CN113947535A

CN113947535A - Low-illumination image enhancement method based on illumination component optimization

Info

Publication number: CN113947535A
Application number: CN202010694542.9A
Authority: CN
Inventors: 王正勇; 龙庆延; 何小海; 卿粼波; 吴小强; 滕奇志; 吴晓红
Original assignee: Sichuan University
Current assignee: Sichuan University
Priority date: 2020-07-17
Filing date: 2020-07-17
Publication date: 2022-01-18
Anticipated expiration: 2040-07-17
Also published as: CN113947535B

Abstract

The invention provides a low-illumination image enhancement method based on illumination component optimization. Firstly, estimating an initial illumination component by using a max-RGB model for a low-illumination original image, and optimizing the initial illumination component by using singular value decomposition and guide filtering to obtain a final illumination map. According to Retinex theory, the low-illumination original image and the final illumination image are divided point by point, and the G channel in RGB three channels of the low-illumination original image is used for guiding, filtering and removing noise of the obtained image. Because singular value decomposition and guide filtering are combined, the illumination component can be estimated more accurately, and the subjective visual effect and various objective indexes of the enhanced picture are improved. The effectiveness of the method is verified in a plurality of image enhancement experiments.

Description

Low-illumination image enhancement method based on illumination component optimization

Technical Field

The invention relates to a low-illumination image enhancement problem in the field of image processing, in particular to a low-illumination image enhancement method based on illumination component optimization.

Background

Low-illumination image enhancement has been a research hotspot in the field of image processing. Under the condition of lacking good illumination conditions, the image acquisition process is inevitably influenced by a plurality of factors such as self parameters of equipment, illumination change and the like, so that the image shooting effect is poor, the problems of uneven illumination, color distortion and the like occur, and certain influence is generated on the subsequent processing such as target detection, target identification and the like. Therefore, how to obtain a high quality image under low illumination conditions is a crucial issue.

Classical low image enhancement methods can be divided into three categories: histogram transform based, Retinex model based and defogging model based. The histogram conversion achieves an enhanced effect by expanding the gray level of the original image and increasing the dynamic range of the image. The Retinex model is based on the principle of color constancy, and the traditional method estimates the illumination in the original image by using Gaussian filtering and removes the illumination by using logarithmic transformation to obtain an enhanced image. In the defogging model, the image is inverted to obtain an effect similar to the foggy image, and the image is inverted again after being processed by the defogging algorithm to obtain an enhanced image. The three methods all make certain progress, but the visual effect and objective index are still not ideal.

Disclosure of Invention

Aiming at the problem that the low-illumination image enhancement result is not ideal, the invention provides a low-illumination image enhancement method based on illumination component optimization. And optimizing initial illumination by using singular value decomposition and guide filtering, and denoising the enhanced image by using the guide filtering again after the enhanced image is obtained by simplifying the Retinex model. The present invention achieves the above object by the following processes:

(1) obtaining an initial illumination component diagram by using a max-RGB model for the original image with low illumination;

(2) processing the initial illumination component graph obtained in the step (1) by using singular value decomposition, and normalizing;

(3) using three-time guided filtering to the result after normalization in the step (2) to obtain an optimized illumination component diagram;

(4) according to the simplified Retinex model, dividing RGB three channels of the low-illumination original image by the illumination component image obtained in the step (3) point by point to obtain an enhanced image;

(5) and (3) taking a G channel in RGB three channels of the low-illumination original image as a guide image, and carrying out denoising processing on the enhanced image obtained in the step (4) to obtain a finally required enhanced image.

Drawings

FIG. 1 is a block diagram of a low-illumination image enhancement method based on illumination component optimization;

Detailed Description

The invention is further described below with reference to the accompanying drawings:

the illumination map calculation and optimization method specifically comprises the following steps:

(1) acquisition of initial illumination map

The initial illumination map is extracted by adopting a max-RGB model, and the calculation formula is as follows:

L(x,y)＝max(R(x,y),G(x,y),B(x,y)) (1)

in the formula, R (x, y), G (x, y), and B (x, y) are RGB three-channel images of the low-illuminance original image, respectively.

(2) Optimization of illumination maps

Singular value decomposition is a common matrix diagonalization decomposition method, and by using singular value decomposition, a matrix a can be decomposed into the following form:

A＝UΣV^T (2)

where U and V are both orthogonal matrices, i.e. UU for matrix U, V^TI and VV^TWhen the matrix is a negative diagonal matrix, the orthogonal matrices U and V contain structural information of the original matrix, and the singular matrix Σ contains energy information of the original matrix. The singular matrix Σ can be expressed as:

Σ＝diag(s₁,s₂,…,s_N)(s₁＞s₂＞…＞s_N) (3)

in the formula, s₁,s₂,…,s_NAre the singular values of the original matrix a.

After singular value decomposition is carried out on a 3 multiplied by 3 area in an initial illumination image, S in a singular matrix is selected₁I.e. the largest singular value as the illumination intensity at that point. And after the whole image is processed, carrying out normalization.

The guide filter is a commonly used edge-preserving filter, which can well preserve the edge of the image while removing the texture detail information. To further optimize the illumination map, the illumination map after singular value decomposition and normalization is reprocessed using cubic guided filtering. If the initial light map is L_ini(x, y), the illumination map obtained after singular value decomposition and normalization is L_s(x,y)，GF_R(A, B) denotes guided filtering with a window radius R, there is the following equation:

wherein the window size is r₁＝15、r₂＝7、r₃＝3，L₃(x, y) is the optimized final used illumination pattern.

The enhanced image is obtained and the denoising method is as follows:

(1) according to Retinex theory, a low-illumination image S (x, y) can be expressed as the product of an illumination component L (x, y) and a reflection component R (x, y), i.e.:

S(x,y)＝L(x,y)×R(x,y) (5)

the above formula can be modified into:

in the formula, τ is a constant that prevents the denominator from being zero, and is taken to be 0.01.

(2) An enhanced image can be obtained according to equation (6). The low-illumination image amplifies the noise originally hidden in the dark while enhancing, and denoising processing is needed after enhancing. The currently used CCD camera and CMOS camera mostly use a bayer array sensor to acquire a color image, and the bayer array is a 4 × 4 array composed of 8 green, 4 blue, and 4 red pixels. Therefore, the noise level of the green channel in the original image is usually low, the green channel can be used as a guide image, the guide filtering is used for denoising RGB three channels of the enhanced result respectively, and the final result is obtained after the three channels are fused.

In order to verify the effectiveness of the low-illumination image enhancement method based on illumination component optimization, experiments are carried out on 20 graphs, and comparison experiments are carried out on the classical low-illumination image enhancement method MSRCR, Dong and SRIE and the method provided by the invention.

Average value of indexes of 120 images in table

Tab.1 Mean value of indicators for 20images

It can be seen from the table that the invention achieves better effect on experimental pictures and has certain practical value, the index of the color entropy is lower than that of the method A, and the indexes of LOE, NIQE and BRISQE exceed that of the comparison method.

Claims

1. A low-illumination image enhancement method based on illumination component optimization is characterized by comprising the following steps:

(2) processing the initial illumination map obtained in the step (1) by using singular value decomposition, and normalizing;

2. The method according to claim 1, wherein the initial illumination map is obtained in step (1) using a max-RGB model, i.e. using the maximum of RGB three channels at point (x, y) of the original image as the illumination of the point.

3. The method of claim 1, wherein the initial illumination map is processed in step (2) using singular value decomposition by the following method:

singular value decomposition is performed for a 3 × 3 region of the image centered at point (x, y), i.e.:

A＝UΣV^T (1)

wherein, a represents a 3 × 3 area in the initial illumination map, U and V are both orthogonal matrices, Σ is a non-negative diagonal matrix, and consists of singular values of a, and there are:

Σ＝diag＝(s₁,s₂,…,s_N)(s₁＞s₂＞…＞s_N) (2)

at the maximum singular value S₁And (4) replacing the point illumination, and performing normalization after the whole image is processed.

4. The method according to claim 1, wherein the illumination is optimized by three times of guided filtering using windows with different sizes in step (3), and the optimization method is as follows:

let L be the initial illumination map obtained in step 1_ini(x, y), the irradiance profile obtained in step 2 is L_s(x,y)，GF_R(A, B) denotes guided filtering with a window radius R, there is the following equation:

in the formula, L₃(x, y) is the optimized final used illumination pattern.

5. The method according to claim 1, wherein the enhanced image is obtained in step (4) by using a simplified Retinex model, which has the following formula:

in the formula, τ is 0.01 to prevent the denominator from being zero.

6. The method according to claim 1, wherein in step (5), the enhanced image is subjected to guided filtering and denoising using the green channel component with lower noise level in the original image, so as to improve the visual effect and various indexes of the image.