Underwater image color correction method based on natural color prior

Abstract

The invention relates to an underwater image color correction method based on natural color prior, and belongs to the technical field of image processing. According to the method, a natural color prior model is constructed by analyzing color channel distribution characteristics of a natural land image in Lab, RGB and HSV color spaces. The method comprises the following steps: firstly, performing color compensation on an underwater image based on a priori model, and adjusting a channel a and a channel b mean value in a Lab color space to be close to zero so as to more accurately compensate a red component; and then, a color stretching method is further adopted, so that the distribution range of R, G and B channels in the RGB color space is wider and more uniform. And finally, converting the image subjected to color stretching into an HSV color space, and performing saturation adjustment on S channels to enable the S channels to be mainly distributed in a suitable saturation interval so as to present natural color information. According to the method, color distortion of various types of underwater images can be effectively corrected, and the visual quality of the underwater images is remarkably improved.

Description

Technical Field

[0001] The invention relates to an underwater image color correction method based on natural color prior, belonging to the technical field of image processing. Background Art

[0002] As an important carrier of marine information, underwater images play a key role in underwater robots, marine engineering and ecological monitoring. However, due to the selective absorption of light by water, different colors of light will experience different degrees of attenuation during propagation, especially the high attenuation of red light, resulting in blue and green color distortion in the image. In order to solve the problem of color distortion of underwater images, a variety of color correction methods have been proposed. However, these methods are often not ideal when processing the red component, which usually leads to the presence of red artifacts in the resulting image and it is difficult to present natural color information. Specifically, the generation of red artifacts mainly comes from two aspects: on the one hand, some color correction algorithms may introduce additional red artifacts during the image enhancement process; on the other hand, underwater images captured by camera sensors may inherently carry red artifacts. Under the interference of these factors, underwater images face serious quality degradation. Therefore, developing an underwater image color correction method that can effectively avoid red artifacts and achieve natural color performance has important application value in the field of underwater image processing. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide an underwater image color correction method based on natural color prior, aiming to solve the technical problem that different colors of light experience different degrees of attenuation during underwater propagation, resulting in underwater image color distortion and difficulty in ensuring the visual quality of underwater images.

[0004] To achieve the above-mentioned purpose, the present invention proposes a method for underwater image color correction based on natural color prior. The method establishes a natural color prior model based on the distribution characteristics of color channels of natural land images in three color spaces: Lab, RGB and HSV. Specifically, the natural color prior model shows that: the average values ​​of a and b channel pixels in the Lab color space are close to zero, the R (red), G (green) and B (blue) channels in the RGB color space are widely and evenly distributed, and the values ​​of the S (saturation) channel in the HSV color space are mainly concentrated in the appropriate saturation range of 0 to 0.5. Based on this prior model, color compensation, color stretching and saturation adjustment are performed on the underwater color distorted image in sequence to effectively suppress red artifacts and correct the color distortion of the underwater image, thereby presenting more natural color information;

[0005] The technical solution of the present invention is: a method for underwater image color correction based on natural color prior, comprising the following steps:

[0006] Step 1: Input the color-distorted underwater image, convert the underwater image into Lab color space, decompose it into three channels: L, a, and b, and perform color compensation on the a and b channels;

[0007] Specifically, the purpose of step 1 is to avoid the influence of brightness on color adjustment in the RGB color space and the instability caused by high color sensitivity in the HSV color space, so color compensation is first performed in the Lab color space that is closer to visual perception characteristics.

[0008] Step 2: Convert the color-compensated underwater image to the RGB color space and decompose it into three channels: R, G, and B. Perform color stretching on the R, G, and B channels.

[0009] Specifically, the purpose of step 2 is that the distribution range of R, G and B channels in the color compensated image is narrow, resulting in a large amount of color information being hidden, which in turn causes the gray phenomenon. In order to eliminate this gray phenomenon, the compensated image is converted into the RGB color space and decomposed into three channels of R, G and B, and color stretching is performed to expand the distribution range of each channel to make it more extensive and uniform. In addition, the color stretching method can also avoid excessive stretching of the red channel and prevent the appearance of a purple-red background.

[0010] Step 3: Convert the underwater image after color stretching into HSV color space and decompose it into three channels: H, S, and V. Adjust the S channel so that its distribution is concentrated in the moderate saturation range of 0 to 0.5 to achieve underwater image color correction.

[0011] Specifically, the purpose of step 3 is that in view of the close relationship between the naturalness and saturation of color, undersaturation and oversaturation will hinder the presentation of natural color information. In order to obtain more natural color information, the color stretched image is converted to the HSV color space and decomposed into three channels: H (hue), S, and V (value). The S channel is appropriately adjusted so that its distribution is concentrated in an appropriate saturation range, thereby enhancing the overall color saturation while avoiding local red artifacts caused by oversaturation, and finally generating an underwater image with natural colors.

[0012] The specific method for color compensation of the a and b channels in step 1 is as follows:

[0013] Step 1.1: Perform initial color compensation on channels a and b, retaining the contour information. The calculation formula is as follows:

[0014] I a1 (x) = I a (x)-μ·W(x)·GI a (x) (1)

[0015] Ib1 (x) = I b (x)-μ·W(x)·GI b (x) (2)

[0016] Among them, I a1 (x) and I b1 (x) represents the a and b channels after the initial color compensation; I a (x) and I b (x) represents the original a and b channels respectively; GI a (x) and GI b (x) represents the Gaussian filtered versions of a and b channels respectively; W(x) is used to maintain the brightness at the light source position; μ is a compensation parameter;

[0017] Step 1.2: Count the pixel mean and variance of channels a and b. The calculation formula is as follows:

[0018]

[0019]

[0020] in, and Respectively represent the average values ​​of a and b channels after the initial color compensation; and Respectively represent the variance of a and b channels after the initial color compensation; H and W represent the height and width of the input channel;

[0021] Step 1.3: Perform enhanced color compensation on the a and b channels to make the mean close to zero, thereby accurately restoring the red component information. The calculation formula is as follows:

[0022]

[0023] Among them, I a2 (x) and I b2 (x) respectively represent the a and b channels after enhanced color compensation; and Respectively represent the maximum values ​​of a and b channels after the initial color compensation; and They represent the minimum values ​​of the a and b channels after the initial color compensation respectively; α and β are both very small values ​​to prevent the denominator from being zero; sign is the sign function used to retain positive and negative color information.

[0024] Specifically, by adjusting the mean values ​​of the a and b channels to be close to zero, the red component information can be compensated more accurately;

[0025] The specific method of color stretching for the R, G and B channels in step 2 is as follows:

[0026] Step 2.1: Find the two key pixel points q for each channel of R, G and B c,1 and q c,2 , used to determine the stretching degree of the R, G, and B channels, the calculation formula is as follows:

[0027]

[0028] q c,2 =1-q c,1 (10)

[0029] in, Represents the color compensation image I cc The average value in channel c∈{R,G,B}, Represent the color compensation image I cc The R, G and B channels, They are color compensated images I cc The average value among the R, G and B channels;

[0030] Step 2.2: Based on the two key pixel points q c,1 and q c,2 , the pixel values ​​of the color-compensated image in the channel c∈{R,G,B} are processed using the quantile function, so that the color-compensated image is c,1 and q c,2 Moderate stretching is achieved between the two images to obtain the image I processed by the quantile function. cs1 (x), the calculation formula is as follows:

[0031]

[0032] Among them, Q(q c,1 ) and Q(q c,2 ) represent the two key pixel points q corresponding to the channel c∈{R,G,B} c,1 and q c,2 The strength value of

[0033] Step 2.3: For image I cs1 (x) performs linear stretching to expand the dynamic range of the R, G, and B channels, thereby generating a color-stretched image. The calculation formula for linear stretching is as follows:

[0034]

[0035] Among them, I cs2 (x) represents the color stretched image; I cs1_max and I cs1_minRespectively represent the image I after being processed by the quantile function cs1 The maximum and minimum values ​​of (x).

[0036] The specific method for adjusting the S channel in step 3 is as follows:

[0037] Step 3.1: Count the pixel mean of the S channel. The calculation formula is as follows:

[0038]

[0039] in, is the color stretched image I cs2 (x) S channel in HSV color space; Represents the average value of the S channel of the color stretched image in the HSV color space; H and W represent the height and width of the input channel;

[0040] Step 3.2: Adjust the saturation of the S channel of the color stretched image. The saturation adjustment formula is as follows:

[0041]

[0042] in, Represents the image obtained after saturation adjustment. When the color stretches the S channel of the image When the pixel value is lower than the midpoint value of 0.5, the item is negative, resulting in a result greater than 1, to increase saturation; conversely, when the saturation value When the midpoint value is 0.5, the item is positive, so that the result is less than 1, so as to prevent the local red artifact effect caused by over-saturation enhancement. Based on the above steps, underwater image color correction is achieved.

[0043] The present invention has outstanding substantive features and significant progress compared to the prior art. Specifically, the beneficial effects of the present invention are:

[0044] 1. The present invention aims at the high attenuation characteristics of red light in underwater environments and can effectively process red component information, which not only avoids the introduction of additional red artifacts in the color correction process, but also eliminates the inherent red artifacts that may be generated by the camera sensor when capturing underwater images;

[0045] 2. The present invention proposes a natural color prior model by statistically analyzing the color channel distribution characteristics of natural land images in multiple color spaces (such as Lab, RGB, and HSV). Based on the prior model, color compensation is performed on the a and b channels of the Lab color space; color stretching is performed on the R, G, and B channels of the RGB color space; and saturation adjustment is performed on the S channel of the HSV color space, thereby solving the color distortion problem of underwater color-biased images and obtaining high-quality underwater natural color images;

[0046] 3. Since the present invention is based on the natural color prior of land images, it is not only suitable for correcting underwater color-distorted images, but also for improving the quality of images in other color-shifted scenes such as nighttime and sandstorms.

[0047] Based on the above reasons, the present invention can be widely promoted in the fields of image processing with color cast degradation. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0049] Figure 1 It is a schematic diagram of the process of the present invention;

[0050] Figure 2 It is the enhancement result of the present invention and other underwater image processing methods for green distorted underwater images;

[0051] Figure 3 It is the enhancement result of the present invention and other underwater image processing methods for blue distorted underwater images;

[0052] Figure 4 The present invention and other underwater image processing methods are used to enhance the yellow distorted underwater image;

[0053] Figure 5 This is the enhancement result of the underwater image with red artifacts obtained by the present invention and other underwater image processing methods. DETAILED DESCRIPTION

[0054] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0055] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0056] Example 1: Figure 1 As shown, a method for underwater image color correction based on natural color prior is provided, comprising the following steps:

[0057] Step 1: Input the color-distorted underwater image, convert the underwater image into Lab color space, decompose it into three channels: L, a, and b, and perform color compensation on the a and b channels;

[0058] Specifically, in order to reduce the interference of brightness on color adjustment in the RGB color space and reduce the instability caused by high color sensitivity in the HSV color space, the present invention preferentially performs color compensation in the Lab color space that is closer to the human visual perception characteristics. First, a color-distorted underwater image is input, the underwater image is converted into the Lab color space, and decomposed into three channels, L, a, and b. The a and b channels are initially color compensated to retain the image contour information and cause the mean of the a and b channels to approach zero. The calculation formula of the initial color compensation model is as follows:

[0059] I a1 (x) = I a (x)-μ·W(x)·GI a (x) (1)

[0060] I b1 (x) = I b (x)-μ·W(x)·GI b (x) (2)

[0061] Among them, I a1 (x) and I b1 (x) represents the a and b channels after the initial color compensation; I a (x) and I b (x) represents the original a and b channels respectively; GI a (x) and GI b(x) represents the Gaussian filtered versions of the a and b channels, respectively, which can effectively smooth out the subtle fluctuations of local colors and make color adjustment more stable; W(x) is used to maintain the brightness at the light source position; μ is a compensation parameter, which is set to 0.7 based on experience in the present invention to optimize the attenuation effect of the channel.

[0062] Since the above operation has limited effect on equalizing the mean of a and b channels for some severely color-distorted underwater images, in order to enhance the applicability of this method to images with strong color cast, the present invention uses the pixel mean and variance of a and b channels to perform enhanced color compensation, so that the mean of a and b channels is closer to zero, thereby more accurately restoring the red component information. The calculation formulas for the pixel mean and variance of a and b channels are as follows:

[0063]

[0064] in, and Respectively represent the average values ​​of a and b channels after the initial color compensation; and Respectively represent the variance of a and b channels after the initial color compensation; H and W represent the height and width of the input channel;

[0065] Using the calculated pixel mean of channels a and b and and variance and To enhance the color compensation of a and b channels. The calculation formula for enhanced color compensation is as follows:

[0066]

[0067] Among them, I a2 (x) and I b2 (x) respectively represent the a and b channels after enhanced color compensation; and Respectively represent the maximum values ​​of a and b channels after the initial color compensation; and They represent the minimum values ​​of the a and b channels after the initial color compensation, respectively; α and β are both very small values ​​to prevent the denominator from being zero, and in the present invention, α and β are set to 0.001; sign is a sign function used to retain positive and negative color information so as to more effectively process the red component information based on the balance between the a and b channels.

[0068] Step 2: Convert the color-compensated underwater image to the RGB color space and decompose it into three channels: R, G, and B. Perform color stretching on the R, G, and B channels.

[0069] Specifically, due to the narrow distribution range of R, G and B channels in the color-compensated image, a large amount of color information is hidden, which in turn causes the gray phenomenon. To eliminate this gray phenomenon, the color-compensated image is converted to the RGB color space, and the R, G and B channels are extracted; the two key pixel points q of each R, G and B channel are calculated. c,1 and q c,2 , to determine the stretching degree of each channel. The calculation formula is as follows:

[0070]

[0071] q c,2 =1-q c,1 (10)

[0072] in, Represents the color compensation image I cc The average value in channel c∈{R,G,B}, Respectively represent the color compensation image I cc The R, G and B channels, They are color compensated images I cc The average value among the R, G and B channels;

[0073] Based on the calculated pixel q c,1 and q c,2 , the pixel values ​​of the image in channel c∈{R,G,B} are processed by quantile function to obtain the image I after quantile function processing cs1 (x). This process ensures that the image pixel q c,1 and q c,2 The image I after the quantile function processing is properly stretched to prevent the purple-red background caused by excessive stretching of the red channel. cs1 (x) is calculated as follows:

[0074]

[0075] Among them, Q(q c,1 ) and Q(q c,2 ) represent the two key pixel points q corresponding to the channel c∈{R,G,B} c,1 and q c,2 The strength value of

[0076] For the image I processed by the quantile function cs1 (x) Perform linear stretching to further expand the dynamic range of the R, G, and B channels, thereby generating a color-stretched image I cs2 (x). Its linear stretching expression is as follows:

[0077]

[0078] Among them, I cs2 (x) represents the color stretched image; I cs1_max and I cs1_min Respectively represent the image I after being processed by the quantile function cs1 (x) maximum and minimum values;

[0079] Step 3: Convert the underwater image after color stretching into HSV color space and decompose it into three channels: H, S, and V. Adjust the S channel so that the S channel distribution is concentrated in the moderate saturation range of 0 to 0.5 to achieve underwater image color correction;

[0080] Specifically, the color stretched image I cs2 (x) Convert to HSV color space and extract S channel; Calculate the pixel mean of S channel, and the calculation formula is as follows;

[0081]

[0082] in, is the color stretched image I cs2 (x) S channel in HSV color space; Represents the average value of the S channel of the color stretched image in the HSV color space; H and W represent the height and width of the input channel;

[0083] According to the distance relationship between the saturation channel mean of the color stretched image and the midpoint value 0.5, the saturation adjustment model is applied to the S channel of the color stretched image. The specific saturation adjustment formula is as follows:

[0084]

[0085] in, Represents the image obtained after saturation adjustment. When the color stretches the S channel of the image When the pixel value is lower than the midpoint value of 0.5, the item is negative, resulting in a result greater than 1, to increase saturation; conversely, when the saturation value When the midpoint value is 0.5, the item is positive, so that the result is less than 1, so as to prevent the local red artifact effect caused by over-saturation enhancement. Based on the above steps, the present invention can solve the color distortion problem of underwater images and provide high-quality underwater natural color images.

[0086] In order to verify the advantages of the present invention over existing methods in processing underwater color-shifted images, especially its effectiveness in eliminating red artifacts and restoring natural color information, various types of color-distorted underwater images were selected as test sets, and compared with the following underwater image processing methods: Adaptive Standardization and Normalization Networks (ASNN), Underwater Image Enhancement Network (WN), Principal Component Fusion of Foreground and Background (PCFB), Color-Balanced Locally Adjustable Underwater Image Enhancement (CBLA), Hybrid Fusion Method (HFM), Adaptive Color Correction and Dehazing (ACCD), Bayesian Retinex (BR), Rank-One Prior Plus (ROP+). The comparison results are shown in the form of test images.

[0087] Figure 2 The comparative test results on correcting green-distorted underwater images are presented. The results show that the ROP+ and CBLA methods will cause local overexposed areas and introduce additional red artifacts. Although the PCFB and HFM methods can avoid the overexposure problem, there is still excessive red shift in some areas. In addition, the ACCD, WN and BR methods do not enhance the red information enough, resulting in blue or green color cast in some areas, while excessive enhancement of red information can also cause red artifacts. Although the ASNN method can effectively remove the green color cast, the method may still have red artifacts in the foreground area. In contrast, the algorithm of the present invention can not only correct the green shift, but also avoid additional red artifacts, so as to have reasonable colors and provide more detailed local texture information.

[0088] Figure 3The comparative test results on correcting blue-distorted underwater images are shown. In the enhanced results, the PCFB method introduces red artifacts in the bright area and produces a purple-red hue in the background. The ASNN and HFM methods over-enhance the local bright area, resulting in purple-red textures in the dark blue background. In addition, the ROP+ and BR methods have overexposure problems and are difficult to effectively correct the blue color shift. Although the WN and CBLA methods can alleviate the overexposure phenomenon, there are still obvious local blue and red areas in their results. The ACCD method can correct the blue deviation, but may over-enhance the red component. In contrast, the present algorithm can not only effectively correct the blue hue shift, but also avoid the generation of local red areas and purple-red backgrounds, thereby restoring a more natural color effect.

[0089] Figure 4 Comparative test results on correcting yellow-distorted underwater images are presented. The results show that the PCFB, CBLA, ACCD, and BR methods over-enhance the red component, resulting in additional red artifacts. The WN method fails to effectively remove the yellow color cast. In addition, the ROP+, ASNN, and HFM methods cause local overexposure with unnatural color performance. In contrast, the proposed method is able to correct the yellow color cast and avoid introducing additional red artifacts, thereby providing a more natural underwater image feature performance.

[0090] like Figure 2 , 3 As shown in Figure 4, when correcting other types of underwater color-distorted images, the introduction of red artifacts can be effectively prevented while obtaining reasonable color effects. Additional tests were conducted on underwater images with inherent red artifacts to evaluate the robustness in processing underwater color-shifted images.

[0091] Figure 5 The correction comparison results for underwater images with red artifacts are shown. It can be observed that the ROP+ method causes local overexposure and leads to the appearance of local red areas. Although the ASNN, WN, PCFB, CBLA, HFM, BR and ACCD methods perform well in preventing local overexposure, they fail to eliminate the inherent red artifacts in the image. In contrast, the present method can effectively remove the inherent red artifacts and obtain ideal color information, which helps to improve the clarity of underwater color-shifted images.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for underwater image color correction based on natural color prior, characterized in that: The following steps are involved: Step 1: Input the color-distorted underwater image, convert the underwater image into Lab color space, decompose it into three channels: L, a, and b, and perform color compensation on the a and b channels; Step 2: Convert the color-compensated underwater image to the RGB color space and decompose it into three channels: R, G, and B. Perform color stretching on the R, G, and B channels. Step 3: Convert the underwater image after color stretching into HSV color space and decompose it into three channels: H, S, and V. Adjust the S channel so that its distribution is concentrated in the moderate saturation range of 0 to 0.5 to achieve underwater image color correction.

2. The underwater image color correction method based on natural color prior according to claim 1, characterized in that: The specific method for color compensation of the a and b channels in step 1 is as follows: Step 1.1: Perform initial color compensation on channels a and b, retaining the contour information. The calculation formula is as follows: I a1 (x)=I a (x)-μ·W(x)·GI a (x) (1) I b1 (x)=I b (x)-μ·W(x)·GI b (x) (2) Among them, I a1 (x) and I b1 (x) represents the a and b channels after the initial color compensation; I a (x) and I b (x) represents the original a and b channels respectively; GI a (x) and GI b (x) represents the Gaussian filtered versions of a and b channels respectively; W(x) is used to maintain the brightness at the light source position; μ is a compensation parameter; Step 1.2: Count the pixel mean and variance of channels a and b. The calculation formula is as follows: in, and Respectively represent the average values ​​of a and b channels after the initial color compensation; and Respectively represent the variance of a and b channels after the initial color compensation; H and W represent the height and width of the input channel; Step 1.3: Perform enhanced color compensation on the a and b channels to make the mean close to zero, thereby accurately restoring the red component information. The calculation formula is as follows: Among them, I a2 (x) and I b2 (x) respectively represent the a and b channels after enhanced color compensation; and Respectively represent the maximum values ​​of a and b channels after the initial color compensation; and They represent the minimum values ​​of the a and b channels after the initial color compensation respectively; α and β are both very small values ​​to prevent the denominator from being zero; sign is the sign function used to retain positive and negative color information.

3. The underwater image color correction method based on natural color prior according to claim 1, characterized in that: The specific method of color stretching for the R, G and B channels in step 2 is as follows: Step 2.1: Find the two key pixel points q for each channel of R, G and B c,1 and q c,2 , used to determine the stretching degree of the R, G, and B channels, the calculation formula is as follows: q c,2 =1-q c,1 (10) in, Represents the color compensation image I cc The average value in channel c∈{R,G,B}, Represent the color compensation image I cc R, G and B channels, They are color compensated images I cc The average value among the R, G and B channels; Step 2.2: Based on the two key pixel points q c,1 and q c,2 , the pixel values ​​of the color-compensated image in the channel c∈{R,G,B} are processed using the quantile function, so that the color-compensated image is c,1 and q c,2 Moderate stretching is achieved between the two images to obtain the image I processed by the quantile function. cs1 (x), the calculation formula is as follows: Among them, Q(q c,1 ) and Q(q c,2 ) represent the two key pixel points q corresponding to the channel c∈{R,G,B} c,1 and q c,2 The strength value of Step 2.3: For image I cs1 (x) performs linear stretching to expand the dynamic range of the R, G, and B channels, thereby generating a color-stretched image. The calculation formula for linear stretching is as follows: Among them, I cs2 (x) represents the color stretched image; I cs1_max and I cs1_min Respectively represent the image I after being processed by the quantile function cs1 The maximum and minimum values ​​of (x).

4. The underwater image color correction method based on natural color prior according to claim 1, characterized in that: The specific method for adjusting the S channel in step 3 is as follows: Step 3.1: Count the pixel mean of the S channel. The calculation formula is as follows: in, is the color stretched image I cs2 (x) S channel in HSV color space; Represents the average value of the S channel of the color stretched image in the HSV color space; H and W represent the height and width of the input channel; Step 3.2: Adjust the saturation of the S channel of the color stretched image. The saturation adjustment formula is as follows: in, Represents the image after saturation adjustment. When the color stretches the S channel of the image When the pixel value is lower than the midpoint value 0.5, the item is negative, resulting in a result greater than 1, to increase saturation; conversely, when the saturation value When the midpoint value is 0.5, the item is positive, so that the result is less than 1, so as to prevent the local red artifact effect caused by over-saturation enhancement. Based on the above steps, underwater image color correction is achieved.

Images