Image enhancement method and device based on vehicle-mounted surround view system

Abstract

The application discloses an image enhancement method and device based on a vehicle-mounted surround view system, converts a low-illumination grayscale image from an RGB color space to an HSV color space to obtain a V-channel luminance value; performs double gamma transformation on the V-channel luminance value to generate a first image luminance value and a second image luminance value; performs Laplace transformation on the V-channel luminance value to generate a third image luminance value; performs weighted fusion on the first image luminance value, the second image luminance value and the third image luminance value to generate a fourth image luminance value; performs multi-scale Retinex transformation on the fourth image luminance value to generate a fifth image luminance value; and resets the grayscale values of the RGB three channels of the low-illumination grayscale image according to the fourth image luminance value and the fifth image luminance value to generate an enhanced image; the beneficial effects are that the low-illumination grayscale image can be enhanced in multiple ways, and the generated multiple luminance values are weighted and fused, so that the luminance enhancement effect can be improved, and color imbalance can be prevented.

Description

Technical Field

[0001] This invention relates to the field of image processing, specifically to an image enhancement method and apparatus based on a vehicle surround view system. Background Technology

[0002] The vehicle surround view system consists of four fisheye cameras. After distortion correction and perspective transformation, the images from each camera are stitched together to obtain a 360-degree bird's-eye view, which can provide complete information about the vehicle's surroundings. However, in low light conditions, the image brightness is low and details are not obvious, which can lead to unreliable automatic parking functions.

[0003] Hohai University disclosed a method, apparatus, and storage medium for enhancing low-light grayscale images based on gamma transform in its patent application, "Method, Apparatus, and Storage Medium for Enhancing Low-Light Grayscale Images Based on Gamma Transform" (Patent Application No. CN202110670167.9, Publication No. CN113393394A). This invention uses gamma transform to enhance images. However, the gamma transform used in this invention has certain limitations, and the image enhancement effect is not significant in nighttime scenes. Summary of the Invention

[0004] To overcome the shortcomings and deficiencies of existing technologies, this invention discloses an image enhancement method based on a vehicle surround view system. This method can enhance low-light grayscale images through multiple techniques and performs weighted fusion of multiple generated brightness values, thereby not only improving brightness enhancement but also preventing color imbalance. The method includes the following steps:

[0005] Obtain a low-light grayscale image;

[0006] The low-light grayscale image is converted from the RGB color space to the HSV color space to obtain the V channel brightness value;

[0007] Perform a dual gamma transform on the V channel brightness value to generate a first image brightness value and a second image brightness value;

[0008] The V channel brightness value is subjected to Laplacian transform to generate the third image brightness value;

[0009] The first image brightness value, the second image brightness value, and the third image brightness value are weighted and fused to generate a fourth image brightness value;

[0010] Perform a multi-scale Retinex transform on the brightness value of the fourth image to generate the brightness value of the fifth image;

[0011] Based on the brightness values ​​of the fourth and fifth images, the grayscale values ​​of the RGB three channels of the low-light grayscale image are reset to generate an enhanced image.

[0012] Furthermore, the step of acquiring the low-light grayscale image includes the following steps:

[0013] A first low-light grayscale image is acquired, which is captured by a fisheye camera;

[0014] Perform distortion correction on the first low-light grayscale image to generate a second low-light grayscale image;

[0015] Multiple second low-light grayscale images are stitched together to generate the low-light grayscale image.

[0016] Furthermore, the step of performing a dual gamma transform on the V channel brightness value to generate a first image brightness value and a second image brightness value includes transforming the V channel brightness value using the following formula:

[0017] I v_Gamma1 (x,y)=[I v (x,y)] γ

[0018] I v_Gamma2 (x,y)=1-[1-I v (x,y)] γ

[0019] Among them, I v_Gamma1 (x,y) represents the brightness value of the first image, i.e., the brightness value of the dual-gamma compressed image; I v_Gamma2 (x,y) represents the brightness value of the second image, i.e., the brightness value of the dual-gamma extended image; I v (x,y) represents the brightness value of the V channel, (x,y) represents the pixel coordinates in the image; γ is a variable that controls the degree of image enhancement.

[0020] Furthermore, the step of performing a Laplacian transform on the V channel brightness value to generate a third image brightness value includes transforming the V channel brightness value using the following formula:

[0021]

[0022] Among them I v_laplace (x,y) represents the brightness value of the third image; I v (x,y) represents the brightness value of the V channel.

[0023] Furthermore, the weighted fusion of the first image brightness value, the second image brightness value, and the third image brightness value generates a fourth image brightness value, which includes the third image brightness value obtained through the following formula:

[0024]

[0025] Among them Iv_en (x,y) represents the brightness value of the fourth image; I v_Gamma1 (x,y) represents the brightness value of the first image; I v_Gamma2 (x,y) represents the brightness values ​​of the second image; I v_laplace (x,y) represents the brightness value of the third image.

[0026] Furthermore, the step of performing a multi-scale Retinex transform on the fourth image brightness value to generate the fifth image brightness value includes the following steps:

[0027] Construct a Gaussian wrap function;

[0028] The RGB channels of the image corresponding to the fourth image brightness value are filtered at multiple scales according to the Gaussian wrapping function and then weighted to generate a preset brightness value.

[0029] The fifth image brightness value is generated by subtracting the preset brightness value from the fourth image brightness value.

[0030] Furthermore, the step of performing a multi-scale Retinex transform on the fourth image brightness value to generate a fifth image brightness value includes generating the fifth image brightness value using the following formula:

[0031]

[0032] Among them, R MSR (x,y) represents the brightness value of the fifth image; N is the number of scales; w i Weights corresponding to multiple scales; I v_en (x,y) represents the brightness value of the fourth image; G i (x,y) is a Gaussian wrapping function with N scales.

[0033] Furthermore, the step of resetting the grayscale values ​​of the three RGB channels of the low-light grayscale image based on the brightness values ​​of the fourth and fifth images to generate the enhanced image includes the following steps:

[0034] The enhancement factor of the RGB three channels of the low-light grayscale image is obtained, and the enhancement factor is the ratio of the brightness value of the fifth image to the brightness value of the fourth image;

[0035] The grayscale values ​​of the three RGB channels of the low-light grayscale image are reset according to the enhancement factor to generate the enhanced image.

[0036] Furthermore, the step of resetting the grayscale values ​​of the three RGB channels of the low-light grayscale image according to the enhancement factor includes resetting the grayscale values ​​of the three RGB channels of the low-light grayscale image using the following formula:

[0037]

[0038] Where, r r (x,y) represents the grayscale value of the R channel in the enhanced image; R MSR (x,y) / / I v_en (x,y) is the enhancement factor, r g (x,y) represents the grayscale value of the G channel in the enhanced image; r b (x,y) represents the gray value of the B channel of the enhanced image; R(x,y) represents the gray value of the R channel of the low-light grayscale image; G(x,y) represents the gray value of the G channel of the low-light grayscale image; B(x,y) represents the gray value of the B channel of the low-light grayscale image.

[0039] On the other hand, this application also provides an image enhancement device based on a vehicle surround view system, comprising:

[0040] Low-light grayscale image acquisition module: used to acquire low-light grayscale images;

[0041] V-channel brightness value acquisition module: used to acquire low-light grayscale images;

[0042] Dual-gamma transform module: used to perform dual-gamma transform on the brightness value of the V channel to generate a first image brightness value and a second image brightness value;

[0043] Laplacian transform module: Performs Laplacian transform on the brightness value of the V channel to generate the brightness value of the third image;

[0044] The fourth image brightness value generation module is used to weightedly fuse the first image brightness value, the second image brightness value, and the third image brightness value to generate a fourth image brightness value.

[0045] Fifth image brightness value generation module: Performs multi-scale Retinex transform on the fourth image brightness value to generate the fifth image brightness value;

[0046] Enhanced image generation module: used to reset the grayscale values ​​of the RGB three channels of the low-light grayscale image based on the brightness values ​​of the fourth image and the fifth image, and generate the enhanced image.

[0047] Implementing this invention has the following beneficial effects:

[0048] A combination of dual gamma transform and Laplacian transform is used to enhance the V channel of low-light images. The enhanced V channel brightness value is then subjected to multi-scale Retinex transform. Based on the gray values ​​of the RGB channels of the low-light grayscale image and the brightness ratio before and after the Retinex transform, the gray values ​​of the RGB channels are reset to improve the image enhancement effect and prevent color imbalance. Attached Figure Description

[0049] To more clearly illustrate the technical solution of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of an image enhancement method based on a vehicle surround view system provided in an embodiment of the present invention;

[0051] Figure 2 This is a flowchart of a method for generating a fifth image brightness value provided in an embodiment of the present invention;

[0052] Figure 3 (a) is a low-light grayscale image;

[0053] Figure 3 (b) is the image after brightness enhancement of the low-light grayscale image;

[0054] Figure 4 This is a structural block diagram of an image enhancement device based on a vehicle surround view system provided in an embodiment of the present invention. Detailed Implementation

[0055] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0056] Example

[0057] In this embodiment, the technical problem to be solved by the present invention is to enhance low-light grayscale images through multiple methods and to weightedly fuse the generated multiple brightness values, which not only improves the brightness enhancement effect but also prevents color imbalance. For example... Figure 1 The method shown includes the following steps:

[0058] S1: Obtain a low-light grayscale image;

[0059] In low-light conditions such as indoors and at night, computer vision monitoring equipment suffers from insufficient illumination from non-natural light sources. This results in weak reflected light from the target surface, leading to insufficient light entering the imaging sensor. Consequently, images acquired at night exhibit severe quality degradation, low recognizability, and significant noise, making it difficult to discern details and greatly reducing their application value. These images are collectively referred to as low-light images. Low-light images captured at night are prone to blurred details. Using low-light images in vehicle surround-view systems can render automatic parking functions unreliable. Therefore, it is necessary to enhance the illumination of low-light grayscale images to acquire them under insufficient ambient light conditions. Specifically, acquiring low-light grayscale images includes the following steps:

[0060] S11: Acquire the first low-light grayscale image, which is captured by a fisheye camera;

[0061] A fisheye camera is an ultra-wide-angle camera with a field of view close to or even exceeding 180°. The range it can observe is much larger than the human eye's field of view. Therefore, using a fisheye camera can acquire a wider range of low-light grayscale images.

[0062] S12: Perform distortion correction on the first low-light grayscale image to generate a second low-light grayscale image;

[0063] Since the large field of view of a fisheye lens comes at the cost of sacrificing the intuitiveness of the image, the images taken with a fisheye lens have a large degree of distortion. Therefore, it is necessary to distort the first low-light grayscale image and perform perspective transformation. This process is existing technology and will not be described in detail here.

[0064] S13: Stitch together multiple second low-light grayscale images to generate a low-light grayscale image.

[0065] A typical vehicle surround view system consists of four fisheye cameras. Each camera corrects the distortion and performs perspective transformation on a first low-light grayscale image. Then, multiple images are stitched together to obtain a 360-degree bird's-eye view, i.e., a low-light grayscale image. By capturing images in real time, a monitoring video centered on the vehicle can be obtained. Therefore, the low-light grayscale image in this application can be a 360-degree bird's-eye view or a video that can capture a certain distance around the vehicle.

[0066] S2: Convert the low-light grayscale image from the RGB color space to the HSV color space to obtain the V channel brightness value;

[0067] The low-light grayscale image has three RGB channels, namely the red component (R), green component (G), and blue component (B); and three HSV channels, namely the chroma component (H), saturation component (S), and luminance component (V). This embodiment mainly focuses on enhancing the luminance component, which can directly convert the color space to obtain the luminance value of the low-light grayscale image.

[0068] S3: Perform a dual gamma transform on the V channel brightness value to generate the first image brightness value and the second image brightness value;

[0069] Traditional algorithms for enhancing image brightness and contrast include linear stretching, logarithmic transformation, and gamma transformation. Linear stretching can linearly stretch the grayscale set of an image to 0-255. While enhancing image contrast, linear stretching also stretches the bright areas of the image, leading to partial overexposure. Traditional logarithmic and gamma transform algorithms can expand pixel values ​​in dark areas but also compress pixel values ​​in bright areas, resulting in loss of detail in bright areas. In traditional gamma transform, a gamma coefficient greater than 0 and less than 1 can enhance the contrast of underexposed images; a gamma coefficient greater than 1 can enhance the contrast of overexposed images; and a gamma coefficient equal to 1 does not process the image. A bidirectional gamma transform method for low-light image contrast enhancement adapts the gamma parameters. Through bidirectional gamma correction, it improves brightness and contrast while avoiding overflow of pixel values ​​in bright areas, overcoming the limitation of single-gamma correction algorithms that cannot simultaneously address both bright and dark areas. A bidirectional gamma transform is performed on the V channel brightness value to generate a first and a second image brightness value, including a transformation of the V channel brightness value using the following formula:

[0070] I v_Gamma1 (x,y)=[I v (x,y)] γ

[0071] I v_Gamma2 (x,y)=1-[1-I v (x,y)] γ

[0072] Among them, I v_Gamma1 (x,y) represents the brightness value of the first image, i.e., the brightness value of the dual-gamma compressed image; I v_Gamma2 (x,y) represents the brightness value of the second image, i.e., the brightness value of the dual-gamma extended image; I v (x,y) represents the brightness value of the V channel, (x,y) represents the pixel coordinates in the image; γ is a variable that controls the degree of image enhancement.

[0073] S4: Perform a Laplacian transform on the brightness value of the V channel to generate the brightness value of the third image;

[0074] The V channel luminance value is transformed using a Laplacian transform to generate the third image luminance value. This transformation involves applying the following formula to the V channel luminance value:

[0075]

[0076]

[0077]

[0078]

[0079] Among them I v_laplace (x,y) represents the brightness value of the third image; I v (x,y) represents the brightness value of the V channel.

[0080] The V channel luminance value of the low-light grayscale image is transformed using Laplacian transform to generate a third image luminance value. This third image luminance value is obtained through different methods than the first and second image luminance values, utilizing multiple acquisition channels. The luminance values ​​are then weighted, resulting in a more convincing luminance value. The weighting process is as follows:

[0081] S5: Weighted fusion of the first image brightness value, the second image brightness value and the third image brightness value to generate the fourth image brightness value;

[0082] The brightness values ​​of the first image, the second image, and the third image are weighted and fused to generate the brightness value of the fourth image, which is obtained from the third image using the following formula:

[0083]

[0084] Among them I v_en (x,y) represents the brightness value of the fourth image; I v_Gamma1 (x,y) represents the brightness value of the first image; I v_Gamma2 (x,y) represents the brightness values ​​of the second image; I v_laplace (x,y) represents the brightness value of the third image.

[0085] S6: Perform multi-scale Retinex transformation on the brightness value of the fourth image to generate the brightness value of the fifth image;

[0086] The image captured by the camera is obtained by the reflection of incident light on the surface of the object. The reflectivity is determined by the object itself and is not affected by the change of incident light. The Retinex theoretical enhancement algorithm obtains the reflection component by removing the luminance component, thereby achieving the image enhancement effect. Therefore, it is only necessary to estimate the luminance component to obtain the reflection component. The estimation of the luminance component directly determines the image dehazing and restoration effect. The Gaussian wrap function (Gaussian convolution function) can better estimate the luminance component from the known image.

[0087] Perform a multi-scale Retinex transform on the brightness values ​​of the fourth image to generate the brightness values ​​of the fifth image, as shown below. Figure 2 As shown, the steps include:

[0088] S61: Construct the Gaussian wrap function;

[0089] S62: Based on the Gaussian wrap function, the RGB three channels of the image corresponding to the brightness value of the fourth image are filtered at multiple scales and then averaged to generate an estimated brightness value.

[0090] S63: Subtract the estimated brightness value from the fourth image brightness value to generate the fifth image brightness value.

[0091] As mentioned earlier, the fourth image brightness value is obtained by subtracting the estimated brightness value from the original image brightness value, generating the fifth image brightness value as the output image. This algorithm can compress the dynamic range of the image while preserving some of the image's color and detail enhancement. Its specific expression is as follows:

[0092]

[0093] Among them, R MSR (x,y) represents the brightness value of the fifth image; N is the number of scales; w i Weights corresponding to multiple scales; I v_en (x,y) represents the brightness value of the fourth image; G i (x,y) is a Gaussian wrapping function with N scales; For convolution operations, specifically, the best results are achieved when N=3, meaning three Gaussian filters of different scales are used to filter the image. These three scales correspond to large, medium, and small scales, respectively. i The weights corresponding to different scales are w1 = w2 = w3 = 1 / 3 respectively.

[0094] S7: Based on the brightness values ​​of the fourth and fifth images, reset the grayscale values ​​of the RGB three channels of the low-light grayscale image to generate the enhanced image.

[0095] The multi-scale Retinex algorithm performs multiple filtering operations at different scales on each channel and then performs weighted summation, which increases the processing time. Furthermore, the use of different scales results in the restored RGB ratios being different from the original image ratios, leading to color distortion. Therefore, an enhancement factor needs to be introduced to reduce the impact of color distortion, i.e., the enhancement ratio. Based on this specific ratio, the image illumination can be enhanced while preventing color imbalance.

[0096] S71: Based on the brightness values ​​of the fourth and fifth images, reset the grayscale values ​​of the three RGB channels of the low-light grayscale image to generate the enhanced image, including the following steps:

[0097] S72: Obtain the enhancement factor of the three RGB channels of the low-light grayscale image. The enhancement factor is the ratio of the brightness value of the fifth image to the brightness value of the fourth image.

[0098] S73: Based on the enhancement factor, reset the grayscale values ​​of the three RGB channels of the low-light grayscale image to generate the enhanced image.

[0099] Images taken at night, such as Figure 3 As shown in (a), insufficient image illumination results in blurred details. To address this, the low-light grayscale image is converted from the RGB color space to the HSV color space. Then, the V channel brightness values ​​undergo dual gamma transformation and Laplacian transformation. The results are weighted and then subjected to multi-scale Retinex transformation. Finally, the grayscale values ​​of the RGB three channels of the low-light grayscale image are reset to generate the enhanced image. Figure 3 (b) As can be seen intuitively from the figure, Figure 3 (b) has high image brightness and clear details, which can better provide technical support for vehicle surround view systems.

[0100] The grayscale values ​​of the three RGB channels of the low-light grayscale image are reset according to the enhancement factor. The enhancement factor is used to reset the grayscale values ​​of the three RGB channels of the low-light grayscale image using the following formula:

[0101]

[0102] Where, r r (x,y) represents the grayscale value of the R channel in the enhanced image; R MSR (x,y) / I v_en (x,y) is the enhancement factor, r g (x,y) represents the grayscale value of the G channel in the enhanced image; r b(x,y) represents the gray value of the B channel of the enhanced image; R(x,y) represents the gray value of the R channel of the low-light grayscale image; G(x,y) represents the gray value of the G channel of the low-light grayscale image; B(x,y) represents the gray value of the B channel of the low-light grayscale image.

[0103] This embodiment also provides an image enhancement device based on a vehicle surround view system, which can implement all the steps of the above method, such as... Figure 4 The device shown includes:

[0104] Low-light grayscale image acquisition module: used to acquire low-light grayscale images;

[0105] V-channel brightness value acquisition module: used to acquire low-light grayscale images;

[0106] Dual Gamma Transform Module: Used to perform dual gamma transformation on the V channel brightness value to generate a first image brightness value and a second image brightness value;

[0107] Laplacian Transform Module: Performs Laplacian transform on the V channel brightness value to generate the third image brightness value;

[0108] The fourth image brightness value generation module is used to weightedly fuse the first image brightness value, the second image brightness value, and the third image brightness value to generate the fourth image brightness value.

[0109] The fifth image brightness value generation module performs a multi-scale Retinex transform on the fourth image brightness value to generate the fifth image brightness value.

[0110] Enhanced image generation module: Used to reset the grayscale values ​​of the RGB three channels of the low-light grayscale image based on the brightness values ​​of the fourth and fifth images, and generate the enhanced image.

[0111] Embodiments of the present invention also provide an electronic device, which includes a processor and a memory. The memory stores at least one instruction, at least one program, code set, or instruction set. The at least one instruction, at least one program, code set, or instruction set is loaded and executed by the processor to implement the image enhancement method based on a vehicle surround view system as described in the method embodiments.

[0112] Embodiments of the present invention also provide a storage medium, which can be disposed in a server to store at least one instruction, at least one program, code set, or instruction set related to implementing the image enhancement method based on the vehicle surround view system in the method embodiments. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the image enhancement method based on the vehicle surround view system provided in the above method embodiments.

[0113] Optionally, in this embodiment, the storage medium may be located at at least one of the multiple network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0114] Implementing this embodiment has the following effects:

[0115] A combination of dual gamma transform and Laplacian transform is used to enhance the V channel of low-light images. The enhanced V channel brightness value is then subjected to multi-scale Retinex transform. Based on the gray values ​​of the RGB channels of the low-light grayscale image and the brightness ratio before and after the Retinex transform, the gray values ​​of the RGB channels are reset to improve the image enhancement effect and prevent color imbalance.

[0116] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0117] The foregoing description has fully disclosed the specific embodiments of the present invention. It should be noted that any modifications made to the specific embodiments of the present invention by those skilled in the art do not depart from the scope of the claims. Accordingly, the scope of the claims is not limited to the foregoing specific embodiments.

Claims

1. An image enhancement method based on a vehicle surround view system, characterized in that, Including the following steps: Obtain a low-light grayscale image; The low-light grayscale image is converted from the RGB color space to the HSV color space to obtain the V channel brightness value; Perform a dual gamma transform on the V channel brightness value to generate a first image brightness value and a second image brightness value; The V channel brightness value is subjected to Laplacian transform to generate the third image brightness value; The brightness values ​​of the first, second, and third images are weighted and fused to generate a fourth image brightness value; the fourth image brightness value is calculated using the following formula: in, This is the brightness value of the fourth image; The first image brightness value; This is the brightness value of the second image; The brightness value of the third image; Perform a multi-scale Retinex transform on the brightness value of the fourth image to generate the brightness value of the fifth image; The enhancement factors of the three RGB channels of the low-light grayscale image are obtained, and the grayscale values ​​of the three RGB channels of the low-light grayscale image are reset according to the enhancement factors to generate an enhanced image; the enhancement factor is the ratio of the brightness value of the fifth image to the brightness value of the fourth image. The step of performing a dual gamma transform on the V channel brightness value to generate a first image brightness value and a second image brightness value includes transforming the V channel brightness value using the following formula: in, This is the brightness value of the first image, i.e., the brightness value of the dual-gamma compressed image; This is the brightness value of the second image, i.e., the brightness value of the dual-gamma extended image; This refers to the brightness value of the V channel. These are the pixel coordinates in the image; Variables that control the degree of image enhancement.

2. The image enhancement method based on a vehicle surround view system according to claim 1, characterized in that, The steps for obtaining the low-light grayscale image include: A first low-light grayscale image is acquired, which is captured by a fisheye camera; Perform distortion correction on the first low-light grayscale image to generate a second low-light grayscale image; Multiple second low-light grayscale images are stitched together to generate the low-light grayscale image.

3. The image enhancement method based on a vehicle surround view system according to claim 1, characterized in that, The step of performing a Laplacian transform on the V channel brightness value to generate a third image brightness value includes transforming the V channel brightness value using the following formula: in Indicates the brightness value of the third image; The luminance value of the V channel; These are the Laplace transform coefficients; The V channel brightness value The second-order Laplace operator.

4. The image enhancement method based on a vehicle surround view system according to claim 1, characterized in that, The step of performing a multi-scale Retinex transform on the brightness value of the fourth image to generate the brightness value of the fifth image includes the following steps: Construct a Gaussian wrap function; The RGB channels of the image corresponding to the fourth image brightness value are filtered at multiple scales according to the Gaussian wrapping function and then weighted to generate a preset brightness value. The fifth image brightness value is generated by subtracting the preset brightness value from the fourth image brightness value.

5. The image enhancement method based on a vehicle surround view system according to claim 4, characterized in that, The step of performing a multi-scale Retinex transform on the brightness value of the fourth image to generate the brightness value of the fifth image includes generating the brightness value of the fifth image using the following formula: in, The brightness value of the fifth image; N is the number of scales; Weights corresponding to multiple scales; This is the brightness value of the fourth image; Let be a Gaussian wrapping function with N scales.

6. The image enhancement method based on a vehicle surround view system according to claim 5, characterized in that, The step of resetting the grayscale values ​​of the three RGB channels of the low-light grayscale image according to the enhancement factor includes resetting the grayscale values ​​of the three RGB channels of the low-light grayscale image according to the enhancement factor using the following formula: in, The grayscale value of the R channel in the enhanced image; As an enhancing factor, The grayscale value of the G channel in the enhanced image; The grayscale value of the B channel in the enhanced image; The grayscale value of the R channel in a low-light grayscale image; The grayscale value of the G channel in a low-light grayscale image; This represents the grayscale value of the B channel in a low-light grayscale image.

7. An image enhancement device based on a vehicle surround view system, characterized in that, include: Low-light grayscale image acquisition module: used to acquire low-light grayscale images; V-channel brightness value acquisition module: used to acquire low-light grayscale images; Dual-gamma transform module: used to perform dual-gamma transform on the brightness value of the V channel to generate a first image brightness value and a second image brightness value; Laplacian transform module: Performs Laplacian transform on the V channel brightness value to generate the third image brightness value; The fourth image brightness value generation module is used to weightedly fuse the first image brightness value, the second image brightness value, and the third image brightness value to generate a fourth image brightness value; the fourth image brightness value is calculated using the following formula: in, This is the brightness value of the fourth image; The first image brightness value; This is the brightness value of the second image; The brightness value of the third image; Fifth image brightness value generation module: Performs multi-scale Retinex transform on the fourth image brightness value to generate the fifth image brightness value; Enhanced image generation module: used to obtain the enhancement factors of the three RGB channels of the low-light grayscale image, and reset the grayscale values ​​of the three RGB channels of the low-light grayscale image according to the enhancement factors to generate the enhanced image; the enhancement factor is the ratio of the brightness value of the fifth image to the brightness value of the fourth image; The dual-gamma transformation module is specifically used to transform the V channel luminance value using the following formula: in, This is the brightness value of the first image, i.e., the brightness value of the dual-gamma compressed image; This is the brightness value of the second image, i.e., the brightness value of the dual-gamma extended image; This refers to the brightness value of the V channel. These are the pixel coordinates in the image; Variables that control the degree of image enhancement.

Images