X s5 Search for the AD value X on the upper boundary of main area 2 s4 ;X s4 The calculation is as follows: From the histogram X p2 Start to X s5 Search, when H(X k )<1 or X k =X s5 When AD value is X s4 ;

[0082] (K) Calculate the number of main area AD X main1 , X main2 、Number of transition zone AD X trans1 , X trans2 , X trans3 ; In this step, X main1 , X main2 , X trans1 , X trans2 , X trans3 The calculation is performed as follows:

[0083] (i) Number of ADs in primary zone 1: X main1 =X s2 -X s1 ;

[0084] (ii) Number of ADs in primary zone 2: X main2 =X s4 -X s3 ;

[0085] (iii) Number of ADs in transition zone 1: X trans1 =X s1 –X s0 ;

[0086] (iv) Number of ADs in transition region 2: X trans2 =X s3 –X s2 ;

[0087] (v) Number of ADs in transition zone 3: X trans3 =X s5 –X s4 .

[0088] (L) Calculate the number of grayscales Y in the transition area trans1 , Y trans2 , Y trans3 ; The maximum grayscale number in the transition zone is recorded as Y trans_max , take the value as 10, calculate the mapping grayscale number Y in the transition area trans1 , Y trans2 , Y trans3 :

[0089] (i) When X trans1 >Y trans_max , Y trans1 =Y trans_max , otherwise Y trans1 =X trans1 ;

[0090] (ii) When X trans2 >Y trans_max , Y trans2 =Y trans_max , otherwise Y trans2 =X trans2 ;

[0091] (iii) When X trans3 >Y trans_max , Y trans3 =Y trans_max , otherwise Y trans3 =X trans3 .

[0092] (M) Calculate the grayscale number Y of the main area mapping main1 , Y main2 ; The maximum value of the mapped grayscale is Y max , calculate the mapping grayscale number Y of the main area main1 , Y main2 Here’s how:

[0093] (a)X main =X main1 +X main2 ;

[0094] (b)Y main =Y max +1-Y trans1 -Y trans2 -Y trans3 ;

[0095] (c)

[0096] (d)

[0097] (N) Calculate AD value X s0 , X s1 , X s2 , X s3 , X s4 , X s5 The corresponding mapping grayscale Y s0 , Y s1 , Y s2 , Y s3 , Y s4 , Y s5 , as shown below:

[0098] (a)Y s0 =0,

[0099] (b)Y s1 =Y trans1 ,

[0100] (c)Y s2 =Y trans1 +Y main1 -1,

[0101] (d)Y s3 =Y trans1 +Y main1 +Y trans2 -1,

[0102] (e)Y s4 =Y trans1 +Y main1 +Y trans2 +Y main2 -1,

[0103] (f)Y s5 =Y trans1 +Y main1 +Y trans2 +Y main2 +Y trans3 -1=Y max .

[0104] (O) Piecewise linear mapping of the AD value of the input infrared image to [Y s0 , Y s5 ] range of gray values, that is, complete I wdr Compressed and mapped to 8-bit grayscale image I grey ; Mapping is performed according to the following formula:

[0105]

[0106] (2) to I grey Smoothing filter calculation basic image I bs ; Smoothing filtering is a spatial low-pass filtering method for images, which is used to smooth the image. Smoothing filtering includes Gaussian filtering, mean filtering, median filtering and bilateral filtering. Taking the 5X5 window template mean filtering as an example, I bs The calculation formula is:

[0107]

[0108] (3)From I grey with I bs Difference extraction of image details I dt ; I in this step dt The calculation formula is:

[0109] I dt =I grey -I bs .

[0110] (4) to I grey Edge detection calculates the image depth factor depth; the edge detection methods in this step include: Roberts operator detection, prewitt operator detection, sobel operator detection, Kirsch operator detection, Laplace operator detection, Log operator detection, Canny operator detection, etc. Taking the sobel operator edge detection as an example, the depth factor depth calculation is performed according to the following steps:

[0111] (a) Calculate the image horizontal gradient factor:

[0112]

[0113] (b) Calculate the vertical gradient factor of the image:

[0114]

[0115] (c) Calculate the image gradient factor:

[0116]

[0117] (d) Find the maximum gradient factor G max ;

[0118] (e) Calculate the depth factor depth:

[0119] depth = G / G max .

[0120] (5) According to the image details I dt , image depth factor depth, image detail gain G dt And grayscale image I grey Compute detail enhanced image I dt_enh ; Among them, the detail enhanced image I dt_enh Calculated using the following formula:

[0121] I dt_enh =I grey +I dt ×depth×G dt .

[0122] (6) Detail-enhanced image I dt_enh with I grey According to the enhancement ratio R enh Fusion output fused image I fuse Among them, the fused image I fuse Calculated using the following formula:

[0123] I fuse =(1-R enh )×I grey +R enh ×I dt , R enh ∈[0, 1].

[0124] (7) For the fused image I fuse Edge-preserving and noise-filtering output optimized image I opt In this step, the edge-preserving noise filtering process is processed using a surface blurring algorithm, and the algorithm formula is as follows:

[0125]

[0126] Among them, r is the surface blur radius, TH is the surface blur threshold, I fuse1 is the current pixel value, I fuse (i) is the i-th pixel value in the template, I optCalculates the resulting value for the current pixel.

[0127] The method of the present invention can effectively enhance details, suppress noise, and improve image clarity.

[0128] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

Claims

1. A method for optimizing infrared image processing based on spatial domain hybrid operation, comprising the following steps: (1) Using the dynamic range compression method to enhance the details, the large dynamic infrared image I wdr Compressed and mapped to 8-bit grayscale image I grey ; The calculation steps are as follows: (A) Histogram of AD values ​​of input infrared image; (B) Find the maximum AD value X in the histogram where the number of pixels is greater than 0 max With the minimum AD value X min ; (C) Calculate the mapped minimum AD value X s0 ; (D) Calculate the mapped maximum AD value X s5 ; (E) Find the number of s0 ,X s5 ] interval with the largest number of pixels AD value X p1 ; (F) From X p1 To X s0 Search for the AD value X at the lower boundary of main area 1 s1 ; (G) From X p1 To X s5 Search for the AD value X on the upper boundary of main area 1 s2 ; (H) Find the number of s2 ,X s5 ] The maximum number of pixels in the interval is AD value X p2 ; (I) From X p2 To X s2 Search for the AD value X at the lower boundary of main area 2 s3 ; (J) From X p2 To X s5 Search for the AD value X on the upper boundary of main area 2 s4 ; (K) Calculate the number of main area AD X main1 , X main2 、Number of transition zone AD X trans1 , X trans2 , X trans3 ; In this step, X main1 , X main2 , X trans1 , X trans2 , X trans3 The calculation is performed as follows: (i) Number of ADs in primary zone 1: X main1 =X s2 -X s1 , (ii) Number of ADs in primary zone 2: X main2 =X s4 -X s3 , (iii) Number of ADs in transition zone 1: X trans1 =X s1 –X s0 , (iv) Number of ADs in transition region 2: X trans2 =X s3 –X s2 , (v) Number of ADs in transition zone 3: X trans3 =X s5 –X s4 ; (L) Calculate the number of grayscales Y in the transition area trans1 , Y trans2 , Y trans3 , where Y trans1 is the number of grayscales mapped in transition zone 1, Y trans2 is the number of grayscales mapped in transition zone 2, Y trans3 is the number of mapped gray levels in transition zone 3; (M) Calculate the grayscale number Y of the main area mapping main1 , Y main2 ,in, Y main1 is the grayscale number of the main area 1, Y main2 The mapping grayscale number of main area 2; (N) Calculate AD value X s0 , X s1 , X s2 , X s3 , X s4 , X s5 The corresponding mapping grayscale Y s0 , Y s1 , Y s2 , Y s3 , Y s4 , Y s5 ; (O) Map the AD value of the input infrared image to [Y s0 ,Y s5 ] range of gray values, that is, complete I wdr Compressed and mapped to 8-bit grayscale image I grey ; (2) to I grey Smoothing filter calculation basic image I bs ; (3)From I grey with I bs Difference Extraction of Image Details I dt ; (4) to I grey Edge detection calculates the image depth factor depth; (5) According to the image details I dt , image depth factor depth, image detail gain G dt And grayscale image I grey Compute detail enhanced image I dt_enh ; (6) Detail-enhanced image I dt_enh with I grey According to the enhancement factor R enh Fusion output fused image I fuse ; (7) For the fused image I fuse Edge-preserving and noise-filtering output optimized image I opt .

2. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: In the step (A), the number of pixels with k-level grayscale in the input infrared image AD value is counted and recorded as H(X k ); In the step (F), X s1 The calculation is as follows: From the histogram X p1 Start to X s0 Search, when H(X k )<1 or X k =X s0 When the corresponding AD value is X s1 ; In the step (G), X s2 The calculation is as follows: From the histogram X p1 Start to X s5 Search, when H(X k )<1 or X k =X s5 When the corresponding AD value is X s2 ; The step (H), X p2 The calculation is as follows: In the histogram [X s2 ,X s5 ]Interval search pixel maximum value H(X p2 ), the corresponding AD value is X p2 ; Among them, X s2 is the AD value of the upper boundary of main area 1, X s5 is the maximum AD value mapped; The step (I) is calculated as follows: from the histogram X p2 Start to X s2 Search, when H(X k )<1 or X k =X s2 When AD value is X s3 ; The step (J), X s4 The calculation is as follows: From the histogram X p2 Start to X s5 Search, when H(X k )<1 or X k =X s5 When AD value is X s4 .

3. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: In the step (L), the maximum grayscale of the transition zone is recorded as Y trans_max , the value is 10, Y trans1 , Y trans2 , Y trans3 The calculation is performed as follows: (a) When X trans1 >Y trans_max , Y trans1 =Y trans_max , otherwise Y trans1 =X trans1 , (b) When X trans2 >Y trans_max , Y trans2 =Y trans_max , otherwise Y trans2 =X trans2 , (c) When X trans3 >Y trans_max , Y trans3 =Y trans_max , otherwise Y trans3 =X trans3 .

4. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: In the step (M), the maximum value of the mapped grayscale is recorded as Y max , Y main1 , Y main2 The calculation is performed as follows: (a)X main =X main1 +X main2 ; (b)Y main =And max +1-Y trans1 -AND trans2 -AND trans3 ; 5. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: In step (C), the minimum AD value X is mapped s0 The calculation is as follows: From the histogram X min Start to accumulate the number of pixels at each gray level in sequence. When the accumulated value exceeds N, the AD value is recorded as X s0 ; Where N is height / 2, and height is the number of pixels in a column of the image.

6. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: The smoothing filter in step (2) includes Gaussian filtering, mean filtering, median filtering and bilateral filtering; In the step (4), the edge detection method is Roberts operator detection, prewitt operator detection, sobel operator detection, Kirsch operator detection, Laplace operator detection, Log operator detection or Canny operator detection.

7. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: The step (3), I dt The calculation formula is: I dt =I grey -I bs 。 8. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: In step (5), the detail enhanced image I dt_enh Calculated using the following formula: I dt_enh =I grey +I dt ×depth×G dt 。 9. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: In step (6), the fusion image I fuse Calculated using the following formula: I fuse =(1-R enh )×I grey +R enh ×I dt ,R enh ∈[0,1]。 10. The infrared image optimization processing method based on spatial domain hybrid operation according to claim 1 is characterized in that: In step (7), the edge-preserving noise filtering process is performed using a surface blurring algorithm, and the algorithm formula is as follows: Among them, r is the surface blur radius, TH is the surface blur threshold, I fuse1 is the current pixel value, I fuse (i) is the i-th pixel value in the template, I opt Calculates the resulting value for the current pixel.

Images