A low-quality image enhancement method

Abstract

This invention discloses a method for enhancing low-quality images. The method involves normalizing the original low-quality RGB image, converting the normalized image from RGB to HSV format to obtain a first intermediate image, and obtaining the luminance component V1 of the first intermediate image. The normalized image is then inverted, and converted from RGB to HSV format to obtain a second intermediate image, and the luminance component V2 of the second intermediate image is obtained. Luminance components V1 and V2 are enhanced separately, and the two enhanced luminance components are weighted and fused to obtain a luminance component V3. Luminance component V3 is then inversely normalized, and the inversely normalized image is converted from HSV to RGB format to obtain the target image. This invention effectively expands the dynamic range in both low and high illumination, improving underexposed pixels while suppressing overexposed pixels. The image processed using this invention has a balanced visual effect and can effectively improve the contrast of low-quality images.

Description

Technical Field

[0001] This invention relates to the field of image processing, and in particular to a method for enhancing low-quality images. Background Technology

[0002] Image enhancement, an important aspect of image processing, aims to improve the visual effect of an image, making an originally blurry image clearer or emphasizing certain features of interest, thereby improving image quality. In image processing, gamma transform is often used to correct overexposed or underexposed grayscale images.

[0003] Gamma transformation is related to the nonlinear response of the cathode ray tube caused by electrostatic effects in electron guns or many imaging devices. The brightness produced by the cathode ray tube is not linearly related to the input voltage, leading to image distortion. Therefore, gamma correction is performed on the original grayscale or color image to compensate for this nonlinearity. The classic gamma enhancement method directly transforms the grayscale image as follows: y = x γ (γ = 1 / 2.2) can be applied to the three components of the RGB image separately, or the image can be converted to other formats before processing using the above formula. However, ubiquitous additive noise and uneven illumination can cause distortion in non-professional photos, so the input-output system remains non-linear, resulting in less than ideal correction effects from traditional gamma transform. Summary of the Invention

[0004] Purpose of the invention: To address the above problems, the purpose of this invention is to provide a method for enhancing low-quality images.

[0005] Technical solution: A low-quality image enhancement method of the present invention includes:

[0006] The original low-quality RGB image is normalized, and the normalized image is converted from RGB format to HSV format to obtain the first intermediate image. The luminance component V1 of the first intermediate image is then obtained.

[0007] The normalized image is inverted, and the inverted image is converted from RGB format to HSV format to obtain the second intermediate image. The luminance component V2 of the second intermediate image is then obtained.

[0008] The luminance component V1 and luminance component V2 are enhanced separately, and the two enhanced luminance components are weighted and fused to obtain luminance component V3.

[0009] The luminance component V3 is inversely normalized, and the inversely normalized image is converted from HSV format to RGB format to obtain the target image, which is then used as the final enhanced image.

[0010] Furthermore, the step of enhancing the luminance component V1 and the luminance component V2 respectively includes:

[0011] The luminance component V1 is enhanced using the following expression:

[0012]

[0013] In the above formula, V out1 This represents the enhanced luminance component, where x and y represent pixel positions; γ represents the enhancement index, expressed as:

[0014]

[0015] In the above formula, V represents the average value of all pixel luminance components in the original low-quality RGB image; This represents the average value of the top 20% of the luminance components in the original low-quality RGB image. This represents the average value of the bottom 20% of the brightness components in the original low-quality RGB image.

[0016] Furthermore, the step of enhancing the luminance component V1 and luminance component V2 respectively also includes:

[0017] The luminance component V2 is enhanced using the following expression:

[0018]

[0019] In the above formula, V out2 This represents the enhanced luminance component.

[0020] Furthermore, the expression for the luminance component V3 is:

[0021] V out (x,y)=(1-a)×V out1 +a×V out2

[0022] In the above formula, 'a' represents the fusion coefficient, and its expression is:

[0023] a = mean{V1(x,y)}

[0024] In the above formula, mean{} represents the mean.

[0025] Beneficial effects: Compared with the prior art, the significant advantages of this invention are as follows: This invention first normalizes the low-quality RGB image, then obtains the luminance component of the image, corrects the luminance components of the original image and the inverse image respectively, and then performs weighted fusion to obtain the luminance component of the enhanced image. Next, it performs inverse normalization and converts from HSV format to RGB format to obtain the final balanced high-quality image, effectively expanding the dynamic range of low and high illumination, improving underexposed pixels and suppressing overexposed pixels. The image processed by this invention has a balanced visual effect. This method is simple and practical, can effectively improve the contrast of low-quality images, and has good robustness. Attached Figure Description

[0026] Figure 1 This is a flowchart of the present invention;

[0027] Figure 2 This is a set of images showing the effects of different image enhancement methods on a set of original images.

[0028] Figure 3 The images show the effects of different image enhancement methods on two sets of original images. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.

[0030] like Figure 1 The diagram shown is a flowchart of a low-quality image enhancement method according to this embodiment. The method includes the following steps:

[0031] Step 1: Normalize the original low-quality RGB image, convert the normalized image from RGB format to HSV format to obtain the first intermediate image, and obtain the luminance component V1 of the first intermediate image.

[0032] Specifically, the RGB components of the original low-quality RGB image are normalized, reducing the pixel values ​​from the [0,255] interval to the [0,1] interval. The expression is as follows:

[0033] R' = R / 255

[0034] G' = G / 255

[0035] B' = B / 255

[0036] The normalized image is converted from RGB format to HSV format and designated as the first intermediate image. Image processing can then be performed in the HSV color space. The luminance component V1 of the first intermediate image is obtained using the following expression:

[0037] V1 = max(R', G', B').

[0038] Step 2: Invert the normalized image, convert the inverted image from RGB format to HSV format to obtain the second intermediate image, and obtain the luminance component V2 of the second intermediate image.

[0039] Specifically, the above image inversion refers to reversing the pixel intensity of the image, that is, turning the foreground of the normalized image into the background and the background into the foreground, as expressed by:

[0040] r = 1 - R'

[0041] g = 1 - G'

[0042] b = 1 - B'

[0043] The inverted image is converted from RGB format to HSV format to obtain the second intermediate image, and the luminance component V2 of the second intermediate image is obtained, with the expression as follows:

[0044] V1 = max(r, g, b)

[0045] Step 3: Enhance the luminance component V1 and luminance component V2 respectively, and then perform weighted fusion of the two enhanced luminance components to obtain luminance component V3.

[0046] Specifically, the steps of enhancing the luminance component V1 and luminance component V2 respectively include:

[0047] The luminance component V1 is enhanced using the following expression:

[0048]

[0049] In the above formula, V out1 This represents the enhanced luminance component, where x and y represent pixel positions; γ represents the enhancement index, expressed as:

[0050]

[0051] In the above formula, This represents the average value of all pixel luminance components in the original low-quality RGB image. This represents the average value of the top 20% of the luminance components in the original low-quality RGB image. This represents the average value of the bottom 20% of the brightness components in the original low-quality RGB image.

[0052] Furthermore, the steps of enhancing the luminance component V1 and luminance component V2 respectively also include:

[0053] The luminance component V2 is enhanced using the following expression:

[0054]

[0055] In the above formula, V out2 This represents the enhanced luminance component.

[0056] Furthermore, the expression for the luminance component V3 is:

[0057] V out (x,y)=(1-a)×V out1 +a×V out2

[0058] In the above formula, 'a' represents the fusion coefficient, and its expression is:

[0059] a = mean{V1(x,y)}

[0060] In the above formula, mean{} represents the mean.

[0061] Step 4: Perform inverse normalization on the luminance component V3, convert the inverse normalized image from HSV format to RGB format to obtain the target image, and use the target image as the final enhanced image.

[0062] Specifically, the luminance component V3 is inversely normalized, and the expression is:

[0063]

[0064] In the above formula, This indicates rounding up to the nearest integer.

[0065] To further illustrate the effectiveness of the low-quality image enhancement method described in this embodiment, please refer to... Figure 2-3 The image shows a comparison of the enhancement effects achieved using different methods. Figure 2 (a) is the original, low-quality color image. Figure 2 (b) The image enhanced using classical gamma transform. Figure 2 (c) An image enhanced using the method in this embodiment; Figure 3 (a) is the original color image. Figure 3 (b) An image enhanced using classical gamma transform. Figure 3 (c) The image enhanced using the method described in the paper "low-light image enhancement using the camera response model" (IEEE Trans. Circuits Syst. Video Technol. 2019, 29, 968–981). Figure 3 (d) is an image enhanced using the method described in this embodiment. From Figure 2-3As can be seen from the above, the effectiveness of the image enhancement method of this embodiment has been verified on multiple images. This method can effectively expand the dynamic range of low and high illumination, both improving underexposed pixels and suppressing overexposed pixels. The image processed by the image enhancement method of this embodiment has a balanced visual effect and contrast. Both the subjective and objective results are superior to the classic gamma transform and other image enhancement methods.

Claims

1. A low-quality image enhancement method, characterized by, include: The original low-quality RGB image is normalized, and the normalized image is converted from RGB format to HSV format to obtain the first intermediate image. The luminance component V1 of the first intermediate image is then obtained. The normalized image is inverted, and the inverted image is converted from RGB format to HSV format to obtain the second intermediate image. The luminance component V2 of the second intermediate image is then obtained. Luminance component V1 and luminance component V2 are enhanced separately, and the two enhanced luminance components are then weighted and fused to obtain luminance component V3; specifically: The luminance component V1 is enhanced using the following expression: In the above formula, V out1 This represents the enhanced luminance component, where x and y represent pixel positions; γ represents the enhancement index, expressed as: In the above formula, This represents the average value of all pixel luminance components in the original low-quality RGB image. This represents the average value of the top 20% of the luminance components in the original low-quality RGB image. This represents the average value of the bottom 20% of the luminance components in the original low-quality RGB image. The luminance component V2 is enhanced using the following expression: In the above formula, V out2 represents the enhanced luminance component; The luminance component V3 is inversely normalized, and the inversely normalized image is converted from HSV format to RGB format to obtain the target image, which is then used as the final enhanced image.

2. The low-quality image enhancement method of claim 1, wherein, The expression for the luminance component V3 is: V out (x,y) = (1 - a) x V out1 + a x V out2 In the above formula, 'a' represents the fusion coefficient, and its expression is: a = mean{V1(x,y)} In the above formula, mean{} represents the mean.

Images