作者：郭雨青1, 曾庆军2, 夏楠2, 孙啸天2, 许赫威2
作者单位：1. 江苏科技大学计算机学院，江苏 镇江 212028;
2. 江苏科技大学电子信息学院，江苏 镇江 212028
关键词：图像增强;目标识别;机器人视觉;YOLOv4;AUV
摘要：
为满足研发的水下自主机器人对水下环境目标识别的需求，针对退化的水下图像无法进行有效的目标检测的问题，提出一种基于水下光衰减先验（ULAP）的场景深度模型与对比度受限直方图均衡化（CLAHE）算法结合的图像增强新方法。该方法基于水下成像数学模型，构建深度图与绿蓝光的最大强度差和红光的线性关系，估计并推断出相对深度图，结合实际的深度场景推断各通道的传输图，获得未退化图像，并采用CLAHE算法来提高其对比度。通过YOLOv4目标检测网络对6种算法增强后的水下图像数据集进行训练与测试，实验表明，该方法可以有效提升各类水下图像清晰度和色彩增强，并且提高水下图像目标识别任务的准确率，为进一步开展水下自主机器人目标识别应用奠定基础。
Research on target recognition of autonomous underwater vehicle based on image enhancement
GUO Yuqing1, ZENG Qingjun2, XIA Nan2, SUN Xiaotian2, XU Hewei2
1. School of Computer Science, Jiangsu University of Science and Technology, Zhenjiang 212028, China;
2. School of Telecommunications, Jiangsu University of Science and Technology, Zhenjiang 212028, China
Abstract: In order to meet the requirements of underwater environment target recognition for underwater autonomous robot, a new image enhancement method based on scene depth model of underwater light attenuation prior (ULAP) and contrast constrained histogram equalization (CLAHE) algorithm is proposed to solve the problem that the degraded underwater image cannot be detected effectively. Based on the mathematical model of underwater imaging, this method constructs the linear relationship between the maximum intensity difference between the depth map and the green blue light and the red light, estimates and infers the relative depth map, infers the transmission map of each channel combined with the actual depth scene, obtains the non degraded image, and uses the CLAHE algorithm to improve its contrast. The enhanced underwater image data sets of six algorithms are trained and tested by YOLOv4 target detection network. Experiments show that this method can effectively improve the clarity and color enhancement of all kinds of underwater images, and improve the accuracy of underwater image target recognition tasks, which lays the foundation for further research and development of underwater robot applications.
Keywords: image enhancement;target recognition;robot vision;YOLOv4;AUV
2021, 47(11):47-52 收稿日期: 2021-07-01;收到修改稿日期: 2021-08-19
基金项目: 国家自然科学基金项目(11574120)；江苏省产业前瞻与共性关键技术项目(BE2018103)；江苏省研究生实践创新计划(SJCX21_1744)
作者简介: 郭雨青（1997-），女，江苏淮安市人，硕士研究生，专业方向为水下机器人视觉
参考文献
[1] ZHANG X, BIAN X Q, YAN Z P. Underwater docking of auv with the dock and virtual simulation[J]. Advanced Materials Research, 2011, 159: 371-376
[2] 郑荣, 宋涛, 孙庆刚, 等. 自主式水下机器人水下对接技术综述[J]. 中国舰船研究, 2018, 13(6): 43-49+65
[3] 李硕, 刘健, 徐会希, 等. 我国深海自主水下机器人的研究现状[J]. 中国科学:信息科学, 2018, 48(9): 1152-1164
[4] 羊云石, 顾海东. AUV 水下对接技术发展现状[J]. 声学与电子工程, 2013(2): 43-46
[5] 杨雪, 马培立, 施子凡, 等. 一种有缆视觉自主型机器鱼系统设计[J]. 中国测试, 2018, 44(12): 62-68
[6] 刘桂雄, 张瑜, 蔡柳依婷. 机器视觉检测图像拼接融合技术研究进展[J]. 中国测试, 2020, 46(1): 1-6
[7] HE K M, SUN J, TAN X O. Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2011, 33(12): 2341-2353
[8] 杨淼, 王海文, 胡珂, 等. 一种基于色彩补偿的水下图像综合增强算法[J]. 图学学报, 2021, 42(1): 59-64
[9] 王昕, 赵飞, 蒋佐富, 等. 迁移学习和卷积神经网络电力设备图像识别方法[J]. 中国测试, 2020, 46(5): 108-113
[10] 李爱莲, 刘浩楠, 郭志斌, 等. 改进ResNet101网络下渣出钢状态识别研究[J]. 中国测试, 2020, 46(11): 116-119+125
[11] YU X L, QU Y Y, HONG M. Underwater-GAN: underwater image restoration via conditional generative adversarial network[C]//2018 International Conference on Pattern Recognition. Cham: Springer International Publishing, 2018: 66-75.
[12] LI C Y, ANWAR S, PORIKLI F. Underwater scene prior inspired deep underwater image and video enhancement[J]. Pattern Recognition, 2020, 98: 107038
[13] YANG M, HU K, DU Y X, et al. Underwater image enhancement based on conditional generative adversarial network[J]. Signal Processing:Image Communication, 2020, 81: 115723
[14] ZHAO X, JIN T, QU S. Deriving inherent optical properties from background color and underwater image enhancement[J]. Ocean Engineering, 2015, 94: 163-172
[15] 李炼, 李维嘉, 吴耀中. 基于红色暗通道先验理论与CLAHE算法的水下图像增强算法[J]. 中国舰船研究, 2019, 14(S1): 175-182
[16] LI C Y, GUO C L, REN W Q, et al. An underwater image enhancement benchmark dataset and beyond.[J]. IEEE Transactions on Image Processing: A Publication of the IEEE Signal Processing Society, 2020, 29: 4376-4389
[17] LIU R, FAN X, ZHU M, et al. Real-world underwater enhancement: challenges, benchmarks, and solutions under natural light[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2020, 30(12): 4861-4875
[18] 李菊霞, 李艳文, 牛帆, 等. 基于YOLOv4的猪只饮食行为检测方法[J]. 农业机械学报, 2021, 52(3): 251-256
[19] DREWS J P, NASCIMENTO E, MORAES F, et al. Transmission estimation in underwater single images[C]// IEEE International Conference on Computer Vision Workshops. IEEE, 2013.
[20] 黄冬梅, 王龑, 宋巍, 等. 不同颜色模型下自适应直方图拉伸的水下图像增强[J]. 中国图象图形学报, 2018, 23(5): 640-651
[21] PENG P,COSMAN C. Underwater image restoration based on image blurriness and light absorption[J]. IEEE Trans Image Process, 2017, 26(4): 1579-1594
[22] LI C, QUO J, PANG Y, et al. Single underwater image restoration by blue-green channels dehazing and red channel correction[C]// 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2016.
[23] GHANI N A,ISA M. Underwater image quality enhancement through composition of dual-intensity images and rayleigh-stretching[J]. SpringerPlus, 2014, 3: 757
