Scholarly article on topic 'Underwater image quality enhancement of sea cucumbers based on improved histogram equalization and wavelet transform'

Underwater image quality enhancement of sea cucumbers based on improved histogram equalization and wavelet transform Academic research paper on "Medical engineering"

CC BY-NC-ND
0
0
Share paper
OECD Field of science
Keywords
{"Sea cucumber" / "Underwater image enhancement" / "Contrast improvement" / De-noising}

Abstract of research paper on Medical engineering, author of scientific article — Xi Qiao, Jianhua Bao, Hang Zhang, Lihua Zeng, Daoliang Li

Abstract Sea cucumbers usually live in an environment where lighting and visibility are generally not controllable, which cause the underwater image of sea cucumbers to be distorted, blurred, and severely attenuated. Therefore, the valuable information from such an image cannot be fully extracted for further processing. To solve the problems mentioned above and improve the quality of the underwater images of sea cucumbers, pre-processing of a sea cucumber image is attracting increasing interest. This paper presents a new method based on contrast limited adaptive histogram equalization and wavelet transform (CLAHE-WT) to enhance the sea cucumber image quality. CLAHE was used to process the underwater image for increasing contrast based on the Rayleigh distribution, and WT was used for de-noising based on a soft threshold. Qualitative analysis indicated that the proposed method exhibited better performance in enhancing the quality and retaining the image details. For quantitative analysis, the test with 120 underwater images showed that for the proposed method, the mean square error (MSE), peak signal to noise ratio (PSNR), and entropy were 49.2098, 13.3909, and 6.6815, respectively. The proposed method outperformed three established methods in enhancing the visual quality of sea cucumber underwater gray image.

Academic research paper on topic "Underwater image quality enhancement of sea cucumbers based on improved histogram equalization and wavelet transform"

Underwater image quality enhancement of sea cucumbers based on improved histogram equalization and wavelet transform

Xi Qiao, Jianhua Bao, Hang Zhang, Lihua Zeng, Daoliang Li

PII: DOI:

Reference:

Accepted Manuscript

To appear in:

Information Processing in Agriculture

S2214-3173(17)30069-0 http://dx.doi.org/10.1016/j.inpa.2017.06.001 INPA 89

Received Date: 5 May 2017

Revised Date: 19 May 2017

Accepted Date: 6 June 2017

Please cite this article as: X. Qiao, J. Bao, H. Zhang, L. Zeng, D. Li, Underwater image quality enhancement of sea cucumbers based on improved histogram equalization and wavelet transform, Information Processing in Agriculture (2017), doi: http://dx.doi.org/10.1016/jinpa.2017.06.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Underwater image quality enhancement of sea cucumbers based on improved histogram

equalization and wavelet transform

Xi Qiao 12, Jianhua Bao 12, Hang Zhang 12,3, Lihua Zeng 12,4, Daoliang Li 12*

1 Key Laboratory of Agricultural Information Acquisition Technology, Ministry of Agriculture, China

Agricultural University, Beijing, P.R. China

2 College of Information and Electrical Engineering, China Agricultural University, Beijing, P.R. China

ure, Chi

sity, Beiji

3 College of Computer and Information Engineering, Tianjin Agricultural University, Tianjin, P.R.

, Hebei Agricultu

4 College of Mathematics and Computer Science, Hebei Agricultural University, Baoding, P.R. China

6-10-627376 £

mail: qia(

Ha Li:

Corresponding author. Tel.: +86-10-62737679 Fax: +86-10-62737741 E-mail address: dliangl@cau.edu.cn

Postal address: P. O. Box 121, China Agricultural University, 17 Tsinghua East Road, Beijing 100083, PR China.

Xi Qiao, Email: qiao-xi@qq.com Jianhua Bao, Email: bjhxz@126.com

Hang zhang, Email: zhanghrz@126.com Lihua Zeng, Email: zenglh@cau.edu.cn

Abstract: Sea cucumbers usually live in an environment where lighting and visibility are generally not controllable, which cause the underwater image of sea cucumbers to be distorted, blurred, and severely attenuated. Therefore, the valuable information from such an image cannot be fully extracted for further processing. To solve the problems mentioned above and improve the

quality of the underwater images of sea cucumbers, pre-processing of a sea cucumber image is attracting increasing interest. This paper presents a new method based on contrast limited adaptive histogram equalization and wavelet transform (CLAHE-WT) to enhance the sea cucumber image quality. CLAHE was used to process the underwater image for increasing contrast based on the Rayleigh distribution, and WT was used for de-noising based on a soft threshold. Qualitative analysis indicated that the proposed method exhibited better performance in enhancing the quality and retaining the image details. For quantitative analysis, the test with 120 underwater images showed that for the proposed method, the mean square error (MSE), peak signal to noise ratio (PSNR), and entropy were 49.2098, 13.3909, and 6.6815, respectively.

The proposed method outperformed three established methods in enhancing the visual quality of sea cucumber underwater gray image.

Keywords: Sea cucumber; Underwater image enhancement; Contrast improvement; De-noising. 1. Introduction

Sea cucumbers (Stichopus japonicus), which belong to the class Holothuroidea, are marine

Middle East [1]. The aquaculture of the sea cucumber has grown rapidly over recent years in China, in response to increasing consumer demand. Indeed, sea cucumbers have become an important sector of the marine industry in northern China, with a culture area exceeding one million acres and a production value over one hundred and twenty million dollars. However, with the rapid growth and the use of non-standard practices and culture techniques, epidemic diseases of A. japonicas now pose increasing problems to the industry [2-4]. To understand the growth of sea cucumbers and prevent the emergence of large-scale diseases in the process of cultivating, non-contact observations of the sea cucumbers in

nderwater ima

water image

real-time are required. Currently, sea cucumber culture depends on frogman, such culture requires high labor intensity and poses a risk to the lives of the frogmen. Furthermore, the rapid growth of underwater robot technology over the past six decades [5-9], provides a means to solve these problems. The computer vision technique is a useful tool that enables an underwater robot to perceive sea cucumbers in the real world. However, the quality of the underwater images is affected by different factors, such as limited visibility range, low contrast, non-uniform lighting, blurring, bright artefacts, diminished color and noise [10, 11]. These factors result in low contrast, color attenuation and noise in the captured underwater sea cucumber images. A part of information from the images is also lost. Therefore, it is crucial for us to restore the contrast, color and lost information and remove the noise from sea cucumber images.

For the last several decades, the underwater image processing area has received considerable attention. Schettini and Corchs summarized some of the most recent methods that have been specifically developed for the underwater environment, and divided them into image restoration and image enhancement [11]. The image restoration methods base on the light spreading model in water [12], the energy attenuation model, and the scattering model[13] were used to enhance visibility, restore color and dehaze[16] of underwater image [17-19]. However, the models of the above references depended priori knowledge of the environment, such as scene depth, spatial location, optical length, illumination, temperature and conductivity. More sensors were required to match the vision system. In addition, the processing time of these models was far greater than that of the image enhancement method [11]. The image enhancement methods are normally simpler and faster than the image restoration techniques[20]. A number of research studies involving the application of this technique to enhance contrast[21] and remove noise [22] of underwater image have been reported. And the methods

[14, 15], a

on a p on a p

were widely used include histogram equalization (HE) [23-26], contrast limited adaptive histogram

equalization (CLAHE) [27, 28], and wavelet transform (WT) [29, 30]. In other field, some advanced methods were used to improve image quality too, such as: smoothing methods[31, 32] and artificial li

model[33].

However, the use of a single method makes it difficult to obtain perfect results in underwater image processing. In addition, the restoration models were complex, and the calculation time was lengthy. The

underwater 1:

aim of this paper is to develop an integrated approach that can improve the contrast of a gray image and remove the noise. We attempt to process the underwater sea cucumber image by contrast limited

adaptive histogram equalization and wavelet transform (CLAHE-WT). In our work, the sea cucumber adaptive histogram equalization and wavelet transform (CLAHE-WT). In our work, the sea cucumber

farm lies in the open air, the water depth is approximately 2 meters, and the brightness of underwater

he color o

_ comes ^ _ color ^ . _ , gra, do . ^ the factors of illumination and color. 2. Methods!

Figure 1 Framework of the proposed method Implementation

The framework of the proposed method to improve the contrast and remove blur and noise is

illustrated in Figure 1. First, the underwater color image was transformed into a gray image. Second, the image histogram is applied with contrast stretching in the gray model. Depending on the gray channel, the peak of the image histogram would be moved to the middle range of the gray level, and the image histogram would be stretched in both upper and lower directions. The stretching process is set to follow the Rayleigh distribution and is limited within a certain range. Third, the image is decomposed into an approximation image and detailed images of the horizontal, vertical and diagonal directions, in which the detail image of the diagonal direction is de-noised using a soft threshold. After completing these steps, the images are composed. An improved contrast and de-noised output image can then be produced at the final stage. All the referred functions are from MATLAB R2011b including the Image Processing Toolbox (The Math

Works Inc., USA). The values of the parameters were determined by trial and error. 2.1 Contrast limited adaptive histogram equalization

The original image was transformed into a gray image using f1 =rgb2gray(f), and the grayscale value was computed by taking the value of RGB, as described by the following equation: A = 0.299 x R + 0.587 x G + 0.114 x B (1)

t, to enhance a low-contrast underwater image, CLAHE was used to process the grayscale ige f1 by J = adapthisteq (Tiles, param1, val1, param2, val2). CLAHE differs from ordinary adaptive histogram equalization in its contrast limiting. CLAHE limits the amplification by clipping the histogram at a user-defined value called clip limit. The clipping level determines how much noise in the histogram should be smoothed and hence how much contrast should be enhanced [28]. In addition to histogram stretching, the pixel distribution is mapped to follow the Rayleigh distribution, which produces a bell-shaped distribution wherein most of the pixels are

Next, t

image image

concentrated in the middle range of the gray level. The upper and lower sides of the histogram intensity level have the fewest pixels. The probability distribution function (PDF) and cumulative distribution function (CDF) of the Rayleigh distribution are given by [27]

PDFRayle ig h = (^)e(~- 2 /2 « 2 ) for x>0,a>0 (2)

CDFRayle ig h = f0x^e(~ - 2 /2 « 2 )dx = 1 - e(" ^ 2 /2 « 2) for i£ (0 ,co) . (3)

The parameters are shown in Table 1. In the proposed method, these values of the parameters are

used for all underwater images. 2.2 Wavelet transform

For removing blur and de-noising, the enhancement image was processed using the WT

technique. The enhancement image was decomposed to approximate image, horizontal edge image technique. The enhancement image was decomposed to approximate image, horizontal edge image

detail, vertical edge image detail and diagonal edge image detail using orthogonal wavelet base

F(x) G L2

'coif2'. Supposing that 'coif2' % (x)eL2(R) (in which L2(R) represents the mean square cumulative one-dimension function's Hilbert space) satisfies the permission condition:

(x) dx = 0 (4)

The function M (x) is called the wavelet function. If we compress and expand the function M (x)

by using 5, which is called the scale factor, then the function can be obtained as follows:

¥s(x)

=M9 (5)

Next, the definition of WT on the scale s and position x is as follows:

<oJ(x) = J(x) * %(x) = J(u) % (x - u)du (6)

Where '*' denotes convolution, and 5 takes consecutive values among the real domain. The image J (x) is decomposed into the approximation image and detailed images of the horizontal, vertical and diagonal directions. Due to several uncontrollable reasons (e.g., the location of sea cucumber,

illumination, camera-to-subject distance, water turbidity) the image gray level intensity distribution varied between the images and the noise was produced. To remove the image noise, an adaptive threshold function was used, it can be written as

THR = aj 2log(M xN) (7)

Where a is the standard devistion of image, M and N are the width and height of the image, respectively. Then the wavelet transform coefficients ( are processed by soft threshold method. In mathematical form, it can be written as

sign ( ) • (|(| - THR), (> THR

( < THR

Where a> is the wavelet coefficient. Finally, the approximate image, unprocessed image detail and processed image detail were combined to form a new image using a function as:

G( x) = X£(X*, (x)

Where A is the constant. In Matlab, functions 4-9 can be represented by [c, s] = wavedec2(J, N, 'wname'); [thr, sorh, keepapp] = ddencmp('IN1', 'IN2', J);G = wdencmp('gbl', c, s, 'wname', N, THR, SORH, keepapp). The relevant parameters are shown in Table 1. Table 1 Parameter setting for CLAHE and WT

method Parameters

CLAHE (adapthisteq) Tiles (the image name): f1

param1 (name): Clip-limit val1 (the range [0 1], default: 0.01): 0.02 param2 (name): Distribution val2 (histogram shape): Rayleigh

WT * N (The layer number of wavelet decomposition):2 wavedec2 'wname' (name of wavelet base ): coif2 ddencmp IN1 ('den' for de-noising or 'cmp' for compression): den, ddencmp IN2 ('wv' for wavelet or 'wp' for wavelet packet): wv wdencmp SORH (threshold method): s wdencmp 'wname' (name of wavelet base ): coif2

2.3 Performance analysis

The Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR) are two error metrics that are frequently used to compare the quality of the processed images [27, 28]. As a quantitative

evaluation, the entropy is interesting to compare the images [27]. The MSE represents the

cumulative squared error between the improved image and the original image. MSE is written as follows:

MSE = ¿{S^S^1 [/ (i, j) - g(i, j)] 2 } (10)

Where M and N are the width and height of the image, respectively, i and j are locations pixel of the image, / { i ,j) is the gray scale value of the original image, and g { i ,j) is the gray scale value of the improved image. The PSNR represents the ratio of the maximum possible signal and the noise. PSNR is written as follows:

P S N R = 1 01 0 g i (11)

The entropy represents the abundance of image information that measures the image information content. Entropy is calculated as the summation of the products of the probability of outcome multiplied by the log of the inverse of the outcome probability [27]. Entropy, H is written as follows:

H = -£ p * log2 {p,) (12)

Where is the probability density function of pixel in the image. The best resultant image is

indicated by a low MSE, high PSNR and high entropy values.

3. Results and discussion

The materials used in our experiment are underwater images of sea cucumbers. These images were

captured by the C-Watch in a sea cucumber farm at Haiyang Qiandao Lake in Shandong province,

China. The C-Watch is a remotely operated underwater vehicle (ROV) for shallow water operations.

The C-Watch contains a battery pack to power the propulsion units and electronics, such as an attitude

and heading reference unit, GPS sensor, pressure sensor and camera (Canon PowerShot G12, Japan).

The camera is downward looking and focused on the sea cucumbers that live on the bottom of sandy

sediments in shallow pool. The image resolution is 1280*720. Some sample images are shown in Figure 2. The sea cucumber objects in a captured image are gray objects, the background of the image is almost in gray, and the underwater image is blurred and contains high noises.

In this paper, 120 underwater images of sea cucumbers were used to validate the effect of the proposed method. Out of the 120 images, three different levels of pollution of sea cucumber images were used as samples for both qualitative and quantitative evaluation. The image contained high noises and contrast was low. The proposed method aims to improve the contrast of the images and remove the noises. In the discussion, HE and CLAHE are compared with the proposed method. In addition, state-of-the-art methods were used to enhance image quality, including the underwater gray images. The results produced by all of the methods are sho in Figures. 3-5.

Figure 2 Sea cucumber image samples 3.1 Qualitative results and discussion

In the results, the main focus of the visual observation is the contrast and noise of the images. The

observation includes the capability of those methods to remove the noises and to reduce the under- and over-e

-enhanced areas, wherein the output images become too dark or too bright, respectively. The effect of de-noising is obviously shown on the histogram of the image (the value of gray level 255).

The image I of the sea cucumbers is the lowest contrast of the underwater images shown in Figure 2. The image areas were lowlighted by the HE method, and the output image is either excessively dark or bright, which results in loss of image details (Figure 3 (b)). The histogram of HE shows that the distribution of gray level are widened and thin (Figure 3 (f)). The image produced by CLAHE is better

in terms of contrast and reveals more of the sea cucumber details from background (Figure 3 (c)). However, the value of gray level 255 in CLAHE histogram is very high (Figure 3 (g)), which proves that some noises existed in the image. Of the state-of-the-art methods examined here, CLAHE-WT produced the better contrast enhancement because the sea cucumber is clearly seen in the image (Figure 3 (d)). Moreover, the noise in the image is reduced because the value of gray level 255 in

CLAHE histogram is mostly zero (Figure 3 (h)). Above observations are shown in Figure 3.

Figure 3 Sea cucumbers image I: (a) gray image; the rest are the images processed using the methods of (b) HE, (c) CLAHE, and (d) CLAHE-WT. Histograms of sea cucumbers images a, b, c, and d are (e) gray image, (f) HE, (g) CLAHE, and (h) CLAHE-WT, respectively.

iy imagi

.. p. e II (Fig

ge II (Figure 2) was processed by HE, as shown in Figure 4. The image is oversaturated, the areas

me too bright and too dark, and some image details are disappeared, e.g., the edge of sea cucumber upper part is covered by background (Figure 4 (b)). CLAHE increased the image contrast, wherein the

sea cucumbers are better differentiated (Figure 4 (c)). However, the value of gray level 255 in CLAHE histogram is very high (Figure 4 (g)), and the noise is distributed in the image. The proposed method successfully increased the contrast and removed the noise because the sea cucumbers are the most clear, the edge of sea cucumber upper part is distinguished from background, no areas are observed that are

excessively dark or bright (Figure 4 (d)), and the value of gray level 255 in CLAHE histogram is

mostly zero (Figure 4 (h)).

(e) gray image (f) HE (g) CLAHE (h) CLAHE-WT

Figure 4 Sea cucumber image II: (a) gray image; the rest are the images processed using the methods of (b) HE, (c) CLAHE and (d) CLAHE-WT. Histograms of sea cucumbers images a, b, c, and d are (e) gray image, (f) HE, (g) CLAHE, and (h) CLAHE-WT, respectively.

(e) gray image (f)HE (g) CLAHE (h) CLAIIE-WT

Figure 5 Sea cucumber image III: (a) gray image; the rest are the images processed using the methods of (b) HE, (c) CLAHE and (d) CLAHE-WT. Histograms of sea cucumbers images a, b, c, and d are (e) gray image, (f) HE, (g) CLAHE, and (h) CLAHE-WT, respectively.

Image III of the sea cucumber is the clearest of the underwater sea cucumber images in Figure 2. The Image III of the sea cucumber is the clearest of the underwater sea cucumber images in Figure 2. The

result demonstrates that some areas in the image that are produced using the HE method, are over-enhanced. The contrast of the output image becomes excessively high, and some image details are lost (Figure 5 (b)). CLAHE could not significantly reduce the effect of noise because the high value of gray level 255 is retained in CLAHE histogram (Figure 5 (g)). However, the contrast in the resultant image is increased and the sea cucumber in the resultant image is clearly differentiated (Figure 5 (c)).

The proposed method is proven to be the best method than HE and CLAHE, because it can significantly increase the contrast and remove the noise. The sea cucumber in the image is also better distinguished from the background and the details of sea cucumbers are most clear (Figure 5 (d) (h)). The abovementioned observations are highlighted by the figures (Figure 5).

Based on visual observation, the proposed method generally provided better improvement of the image contrast than the other state-of-the-art methods. The sea cucumbers in the images are better distinguished from the background. Moreover, the effect of noise is significantly reduced. The output images of proposed method show the best visualization. 3.2 Quantitative result and discussion

Table 2 Quantitative results in terms of MSE, PSNR, and entropy values for the images in Figures 3-5. Image Method MSE PSNR Entropy

; is signitic

ind entropy

Note: T

Gray 3.3303

HE 118.7194 5.5532 2.9426

CLAHE 49.0397 13.2327 3.8259

CLAHE-WT 35.4256 16.0573 6.1775

Gray 3.6153

HE 119.7575 5.4776 3.1738

CLAHE 53.689 12.4459 4.2089

CLAHE-WT 41.2903 14.7267 6.5503

Gray 3.4266

HE 117.3105 5.6569 2.9574

CLAHE 49.2101 13.2026 3.9550

CLAHE-WT 38.0605 15.4342 6.4711

he values in bold typeface represent the best results obtained from the comparison. The comparative values of MSE, PSNR, and entropy for the images in Figures 3-5 are shown in

Table 2. The best results are represented by bold face values. As previously shown, the quantitative

performance of CLAHE-WT is distinguished among the other methods in terms of MSE, PSNR, and

entropy (Table 2). As stated in the table, HE produced the highest MSE and the lowest PSNR and

entropy values, the entropy value is even lower than the gray image, a number of image information is

lost i.e., it is the worst method. CLAHE produces the second highest MSE values and the second

lowest PSNR values, which are close to the bold face values, CLAHE consistently showed the better performances in keeping image information.

The proposed method produces the lowest MSE values and highest PSNR values and entropy values compared with other methods. The differences of MSE, PSNR, and entropy values between CLAHE-WT and the other state-of-the-art methods are big, as shown in Table 2. These values show that the proposed method can sufficiently reduce noise and increase the valuable information of the image. These qualitative and quantitative evaluations prove that the proposed method generally improved the quality of sea cucumber underwater images.

Table 3 Average values of entropy, MSE, PSNR, and entropy for 120 underwater images. Method MSE PSNR Entropy

Gray 4.1273

HE 117.6261 5.6484 3.6114

CLAHE 59.9465 11.5838 4.7587

CLAHE-WT 49.2098 13.3909 6.6815

Note: The values in bold typeface represent the best results obtained from the comparison.

The average values of the MSE, PSNR, and entropy for 120 tested underwater sea cucumber images are shown in Table 3. The MSE value of the HE method indicates that this method produced over-enhancement and noise in the resultant images. CLAHE has the second average value of entropy, which demonstrates that CLAHE is the better method in image information restoration than HE. However, the MSE and PSNR values of CLAHE indicate that noise was produced in the resultant image. The table also shows that the proposed method has the best average values for MSE, PSNR, and entropy, which validates that the contrast enhancement, de-noising and image information restoration were best obtained by using CLAHE-WT. However, the proposed method is fused two methods, the processed time is a disadvantage. 4. Conclusion and future work

In this paper, a simple and useful algorithm combining CLAHE and WT was proposed. In the

proposed method, the gray histogram distribution is shaped to follow the Rayleigh distribution, in which the gray histogram is stretched toward the lower and upper sides within the limits and the high value of gray level 255 is reduced. This technique significantly improves the contrast and reduces the noise of the sea cucumber underwater image. In addition, the proposed technique avoids the effect of

ntitative re

under- and over-enhanced areas in the image output. The qualitative and quantitative results demonstrated the superior performance of proposed method over the traditional HE and improved HE-based enhancement. This method could effectively improve the quality and visualization of the sea cucumber underwater gray images and lay the groundwork for further processing of the image. Authors' contributions

Xi Qiao collected the image data, designed algorithm for processing images, wrote the paper, and contributed to analysis the experimental data.

Jian Zou collected the image data, wrote the paper, and contributed to analysis the experimental data.

Jianhua Bao collected the image data. Hang Zhang designed algorithm for processing images.

nental dal e data, wrote the

algorithm for proce

Wenzh Daoliai Ac

nzhu Yang designed algorithm for processing images.

ng Li wrote the paper, and contributed to analysis the experimental data. cknowledgment

The authors thank native English speaking expert Dr. S. G. Hassan from China Agricultural University for reviewing the language fluency of our paper, and C. Alan from the editing team of American Journal Experts for polishing our paper. The work in this paper was supported by the International Science & Technology Cooperation Program of China (2015DFA00090) and Special Fund for Agro-scientific Research in the Public Interest (201203017).

Reference

[1] Bordbar S, Anwar F, Saari N. High-Value Components and Bioactives from Sea Cucumbers for Functional Foods—A Review. Marine Drugs, 2011,9(12):1761~1805

[4 [5 [6 [7

[9 [10 [11

[12 [13 [14 [15 [16 [17

[18 [19 [20

of an Autonomous Underwater Vehicle 3gy Society Journal, 2016,50(6):58~68 lignment approach for underwater optical mapping.

Hongzhan L, Fengrong Z, Xiuqin S, et al. Identification of the pathogens associated with skin ulceration and peristome tumescence in cultured sea cucumbers Apostichopus japonicus (Selenka). Journal of Invertebrate Pathology, 2010,105(3):236~242

Dang H, Zhang X, Song L, et al. Molecular characterizations of oxytetracycline resistant bacteria and their resistance genes from mariculture waters of China. Marine Pollution Bulletin, 2006,52(11): 1494~1503

Poot-Salazar A, Hernández-Flores Á, Ardisson P. Use of the SLW index to calculate growth function in the sea cucumber Isostichopus badionotus. Scientific Reports, 2015,4(1):5151~5157 Bogue R. Underwater robots: a review of technologies and applications. INDUSTRIAL ROBOT-AN INTERNATIONAL JOURNAL, 2015,42(3):186~191

Mallet D, Pelletier D. Underwater video techniques for observing coastal marine biodiversity: A review of sixty years of publications (1952 - 2012). Fisheries Research, 2014,154:44~62 Karimanzira D, Jacobi M, Pfuetzenreuter T, et al. First testing of an AUV mission planning and guidance system for water quality monitoring and fish behavior observation in net cage fish farming. Information Processing in Agriculture, 2014,1(2): 131~140

Coleman D F, Newman J B, Ballard R D. Design and Implementation of Advanced Underwater Imaging Systems for Deep Sea Marine Archaeological Surveys.2000: PROVIDENCE, RI. 661~665

Li D, Zhang T, Yang C. Terminal Underw Using One Camera and One Light. Mari Elibol A, Garcia R, Gracias N. A new glol Ocean Engineering, 2011,38(10):1207~1219

Schettini R, Corchs S. Underwater Image Processing: State of the Art of Restoration and Image Enhancement Methods. EURASIP Journal on Advances in Signal Processing, 2010,2010(746052):1~15

Lee D, Kim G, Kim D, et al. Vision-based object detection and tracking for autonomous navigation of underwater robots. Ocean Engineering, 2012,48:59~68

Chen Y, Xia M, Li W, et al. Comparison of point spread models for underwater image restoration. Optik - International Journal for Light and Electron Optics, 2012,123(9):753~757 Mohd Naim M, Mat Isa N A. Pixel distribution shifting color correction for digital color images. Applied Soft Computing, 2012,12(9):2948~2962

Panetta K, Gao C, Agaian S. Human-Visual-System-Inspired Underwater Image Quality Measures. IEEE Journal of Oceanic Engineering, 2016,41(3):541~551

Carlevaris-Bianco N, Mohan A, Eustice R M. Initial Results in Underwater Single Image Dehazing.2010: Seattle, WA, USA. 1~8

Chen Z, Wang H, Shen J, et al. Region-specialized underwater image restoration in inhomogeneous optical environments. Optik - International Journal for Light and Electron Optics, 2014,125(9):2090~2098

Hou W, Woods S, Jarosz E, et al. Optical turbulence on underwater image degradation in natural environments. Applied Optics, 2012,51(14):2678~2686

Chiang J Y, Ying-Ching C. Underwater Image Enhancement by Wavelength Compensation and Dehazing. IEEE Transactions on Image Processing, 2012,21(4):1756~1769 BANERJEE J, RAY R, VADALI S R K, et al. Real-time underwater image enhancement: An improved approach for imaging with AUV-150. Sadhana, 2016,41(2):225~238

ia ■

ility.2013:

[21] Çelebi A T, Ertürk S. Visual enhancement of underwater images using Empirical Mode Decomposition. Expert Systems with Applications, 2012,39(1):800~805

[22] Sheng M W, Pang Y J, Huang H, et al. A Novel Adaptive Underwater Image Biorthogonal Basic Wavelet Transform De-Noising Approach. Applied Mechanics and Materials. 2013,411-414:1335—1340

[23] Wen Y, Tian W, Zheng B, et al. New approach for underwater imaging and processing.201 Baltimore

[24] Hu B, Zheng B, Yang Y, et al. Underwater image color correct in extremely poor visibi San Diego. 1—4

[25] Shamsuddin N B, Ahmad W, Baharudin B B, et al. Image Enhancement of Underwater Habitat Using Color Correction Based on Histogram//Zaman H B, Robinson P, Petrou M, et al. Lecture Notes in Computer Science. Berlin: Springer-Verlag Berlin, 2011:289—299

[26] Singhai J, Rawat P. Image enhancement method for underwater, ground and satellite images using brightness preserving histogram equalization with maximum entropy.2007: Sivakasi. 507—512

[27] Abdul Ghani A S, Mat Isa N A. Underwater image quality enhancement through integrated color model with Rayleigh distribution. Applied Soft Computing, 2015,27:219—230

[28] Hitam M S, Yussof W N J H, Awalludin E A, et al. Mixture Contrast Limited Adaptive Histogram Equalization for Underwater Image Enhancement.2013: Sousse. 1—5

[29] Li Y, Lu H, Zhang L, et al. Real-Time Visualization System for Deep-Sea Surveying. Mathematical Problems in Engineering, 2014,2014:1—10

[30] Fei L, Wang Y. The research of underwater image de-noising method based on adaptive wavelet transform.2014: Changsha. 2521—2525

[31] de Araujo A F, Constantinou C E, Tavares J M R S. Smoothing of ultrasound images using a new selective average filter. Expert Systems with Applications, 2016,60:96—106

[32] Gulo C A S J, de Arruda H F, de Araujo A F, et al. Efficient parallelization on GPU of an image smoothing method based on a variational model. Journal of Real-Time Image Processing, 2016

[33] de Araujo A F, Constantinou C E, Tavares J M R S. New artificial life model for image enhancement. Expert Systems with Applications, 2014,41(13):5892—5906

Distinguishing the image noise according to soft threshold.