CN105205792A - Underwater image enhancement method based on brightness and chrominance separation - Google Patents

Abstract

The invention relates to an underwater image enhancement method based on luminance and chrominance separation, comprising: separation of luminance and chrominance components of an input underwater color image; denoising of luminance components based on a median filter; brightness equalization based on an adaptive histogram Component contrast enhancement; 4) chroma component color cast correction; increase chroma component saturation; restore image. The invention can improve the visible quality of underwater images.

Description

An Underwater Image Enhancement Method Based on Luminance and Chromaticity Separation

technical field

The invention relates to the image enhancement technology in the field of computer vision, in particular to the degraded image enhancement technology for underwater cameras.

Background technique

my country is a large ocean country, with a coastline of 18,000 kilometers and an ocean territory of more than 3 million square kilometers. The exploration and development of marine resources are of great significance and value to our national economy and military. Autonomous underwater vehicles (AUV for short) are widely used in the exploration and development of offshore resources (pipeline laying, seabed investigation, etc.), marine ecological monitoring, and military fields. Most of the current AUVs are equipped with optical imaging and image processing recognition systems, and visible light images are one of the main sources of information for AUVs. However, due to the limitation of the harsh underwater imaging environment, the acquired underwater images generally have unfavorable factors such as low contrast, blur, and color cast. The degradation of image visual quality will seriously affect the performance of subsequent feature extraction and object recognition. Therefore, the research on improving the visual quality of the original underwater image with the help of image processing technology has attracted more and more attention from researchers, and the related research has been increasing, which has become one of the hot research issues in current image processing ^[1] .

The research on underwater image enhancement technology at home and abroad began at the beginning of this century. After more than ten years of development, researchers have proposed many methods. According to the different image enhancement techniques used in various methods, the existing methods can be divided into three categories: methods based on classical image enhancement techniques ^[2,3,4] , methods based on color constancy theory ^[5,6] , and hybrid methods ^[7,8] . In terms of domestic invention patents, Chen Mingsong et al. ^[9] (application number CN201310628073) proposed an underwater image enhancement technology based on the combination of homomorphic filtering and biorthogonal wavelet threshold filtering. They first reduced the influence of uneven illumination with the help of homomorphic filtering technology, and then biorthogonal wavelet threshold filtering method removed the noise of the image. Li Yibing et al. ^[10] (application number CN201310628073) proposed an underwater image enhancement method based on Retinex theory. They first converted the underwater image to the HSV color space, using the S component for denoising and the V component for Retinex enhancement.

references

[1] Raimondo Schettini and Silvia Corchs, "Underwater Image Processing: State of the Art of Restoration and Image Enhancement Methods", EURASIP Journalon Advances in Signal Processing, 2012.

[2] Garcia, R., Nicosevici, T., and Cufí, X., "On The Way to Solve Lighting Problems in Underwater Imaging", Proceeding of the IEEE OCEANS Conference (OCEANS), 2002, pp.1018-1024.

[3] Hitam, M.S., Kuala Terengganu, Yussof, W.N.J.H.W., Awalludin, E.A., Bachok, Z., "Mixturecontrastlimitedadaptivehistogramequalizationforunderwaterimageenhancement", ComputerApplicationsTechnology (ICCAT), 2013InternationalConferenceon, 2013, 1-5.

[4] Chen Congping, Wang Jian, Zou Lei, Zhang Fajun, "An effective method for enhancing low-contrast underwater images", Laser and Infrared, 42(5), 2012, pp.567-571.

[5] Birgit Henke, Matthias Vahl, and Zhiliang Zhou, "Removing Color Cast of Underwater Images through Non-Constant Color Constancy Hypothesis", ISPA2013, 20-24

[6] Xueyang Fu, Peixian Zhuang, and Yinghao Liao, "ARetinex-based Enhancement Approach for Single Underwater Image", IEEE International Conference on Image Processing (ICIP), pp.4572-4576, 2014.

[7] K.Iqbal, Odetayo, M., James, A., Salam, R.A., Talib, A.Z.H., "Enhancing the low quality images using Unsupervised ColourCorrection Method", 2010 IEEE International Conference on Systems Man and Cybernetics (SMC), vol., no., pp.1703, 1709, 10-13 Oct. 2010.

[8] Cosmin Ancuti, Codruta Orniana Ancuti, Tom Haber and Philippe Bekaert, "Enhancing Underwater Images and Videos by Fusion", CVPR2012.

[9] Chen Mingsong, Qin Lin, Kang Yanmei, etc., "Underwater image enhancement processing method", Chinese invention patent, application number: CN201310628073

[10] Li Yi, Fu Qiang, Ye Fang, etc. "Underwater image enhancement method based on HSV color space combined with Retinex", application number: CN201210359018,

[11] Lin Fuzong, "Basics of Multimedia Technology" (3rd Edition), 2009, Tsinghua University Press

Contents of the invention

Aiming at underwater degraded images, the present invention proposes an underwater image enhancement method based on luminance and chrominance separation, so as to achieve the purpose of improving the visual quality of underwater images. Technical scheme of the present invention is as follows:

An underwater image enhancement method based on luminance and chromaticity separation, comprising the following steps:

1) Separate the luminance and chrominance components of the input underwater color image

Convert the input underwater color image from the RGB color space to the YCbCr color space to separate the luminance and chrominance components. In the YC _b C _r color space, Y represents the luminance component, and C _b and C _r represent the blue and red biases respectively. displacement.

2) Luminance component denoising based on median filter

Select the median filter to filter the brightness component Y, and the processing result is represented by _Y1 ;

3) Brightness component contrast enhancement based on adaptive histogram equalization

The contrast-limited histogram equalization method (CLAHE) is selected to improve the contrast of the denoised brightness component _Y1 , and the processing result is represented by _Y2 ;

4) Chroma component color cast correction

Use C _b and C _r to represent the two chromaticity component diagrams in the YCbCr space, use M _b and M _r to represent the mean values of C _b and C _r respectively, and use the following formula to adjust the values of the two chromatic components of C _b and C _r , Complete the color cast correction of the chroma component:

${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; 128</span>}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}}$

${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; 128</span>}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}}$

In the formula, use C _b1 and C _r1 to represent the color cast correction results;

5) Increase the saturation of the chroma component

Use the Sigmoid function to stretch the chroma component that has undergone color cast correction to obtain enhanced chroma saturation. The Sigmoid function used is:

$\mathrm{<span\; class="notranslate">\; g</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 256</span>}}{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\frac{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 128</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; 16</span>}}}}$

In the formula, when the input f is C _b1 , the output g represents the saturation enhancement result of the blue chroma component, represented by C _b2 ; when the input f is C _r1 , the output g represents the saturation enhancement result of the red chroma component, represented by C _r2 means;

6) Restoring the image

Combine the luminance component Y ₂ and the chrominance components C _b2 and C _r2 processed in step 3), and re-transform back to the RGB color space.

Step 3 can use "Rayleigh distribution".

Description of drawings

Fig. 1 is a flow chart of the present invention.

Figure 2 lists some underwater image construction test sample sets and processing results examples. The left side is the color degraded underwater image, and the right side is the enhanced result image.

Fig. 3 is a comparison chart of the processing results of the present invention and the "automatic color tone + automatic contrast adjustment" scheme provided by PhotoshopCS12. Among them, the first row is the original input image, the second row is the Photoshop processing result, and the third row is the processing result of the proposed method.

Detailed ways

The invention uses an underwater image enhancement method based on brightness and chromaticity separation. The proposed method first converts the input image from the RGB color space to the YCbCr color space; for the Y component representing the brightness component, combined with the median filter technology to filter out the image noise, and using the adaptive histogram equalization technology to stretch the contrast of the image; for the representative For the Cb and Cr components of the chroma component, a new scheme is used to correct the color cast and improve the color saturation; finally, the enhanced components are recombined and transformed back to the RGB color space. See Figure 1, the specific steps are as follows:

1 input color image luma chrominance component separation

The proposed method of the present invention separates luminance and chrominance components by converting an input color image from RGB color space to YCbCr color space. The YC _b C _r color space is often used in the field of image/video coding, and well-known file formats such as JPEG and MPEG all use this color space. In the YC _b C _r color space, Y represents the luminance component, and C _b and C _r represent blue and red offsets, respectively. For the conversion formula from RGB color space to YC _b C _r color space, please refer to related literature [11].

2 Luminance component denoising based on median filter

During the imaging process of the underwater image, it is affected by suspended particles in the water and ambient light, and there is noise in some areas. It is necessary to remove the noise before image enhancement processing. The noise energy in the image is usually concentrated in high frequency, so a low-pass filter can be used for denoising. Compared with the ordinary Gaussian smoothing filter, the median filter can preserve the detail information well while denoising. Based on the above analysis, the proposed method uses a median filter to smooth the luminance component to achieve the purpose of denoising.

The method proposed in the present invention selects a median filter with a window size of 5×5 to perform filtering processing on the brightness component Y, and the processing result is represented by Y ₁ .

3 Contrast enhancement of luminance component based on adaptive histogram equalization

The traditional histogram equalization technique (HE for short) does not consider the location information, and the enhancement result is not good. Adaptive histogram equalization (AHE for short) overcomes the shortcomings of HE well, but there is still the problem of excessive amplification of noise in local areas. Contrast-Limited Histogram Equalization (CLAHE for short), as a typical AHE method, can effectively overcome the shortcomings of the former, and is widely used in image enhancement technology.

The present invention selects CLAHE technology to improve the contrast of the brightness component after denoising, wherein the size of each sub-block is 8×8, the clipping value is set to 0.01, the distribution adopts “Rayleigh distribution”, and the processing result is represented by _Y2 .

4 chroma component color cast correction

When the lighting conditions change, the reflection component of the object in the scene changes, which is prone to color cast. For underwater images, when visible light is transmitted in water, it is affected by the combined effects of scattering and absorption of suspended particles in the water, resulting in obvious color cast. Experiments have shown that light with longer wavelengths (such as red light and yellow light) in visible light is more easily absorbed by water molecules than light with shorter wavelengths (such as blue light and green light), that is, the longer the wavelength, the more obvious the absorption attenuation. Therefore, common underwater scenes usually appear blue (or green).

The invention proposes a new color cast correction scheme based on Cb-Cr chromaticity space. Use C _b and C _r to represent the two chromaticity component diagrams in the YCbCr space, and use M _b and M _r to represent the mean values of C _b and C _r , respectively. If there is no color cast in the image, M _b and M _r are equal in the YCbCr space, that is:

M _b =M _r =128(1)

The proposed method uses the following formula to adjust the values of the two chrominance components C _b and C _r to complete the color cast correction, namely:

${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; 128</span>}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; 128</span>}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

In the formula, C _b1 and C _r1 represent the color cast correction results.

5 Chroma component saturation increased

The color of the image after color cast correction is relatively dull. Observing the histograms of C _b1 and C _r1 , it is found that most of the values are concentrated around the median value of 128, and the distribution range is narrow. Therefore, the method proposed in the present invention uses the Sigmoid function to stretch the chroma component to obtain enhanced chroma saturation. The Sigmoid function used is defined as:

$\mathrm{<span\; class="notranslate">\; g</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 256</span>}}{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\frac{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 128</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; 16</span>}}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

In the formula, the input f represents C _b1 and C _r1 , and the output g represents the saturation enhancement result of the two chroma components, represented by C _b2 and C _r2

6 Restoring the image

Combine the processed luminance component _Y2 and chrominance components _Cb2 and _Cr2 , and re-transform back to the RGB color space for easy display or storage.

In order to verify the effectiveness of the proposed method, 50 underwater images are used to construct a test sample set. Figure 2 lists some examples of processing results, the left side is the color degraded underwater image, and the right side is the enhanced result image. Compared with the original image, it is not difficult to find that the contrast and color information of the processed image have been significantly improved, which verifies the effectiveness of our method.

Figure 3 shows the comparison of the processing results of the proposed method and PhotoshopCS12's "automatic tone + automatic contrast adjustment" scheme. Among them, the first row is the original input image, the second row is the Photoshop processing result, and the third row is the processing result of the proposed method. Comparing the processing results of the two methods, it is not difficult to find that the contrast and color saturation of the processing results of the proposed method are better than those of the Photoshop method.

Claims (2)

1. A method for enhancing underwater images based on brightness and chromaticity separation, comprising the following steps: 1) Separate the luminance and chrominance components of the input underwater color image Convert the input underwater color image from the RGB color space to the YCbCr color space to separate the luminance and chrominance components. In the YC _b C _r color space, Y represents the luminance component, and C _b and C _r represent the blue and red biases respectively. displacement. 2) Luminance component denoising based on median filter Select the median filter to filter the brightness component Y, and the processing result is represented by _Y1 ; 3) Brightness component contrast enhancement based on adaptive histogram equalization The contrast-limited histogram equalization method (CLAHE) is selected to improve the contrast of the denoised brightness component _Y1 , and the processing result is represented by _Y2 ; 4) Chroma component color cast correction Use C _b and C _r to represent the two chromaticity component diagrams in the YCbCr space, use M _b and M _r to represent the mean values of C _b and C _r respectively, and use the following formula to adjust the values of the two chromatic components of C _b and C _r , Complete the color cast correction of the chroma component: ${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; 128</span>}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}}$ ${\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; 128</span>}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}}$ In the formula, use C _b1 and C _r1 to represent the color cast correction results; 5) Increase the saturation of the chroma component Use the Sigmoid function to stretch the chroma component that has undergone color cast correction to obtain enhanced chroma saturation. The Sigmoid function used is: $\mathrm{<span\; class="notranslate">\; g</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 256</span>}}{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\frac{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 128</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; 16</span>}}}}$ In the formula, when the input f is C _b1 , the output g represents the saturation enhancement result of the blue chroma component, represented by C _b2 ; when the input f is C _r1 , the output g represents the saturation enhancement result of the red chroma component, represented by C _r2 means; 6) Restoring the image Combine the luminance component Y ₂ and the chrominance components C _b2 and C _r2 processed in step 3), and re-transform back to the RGB color space. 2. The underwater image enhancement method according to claim 1, characterized in that, step 3 adopts "Rayleigh distribution".