Low-light image enhancement method and system based on guided filter camera characteristic

Abstract

The application provides a low-illumination image enhancement method and system based on guided filtering and camera characteristics, and the method comprises the following steps: acquiring an original low-illumination image by using an image sensor, and processing the original low-illumination image into RGB space data; converting the RGB space data into HSV space data, and obtaining a color component H, a brightness component V and a saturation component S; processing the brightness component V to obtain illumination image data, calculating the illumination image data by using guided filtering, and obtaining illumination image calculation data; obtaining a camera response model, and enhancing the brightness component V according to the illumination image calculation data by using the camera response model to obtain an image enhancement component; and obtaining brightness enhancement HSV space data according to the color component H, the image enhancement component and the saturation component S, and processing the brightness enhancement HSV space data to obtain an enhanced result image. The application solves the technical problems of large amount of calculation, difficulty in real-time implementation, appearance of light halo artifacts and color distortion in the prior art.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of image optimization processing, and relates to a low-illumination image enhancement method and system based on guided filtering of camera characteristics. BACKGROUND

[0002] With the rapid development of digital image technology, various imaging optical systems are applied more and more widely, such as the world's largest video monitoring network "China Sky Eye" has been built in China. However, in the evening, night or indoor poor lighting and the like, the images obtained by the imaging device usually have the prominent problems of too low brightness and contrast and serious loss of image details, which seriously affects the efficiency of target detection, identification and the like. How to efficiently and quickly effectively enhance the low-illumination image is an important problem to be solved in the field of image processing. The low-illumination image enhancement methods mainly include three types: histogram-based method, Retinex model-based method and deep learning-based method. The histogram-based enhancement method includes histogram equalization, Gamma correction and the like. This kind of method aims to pull up the dynamic range of the image under the assumption of certain histogram distribution, so as to improve the brightness and contrast. However, since this method does not fully consider the physical process of image formation and the non-uniform distribution of light in complex environments, it is easy to cause over-enhancement or under-enhancement, and the visual effect of the image is not natural enough. The Retinex model-based method includes single-scale Retinex (SSR), multi-scale Retinex (MSR) and MSRCR with color restoration and the like. The enhancement results of this kind of method are easy to appear halos and color distortion. In recent years, the low-illumination image enhancement method based on deep learning has been valued by researchers. However, this kind of method often needs to use a very deep deep learning convolution network, and has the prominent problems of large amount of calculation and difficulty in real-time implementation.

[0003] The patent CN 111861899A discloses a low-illumination image enhancement method based on path optimization. The method first converts the original color image from the RGB space to the HSV space, obtains the luminance component V of the image, then processes the luminance component V using the path-optimized MR algorithm, and finally converts the processed HSV space to the RGB space to synthesize a new image. The method needs to continuously iterate multiple paths during the processing of the luminance component, which has poor real-time performance. Moreover, the processing of the luminance component does not consider the response characteristics of the camera under different lighting conditions, and the enhancement effect may drown some details of the image. A new low-light image enhancement algorithm using camera response model is proposed, which first uses an optimization method to estimate the illumination based on global image information, and then enhances the R, G and B channels of the original low-illumination image in the RGB space based on the camera response model. The method utilizes the nonlinear response characteristics of the camera and can improve the brightness distribution of the image, but has the problems of large calculation amount and poor real-time performance in the estimation process of the illumination image, and the enhancement of the R, G and B channels may cause color distortion. SUMMARY

[0004] The technical problem to be solved by the present application is how to solve the technical problems of poor real-time performance, low brightness and contrast, and serious loss of image details in the conventional technology.

[0005] The present application solves the above technical problems by adopting the following technical scheme: a guided filtering camera characteristic low-illumination image enhancement method comprises:

[0006] An image sensor is used to acquire an original low-illumination image, which is processed into RGB space data;

[0007] The RGB space data is converted into HSV space data, from which a color component H, a luminance component V and a saturation component S are obtained;

[0008] The luminance component V is processed to obtain illumination image data, and guided filtering is used to calculate the illumination image data to obtain illumination image calculation data;

[0009] A camera response model is obtained, and the luminance component V is enhanced and processed according to the illumination image calculation data using the camera response model to obtain an image enhancement component;

[0010] Luminance enhancement HSV space data is obtained according to the color component H, the image enhancement component and the saturation component S, and an enhanced result image is obtained by processing.

[0011] In the HSV color space, the luminance component V is enhanced according to the estimated illumination image by using a camera response model, while the hue component H and the saturation component S are kept unchanged, so that good luminance enhancement effect can be obtained and color distortion of the enhanced image can be effectively reduced.

[0012] In a more specific technical solution, the step of converting the RGB space data into HSV space data to obtain the hue component H, the luminance component V and the saturation component S comprises:

[0013] processing the RGB space data to obtain a low-illumination conversion parameter;

[0014] The original low-illumination image I is converted from the RGB space to the HSV space according to the following logic:

[0015]

[0016]

[0017]

[0018] wherein B is the blue channel value of the image I, G is the green channel value of the image I, R is the red channel value of the image I, max(R, G, B) is the maximum value among R, G and B, and min(R, G, B) is the minimum value among R, G and B.

[0019] The image is first converted from the color space RGB to the color space HSV, and the luminance component V is subjected to fast illumination image estimation by using a guided filter with high parallelism, so that the real-time performance of the algorithm is effectively improved.

[0020] In a more specific technical solution, the step of processing the RGB space data to obtain a low-illumination conversion parameter comprises:

[0021] The RGB space data is processed to obtain the low-illumination conversion parameter according to the following logic:

[0022]

[0023] wherein arccos is an inverse cosine function, B is the blue channel value of the image I, G is the green channel value of the image I, and R is the red channel value of the image I.

[0024] In a more specific technical solution, the step of processing the luminance component V to obtain illumination image data, and calculating the illumination image data by using a guided filter to obtain illumination image calculation data comprises:

[0025] The acquired luminance component V is subjected to fast illumination image estimation according to the following logic by using guided filtering, to obtain the illumination image calculation data T:

[0026]

[0027] Wherein, i is the coordinate position of the pixel point; μ k and Respectively represent the mean and variance of the image V in the local window w; |w| is the number of pixels in the window; ε is a constant.

[0028] The present application can pay attention to important details of the image by using guided filtering and camera response model, and obtain good luminance enhancement effect, and effectively reduce color distortion of the enhanced image.

[0029] In a more specific technical solution, the step of acquiring the camera response model and enhancing the luminance component V according to the illumination image calculation data by using the camera response model to obtain the image enhancement component, comprises:

[0030] Processing the model parameter K of the acquired camera response model;

[0031] Enhancing the luminance component V by using the camera response model and the illumination image calculation data T to obtain the image enhancement component V':

[0032]

[0033] Wherein, a is a constant, generally taken as-0.3293; b is a constant, generally taken as 1.1258.

[0034] In a more specific technical solution, the step of processing the model parameter K of the acquired camera response model, comprises:

[0035] Processing the model parameter of the camera response model according to the following logic:

[0036] K = min (1 / T, K0)

[0037] Wherein, K0 is a constant, generally taken as 6-8.

[0038] In a more specific technical solution, the step of acquiring the luminance enhancement HSV space data from the toning component H, the image enhancement component and the saturation component S, and processing the enhanced result image according to the step, comprises:

[0039] Processing the toning component H, the image enhancement component and the saturation component S to obtain conversion parameters;

[0040] According to the conversion parameter, the hue component H, the image enhancement component and the saturation component S are converted to RGB color space by using preset logic, so as to obtain the enhanced result image.

[0041] The application converts the enhanced V component, H component and S component to RGB color space, and uses a camera response model to enhance the brightness component V according to an estimated illumination image, while keeping the hue component H and the saturation component S unchanged, so as to effectively reduce the color distortion of the enhanced image.

[0042] In a more specific technical solution, the step of processing the hue component H, the image enhancement component and the saturation component S to obtain the conversion parameter comprises:

[0043] The hue component H, the image enhancement component and the saturation component S are processed by using the following logic to obtain the conversion parameter:

[0044]

[0045]

[0046] p = V'(1-S)

[0047] t = V'[1-(1-f)S]

[0048] Wherein mod is the remainder.

[0049] In a more specific technical solution, the step of processing the hue component H, the image enhancement component and the saturation component S to obtain the conversion parameter comprises:

[0050] The hue component H, the image enhancement component and the saturation component S are converted to RGB color space according to the following logic to obtain the enhanced result image I':

[0051]

[0052] In a more specific technical solution, a low-illumination image enhancement system based on guided filtering and camera characteristics, the system comprises:

[0053] An RGB processing module is configured to acquire an original low-illumination image by using an image sensor, and process the original low-illumination image into RGB space data.

[0054] A space conversion module is configured to convert the RGB space data into HSV space data, so as to obtain a hue component H, a brightness component V and a saturation component S, and the space conversion module is connected with the RGB processing module.

[0055] A light image guided filter processing module is configured to process light image data of the luminance component V, and calculate the light image data by guided filter to obtain light image calculation data, and the light image guided filter processing module is connected to the spatial conversion module;

[0056] An image enhancement module is configured to obtain a camera response model, and enhance the luminance component V according to the light image calculation data by using the camera response model to obtain an image enhancement component, and the image enhancement module is connected to the light image guided filter processing module;

[0057] An enhanced image module is configured to obtain luminance enhancement HSV space data according to the hue component H, the image enhancement component and the saturation component S, and obtain an enhanced result image according to the luminance enhancement HSV space data, and the enhanced image module is connected to the image enhancement module.

[0058] Compared with the prior art, the present application has the following advantages: the present application firstly converts an image from a color space RGB to a color space HSV, and then performs fast light image estimation on a luminance component V in the color space HSV by using guided filter with high parallelism; then, the luminance component V is enhanced according to the estimated light image by using a camera response model, while the hue component H and the saturation component S remain unchanged; finally, the enhanced V component and the H component and the S component are converted to the color space RGB. In the above process, on the one hand, fast calculation can be performed to effectively improve the real-time performance of the algorithm; on the other hand, the guided filter and the camera response model can pay attention to important details of the image, obtain good luminance enhancement effect, and effectively reduce color distortion of the enhanced image. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 is a low-illumination image enhancement flow method flow diagram based on guided filter and camera characteristics;

[0060] Figure 2 is a first outdoor scene low-light image diagram in the embodiment;

[0061] Figure 3 is a second outdoor scene low-light image diagram in the embodiment;

[0062] Figure 4 is a first outdoor scene low-illumination image enhancement effect diagram in the embodiment;

[0063] Figure 5 is a second outdoor scene low-illumination image enhancement effect diagram in the embodiment. DETAILED DESCRIPTION

[0064] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0065] Embodiments

[0066] As shown in the following Figure 1 , the specific steps of the low-illumination image enhancement process method based on guided filtering and camera characteristics provided by the present application are as follows:

[0067] Step one: as shown in the following Figure 2 and Figure 3 , input any low-illumination image: I;

[0068] Step two: convert the original low-illumination color image I from the RGB space to the HSV space:

[0069]

[0070]

[0071]

[0072]

[0073] wherein arccos is the inverse cosine function, B is the blue channel value of the image I, G is the green channel value of the image I, R is the red channel value of the image I, max(R, G, B) is the maximum value among the three values of R, G and B, and min(R, G, B) is the minimum value among the three values of R, G and B.

[0074] Step three: use guided filtering to perform fast illumination image estimation on the obtained V component, and obtain the estimated illumination image T:

[0075]

[0076] wherein i is the coordinate position of the pixel point; μ k and respectively represent the mean value and the variance of the image V in the local window w; |w| is the number of pixels in the window; and ε is a constant.

[0077] Step four: use the camera response model and the estimated illumination image T to enhance the V component, and obtain the enhanced V':

[0078]

[0079] wherein a is a constant, generally taken as -0.3293; b is a constant, generally taken as 1.1258; K is:

[0080] K = min(1 / T, K0)

[0081] wherein K0 is a constant, generally taken as 6-8.

[0082] Step five: as shown in Figure 4 and Figure 5 , H, S, V' are converted to RGB color space to obtain the enhanced image I':

[0083]

[0084] wherein:

[0085]

[0086]

[0087] p = V'(1-S)

[0088] t = V'[1-(1-f)S]

[0089] mod is the remainder.

[0090] In summary, the present application first converts the image from color space RGB to color space HSV, and uses the highly parallel guided filtering to estimate the illumination image of the luminance component V; then uses the camera response model to enhance the luminance component V according to the estimated illumination image, while keeping the hue component H and the saturation component S unchanged; finally, converts the enhanced V component and H component, S component to color space RGB. In the above process, on the one hand, fast calculation can be performed to effectively improve the real-time performance of the algorithm; on the other hand, the guided filtering and the camera response model can pay attention to important details of the image, obtain good brightness enhancement effect, and effectively reduce the color distortion of the enhanced image, solving the technical problems of large amount of calculation, difficulty in real-time implementation, appearance of light halo artifacts and color distortion in the prior art.

[0091] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A low-light image enhancement method based on guided filtering and camera characteristics, characterized in that, The method is applied to low-illumination image processing and comprises the following steps: An original low-illumination image is acquired by using an image sensor, and RGB space data is processed based on the original low-illumination image; The RGB space data is converted into HSV space data, and a color component H, a brightness component V and a saturation component S are obtained based on the RGB space data; wherein the RGB space data is processed to obtain a low-illumination conversion parameter; The original low-illumination image I is converted from the RGB space to the HSV space according to the following logic: wherein B is a blue channel value of the original low-illumination image I, G is a green channel value of the original low-illumination image I, R is a red channel value of the original low-illumination image I, max(R, G, B) is the maximum value among the three values R, G and B, and min(R, G, B) is the minimum value among the three values R, G and B; The brightness component V is processed to obtain illumination image data, and the illumination image data is calculated by using guided filtering to obtain illumination image calculation data; A camera response model is obtained, and the brightness component V is enhanced based on the illumination image calculation data by using the camera response model to obtain an image enhancement component; wherein the brightness component V obtained is quickly estimated to obtain the illumination image calculation data T by using guided filtering according to the following logic: where i is the coordinate position of the pixel; μ k and respectively represent the mean and variance of the image V in the local window w; |w| is the number of pixels within the window; ε is a constant; The brightness enhancement HSV space data is obtained based on the color component H, the image enhancement component and the saturation component S, and an enhanced result image is processed based on the brightness enhancement HSV space data.

2. The low-light image enhancement method based on guided filtering and camera characteristics according to claim 1, characterized in that, The step of processing the RGB space data to obtain a low-illumination conversion parameter comprises the following logic: The RGB space data is processed to obtain the low-illumination conversion parameter according to the following logic: wherein arccos is an inverse cosine function, B is a blue channel value of an image I, G is a green channel value of the image I, and R is a red channel value of the image I. 3.The low-light image enhancement method based on guided filtering and camera characteristics of claim 1, wherein, The step of obtaining a camera response model and enhancing the brightness component V based on the illumination image calculation data by using the camera response model to obtain an image enhancement component comprises the following steps: A model parameter K of the camera response model is processed; enhancing the luminance component V using the camera response model and the illumination image data T to obtain the image enhancement component V ′ : wherein a is a constant value, and is generally taken as -0.3293; and b is a constant value, and is generally taken as 1.1258.

4. The low-light image enhancement method based on guided filtering and camera characteristics according to claim 3, characterized in that, The step of processing the model parameter K of the camera response model comprises the following logic: The model parameter K of the camera response model is processed according to the following logic: K = min(1 / T, K0) wherein K0 is a constant value, and is generally taken as 6-8.

5. The low-light image enhancement method based on guided filtering and camera characteristics according to claim 1, characterized in that, The step of obtaining brightness enhancement HSV space data based on the color component H, the image enhancement component and the saturation component S, and processing an enhanced result image based on the brightness enhancement HSV space data comprises the following steps: The color component H, the image enhancement component and the saturation component S are processed to obtain a conversion parameter; The color component H, the image enhancement component and the saturation component S are converted into an RGB color space by using a preset logic based on the conversion parameter, and the enhanced result image is obtained based on the color component H, the image enhancement component and the saturation component S converted into the RGB color space.

6. The low-light image enhancement method based on guided filtering and camera characteristics according to claim 5, characterized in that, The step of processing the color component H, the image enhancement component and the saturation component S to obtain a conversion parameter comprises the following logic: The hue component H, the image enhancement component and the saturation component S are processed according to the following logic, so as to obtain the conversion parameter: p = V'(1-S) t = V'[1-(1-f)S] Wherein mod is the remainder.

7. The low-light image enhancement method based on guided filtering and camera characteristics according to claim 5, characterized in that, The step of converting the hue component H, the image enhancement component and the saturation component S to the RGB color space according to the conversion parameter by using a preset logic, so as to obtain the enhanced result image, comprises: The shading component H, the image enhancement component and the saturation component S are converted to the RGB color space according to the following logic, from which the enhanced result image I is obtained ′ :

8. A low-light image enhancement system based on guided filtering and camera characteristics, configured to perform a low-light image enhancement method based on guided filtering and camera characteristics according to any one of the preceding claims 1 to 7, characterized in that, The system is applied to low-illumination image processing, and comprises: An RGB processing module is configured to acquire an original low-illumination image by using an image sensor, and process the original low-illumination image into RGB space data; A space conversion module is configured to convert the RGB space data into HSV space data, so as to obtain a hue component H, a brightness component V and a saturation component S, and the space conversion module is connected with the RGB processing module; An illumination image guided filter processing module is configured to process the brightness component V to obtain illumination image data, and calculate the illumination image data by using guided filter, so as to obtain illumination image calculation data, and the illumination image guided filter processing module is connected with the space conversion module; An image enhancement module is configured to obtain a camera response model, and enhance the brightness component V according to the camera response model and the illumination image calculation data, so as to obtain an image enhancement component, and the image enhancement module is connected with the illumination image guided filter processing module; An enhanced image module is configured to obtain brightness enhanced HSV space data according to the hue component H, the image enhancement component and the saturation component S, and process the brightness enhanced HSV space data to obtain an enhanced result image, and the enhanced image module is connected with the image enhancement module.

Images