CN117422631A - Infrared image enhancement method based on adaptive filtering layering - Google Patents
Infrared image enhancement method based on adaptive filtering layering Download PDFInfo
- Publication number
- CN117422631A CN117422631A CN202311565110.8A CN202311565110A CN117422631A CN 117422631 A CN117422631 A CN 117422631A CN 202311565110 A CN202311565110 A CN 202311565110A CN 117422631 A CN117422631 A CN 117422631A
- Authority
- CN
- China
- Prior art keywords
- image
- base layer
- layer image
- entropy
- adaptive
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 23
- 238000000034 method Methods 0.000 title claims abstract description 22
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 18
- 230000002146 bilateral effect Effects 0.000 claims abstract description 16
- 230000011218 segmentation Effects 0.000 claims description 13
- 230000004927 fusion Effects 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 5
- 230000009466 transformation Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention relates to an infrared image enhancement method based on adaptive filtering layering. The invention aims to improve the visibility and the resolvable property of an infrared image by improving an SSR theoretical center surrounding function and adopting two main measures of a proper enhancement algorithm for a layered image. The algorithm mainly comprises three steps of SSR image layering based on self-adaptive bilateral filtering, base layer image enhancement and detail layer image enhancement. The algorithm has good enhancement effect in the data set in the image contrast histogram equalization algorithm HE.
Description
Technical field:
the invention belongs to the field of image processing, and provides an infrared enhancement method of a self-adaptive filtering layering, aiming at the problem of low contrast weak texture expressive property caused by unreasonable distribution of infrared image gray level. Specifically, image layering is realized by adopting a bilateral filtering SSR (simple-Scale Retinex) theory fused with local information entropy, and contrast enhancement and detail enhancement of an infrared image are realized by utilizing a double histogram equalization algorithm based on maximum entropy threshold segmentation and an adaptive Gamma transformation algorithm so as to improve the quality and the identifiability of the image.
The background technology is as follows:
the infrared image is generated by capturing heat radiated by an object through an infrared sensor, and is widely applied to the fields of military reconnaissance, monitoring, medical diagnosis, industrial detection and the like. However, due to the restrictions of environmental factors, equipment performance and the like, the infrared image often has the problems of low image quality, insufficient contrast, loss of detail information and the like, which brings great trouble to subsequent works such as feature extraction, identification or tracking. Therefore, the infrared image is processed by adopting an effective enhancement method so as to improve the visibility and the resolvable property of the infrared image, and the infrared image has important practical significance.
The method for enhancing the infrared image mainly comprises the following steps: histogram-based enhancement and hierarchical enhancement of images. Histogram-based image enhancement methods are a technique that optimizes image contrast and brightness by adjusting image histogram distribution. Among the many histogram-based enhancement methods are standard histogram equalization, dual histogram equalization, plateau histogram equalization, adaptive histogram equalization, contrast-limiting adaptive histogram equalization, and the like. The histogram equalization method can effectively improve the contrast of the infrared image, so that the targets and the background in the image are more obvious. However, the histogram-based image enhancement method generally has a common problem that spatial distribution of gray levels of an image and correlation thereof are ignored, and the redistribution of the gray levels destroys the spatial correlation of partial pixels, so that the local texture detail characteristics of the image cannot be represented. Image-based layered enhancement stems from the Retinex theory, which believes that the human visual system can adapt to different lighting conditions by contrast and color constancy, and that images can be decomposed into base layer images and detail layer images. The infrared image enhancement algorithm based on the Retinex theory comprises a single-scale Retinex algorithm SSR. A multi-scale Retinex algorithm MSR and a multi-scale MSRCR algorithm with color recovery. The Retinex algorithm has obvious advantages, can easily enhance the contrast of the image, and the methods mainly carry out illumination estimation through a Gaussian convolution template, have the defects of color cast, color distortion, halation and the like of the integral enhancement effect of the image because the method does not have the capability of retaining edges, and can not well display the original information of the image.
In order to further improve the brightness and contrast of the infrared image and simultaneously maintain the natural effect of the image, the invention highlights the infrared image enhancement method based on the adaptive filtering layering, can avoid the problems of underenhancement, overenhancement and low contrast in the infrared image enhancement, and can maintain good visual effect while improving the image quality in various scenes. Firstly, the self-adaptive bilateral filtering is used as a center surrounding function of an SSR theory, and the edge characteristics of the image are effectively reserved while denoising is performed. And secondly, a double-histogram equalization algorithm based on the maximum entropy is adopted to enhance the base layer image, so that the gray level combination and detail loss condition of the histogram can be measured, and the contrast is enhanced most moderately. In addition, the Gamma transformation method which is adaptive to the brightness distribution of the detail layer image is further adopted to strengthen the detail layer image. And finally, fusing the base layer image and the detail layer image to obtain a final enhanced image.
The invention is a means aiming at the problems of low overall contrast and insufficient detail display of an infrared image by adopting a corresponding enhancement algorithm from the angle of filtering layering, wherein low-frequency information of which a base layer image is an image contains large dynamic information, and high-frequency information of which a detail layer image is an image contains characteristic difference of small dynamic information.
The invention comprises the following steps:
according to the invention, an original image is decomposed into a base layer image and a detail layer image by adopting a self-adaptive bilateral filtering SSR algorithm fused with local information entropy, then the contrast and brightness presentation of the base layer are improved by utilizing a double histogram equalization algorithm based on maximum entropy, the detail layer texture is enhanced by utilizing Gamma transformation, and finally the enhanced base layer image and detail layer image are subjected to Retinex inverse transformation to obtain an enhanced infrared image, so that the visibility and the target detection effect of the infrared image are improved.
The technical scheme adopted by the invention is as follows:
the first step: reading an original infrared image I, wherein the size of the image is M multiplied by N, M is the number of width pixels of the image I, and N is the number of height pixels of the image I;
and a second step of: decomposing the image into a base layer image L and a detail layer image R based on an SSR theory thought by a formula (1), and in actual processing, equivalently transforming the formula (1) into a logarithmic domain by a formula (2);
I=L×R (1)
ln I=ln L+ln R (2)
and a third step of: calculating an adaptive bilateral filter function f (x, y) fused with local information entropy through formulas (3) - (7); omega (x, y) is a neighborhood pixel set with a pixel (x, y) as a center point, and pixel point coordinates in omega (x, y) can be expressed as (I, j), I (x, y) is a (x, y) pixel value, and I (I, j) is a (I, j) pixel value; calculating a spatial domain kernel function w s (i, j) and a gray domain kernel function wr (i, j), where σ s Is the spatial domain standard deviation; sigma (sigma) r Is the gray domain standard deviation; to improve the edge preserving capability of bilateral filtering, a weight parameter alpha is introduced, an adaptive bilateral filtering kernel function w (I, j) is calculated by using a formula (6), wherein Entropy (I) is the Entropy of image information, and Entropy [ omega (x, y)]Entropy is set for the neighborhood pixels;
w(i,j)=(1-α)w s (i,j)+αw r (i,j) (6)
fourth step: taking the self-adaptive bilateral filtering function f obtained in the formula (7) as a center surrounding function of an SSR theory, and filtering an original infrared image through the formula (8) to obtain a base layer image L';
fifth step: removing the base layer image from the original infrared image through the method (9) to obtain a detail layer image R';
R′=exp(ln I-ln L) (9)
sixth step: based on the assumption of the segmentation threshold t ', the base layer image L' is segmented into a background region and a target region, and the background region information entropy H is calculated by the equations (10) and (11), respectively B (t') target region information entropy H O (t') wherein p g Probability for each gray level pixel value; p (P) t' The gray level distribution probability of the background area is that l is the gray level of the image, and l is 256; equation (12) calculates a segmentation threshold that maximizes the sum of the background region information entropy and the target region information entropy, denoted as a maximum entropy segmentation threshold t me ;
t me =arg max(H O (t′)+H B (t′)) (12)
Seventh step: calculating a data segmentation point t of a base layer image L' by adopting an Ostu algorithm ostu Will t me And t ostu In combination, a dual-histogram equalization method based on maximum entropy is adopted to enhance the base layer image L' to obtain a base layer enhanced image L e ;
Eighth step: enhancement of detail layer image R' by adaptive Gamma using equations (13) - (14) to obtain detail layer enhanced image R e Gamma is the brightness distribution characteristic of the self-adaptive detail layer image, h i I=0, 1,2, …, L-1 is defined as the gray variable of the gray histogram of the base layer image L', parameter C is defined as the dark area pixel range, C takes 90;
R e =R γ (13)
ninth step: equation (15) performs the Retinex inverse transform fusion of the base layer image and the detail layer image,
I e =exp(ln L e +ln R e ) (15)
obtaining enhanced image I e 。
The invention has the advantages that:
1. to avoid the problems of under-enhancement, over-enhancement, low contrast and the like in the enhancement of the infrared image;
2. reliable enhancement effect is obtained in various complex scenes;
3. the enhanced image has clearer visual effect and richer detail information
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 (a) is an original infrared image;
fig. 2 (b) shows the base layer image L after the fourth step of processing;
fig. 2 (c) is a detail layer image R after the fifth step of processing;
FIG. 3 (b) shows the base layer enhanced image L after the seventh step of processing e ;
FIG. 3 (c) shows the detail layer enhanced image R after the eighth step e ;
FIG. 3 (a) shows an IR enhanced image I after the ninth step e 。
Fig. 4 shows simulation comparison results of the method and the histogram equalization algorithm in different scenes.
The specific embodiment is as follows:
the present invention will be described in detail with reference to specific examples.
1: reading an original infrared image I, wherein the size of the image is M multiplied by N, M is the number of width pixels of the image I, and N is the number of height pixels of the image I;
2: decomposing the image into a base layer image L and a detail layer image R based on an SSR theory thought by a formula (1), and in actual processing, equivalently transforming the formula (1) into a logarithmic domain by a formula (2);
I=L×R (1)
ln I=ln L+ln R (2)
3: calculating an adaptive bilateral filter function f (x, y) fused with local information entropy through formulas (3) - (7); omega (x, y) is a neighborhood pixel set with a pixel (x, y) as a center point, and pixel point coordinates in omega (x, y) can be expressed as (I, j), I (x, y) is a (x, y) pixel value, and I (I, j) is a (I, j) pixel value; calculating a spatial domain kernel function w s (i, j) and a gray domain kernel function wr (i, j), where σ s Is the spatial domain standard deviation; sigma (sigma) r Is the gray domain standard deviation; to improve the edge preserving capability of bilateral filtering, a weight parameter alpha is introduced, an adaptive bilateral filtering kernel function w (I, j) is calculated by using a formula (6), wherein Entropy (I) is the Entropy of image information, and Entropy [ omega (x, y)]Entropy is set for the neighborhood pixels;
w(i,j)=(1-α)w s (i,j)+αw r (i,j) (6)
4: taking the self-adaptive bilateral filtering function f obtained in the formula (7) as a center surrounding function of an SSR theory, and filtering an original infrared image through the formula (8) to obtain a base layer image L';
5: removing the base layer image from the original infrared image through the method (9) to obtain a detail layer image R';
R′=exp(ln I-ln L) (9)
6: based on the assumption of the segmentation threshold t ', the base layer image L' is segmented into a background region and a target region, and the background region information entropy H is calculated by the equations (10) and (11), respectively B (t') target region information entropy H O (t') wherein p g Probability for each gray level pixel value; p (P) t' The gray level distribution probability of the background area is that l is the gray level of the image, and l is 256; equation (12) calculates a segmentation threshold that maximizes the sum of the background region information entropy and the target region information entropy, denoted as a maximum entropy segmentation threshold t me ;
t me =arg max(H O (t′)+H B (t′)) (12)
7: calculating a data segmentation point t of a base layer image L' by adopting an Ostu algorithm ostu Will t me And t ostu In combination, a dual-histogram equalization method based on maximum entropy is adopted to enhance the base layer image L' to obtain a base layer enhanced image L e ;
8: enhancement of detail layer image R' by adaptive Gamma using equations (13) - (14) to obtain detail layer enhanced image R e Gamma is the brightness distribution characteristic of the self-adaptive detail layer image, h i I=0, 1,2, …, L-1 is defined as the gray variable of the gray histogram of the base layer image L', parameter C is defined as the dark area pixel range, C takes 90;
R e =R γ (13)
9: equation (15) performs the Retinex inverse transform fusion of the base layer image and the detail layer image,
I e =exp(ln L e +ln R e ) (15)
obtaining enhanced image I e 。
Claims (1)
1. The infrared image enhancement method based on the adaptive filtering layering is characterized by comprising the following steps of:
the first step: reading an original infrared image I, wherein the size of the image is M multiplied by N, M is the number of width pixels of the image I, and N is the number of height pixels of the image I;
and a second step of: decomposing the image into a base layer image L and a detail layer image R based on an SSR theory thought by a formula (1), and in actual processing, equivalently transforming the formula (1) into a logarithmic domain by a formula (2);
I=L×R (1)
lnI=lnL+lnR (2)
and a third step of: calculating an adaptive bilateral filter function f (x, y) fused with local information entropy through formulas (3) - (7); omega (x, y) is a neighborhood pixel set with a pixel (x, y) as a center point, and pixel point coordinates in omega (x, y) can be expressed as (I, j), I (x, y) is a (x, y) pixel value, and I (I, j) is a (I, j) pixel value; calculating a spatial domain kernel function w s (i, j) and a gray domain kernel function wr (i, j), where σ s Is the spatial domain standard deviation; sigma (sigma) r Is the gray domain standard deviation; to improve the edge preserving capability of bilateral filtering, a weight parameter alpha is introduced, an adaptive bilateral filtering kernel function w (I, j) is calculated by using a formula (6), wherein Entropy (I) is the Entropy of image information, and Entropy [ omega (x, y)]Entropy is set for the neighborhood pixels;
w(i,j)=(1-α)w s (i,j)+αw r (i,j) (6)
fourth step: taking the self-adaptive bilateral filtering function f obtained in the formula (7) as a center surrounding function of an SSR theory, and filtering an original infrared image through the formula (8) to obtain a base layer image L';
fifth step: removing the base layer image from the original infrared image through the method (9) to obtain a detail layer image R';
R′=exp(lnI-lnL) (9)
sixth step: based on the assumption of the segmentation threshold t ', the base layer image L' is segmented into a background region and a target region, and the background region information entropy H is calculated by the equations (10) and (11), respectively B (t') target region information entropy H O (t') wherein p g Probability for each gray level pixel value; p (P) t' The gray level distribution probability of the background area is that l is the gray level of the image, and l is 256; equation (12) calculates a segmentation threshold that maximizes the sum of the background region information entropy and the target region information entropy, denoted as a maximum entropy segmentation threshold t me ;
t me =argmax(H O (t′)+H B (t′)) (12)
Seventh step: calculating a data segmentation point t of a base layer image L' by adopting an Ostu algorithm ostu Will t me And t ostu In combination, a dual-histogram equalization method based on maximum entropy is adopted to enhance the base layer image L' to obtain a base layer enhanced image L e ;
Eighth step: enhancement of detail layer image R' by adaptive Gamma using equations (13) - (14) to obtain detail layer enhanced image R e Gamma is the brightness distribution characteristic of the self-adaptive detail layer image, h i I=0, 1,2, …, L-1 is defined as the gray variable of the gray histogram of the base layer image L', parameter C is defined as the dark area pixel range, C takes 90;
R e =R γ (13)
ninth step: equation (15) performs the Retinex inverse transform fusion of the base layer image and the detail layer image,
I e =exp(lnL e +lnR e ) (15)
finally, the enhanced image I is obtained e 。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311565110.8A CN117422631A (en) | 2023-11-22 | 2023-11-22 | Infrared image enhancement method based on adaptive filtering layering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311565110.8A CN117422631A (en) | 2023-11-22 | 2023-11-22 | Infrared image enhancement method based on adaptive filtering layering |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117422631A true CN117422631A (en) | 2024-01-19 |
Family
ID=89526561
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311565110.8A Pending CN117422631A (en) | 2023-11-22 | 2023-11-22 | Infrared image enhancement method based on adaptive filtering layering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117422631A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117893742A (en) * | 2024-03-15 | 2024-04-16 | 深圳大学总医院 | Uveitis image feature recognition method |
-
2023
- 2023-11-22 CN CN202311565110.8A patent/CN117422631A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117893742A (en) * | 2024-03-15 | 2024-04-16 | 深圳大学总医院 | Uveitis image feature recognition method |
CN117893742B (en) * | 2024-03-15 | 2024-05-28 | 深圳大学总医院 | Uveitis image feature recognition method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Kim et al. | Optimized contrast enhancement for real-time image and video dehazing | |
CN109064426B (en) | Method and device for suppressing glare in low-illumination image and enhancing image | |
CN102222328B (en) | Edge-preserving self-adaptive weighted filtering method for natural scene images | |
CN107527332A (en) | Enhancement Method is kept based on the low-light (level) image color for improving Retinex | |
Ma et al. | An effective fusion defogging approach for single sea fog image | |
Wang et al. | Low-illumination image enhancement for night-time UAV pedestrian detection | |
Li et al. | Low-light image enhancement via pair of complementary gamma functions by fusion | |
CN113222877B (en) | Infrared and visible light image fusion method and application thereof in airborne photoelectric video | |
CN110428389B (en) | Low-light-level image enhancement method based on MSR theory and exposure fusion | |
CN117422631A (en) | Infrared image enhancement method based on adaptive filtering layering | |
Parihar et al. | A comprehensive analysis of fusion-based image enhancement techniques | |
CN113344810A (en) | Image enhancement method based on dynamic data distribution | |
Liu et al. | Enhancement of low illumination images based on an optimal hyperbolic tangent profile | |
CN115587945A (en) | High dynamic infrared image detail enhancement method, system and computer storage medium | |
CN111311503A (en) | Night low-brightness image enhancement system | |
Wei et al. | An image fusion dehazing algorithm based on dark channel prior and retinex | |
CN113592750A (en) | Infrared enhancement method based on gradient histogram | |
CN117252773A (en) | Image enhancement method and system based on self-adaptive color correction and guided filtering | |
CN116309233A (en) | Infrared and visible light image fusion method based on night vision enhancement | |
CN116630198A (en) | Multi-scale fusion underwater image enhancement method combining self-adaptive gamma correction | |
CN116452447A (en) | Low-illumination high-definition image processing method | |
CN110992287A (en) | Method for clarifying non-uniform illumination video | |
CN110930358A (en) | Solar panel image processing method based on self-adaptive algorithm | |
CN115564682A (en) | Uneven-illumination image enhancement method and system | |
CN113240611B (en) | Foreign matter detection method based on picture sequence |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |