摘要
针对采用自主式水下机器人(AUV)巡检和拍摄输水隧洞中的裂缝时水下环境会极大地限制光视觉图像的可视范围和分辨率,单幅水下图像获得的视角范围有限的问题,提出一种基于加速稳健特征(SURF)算法的AUV水下图像拼接方法。在图像预处理阶段对水下图像进行去畸变和限制对比度自适应直方图均衡化处理,用于解决水下图像存在的畸变、对比度低和噪声严重等问题。将SURF特征点检测算法应用到水下图像配准中,并与RANSAC算法相结合,对特征点进行精确匹配。应用线性渐变融合算法实现水下图像融合,有效去除拼接缝隙,完成水下图像拼接,最终通过水池和外场试验拼接图像验证所提方法的正确性。
Underwater environment greatly limits the visual range and resolution of optical visual images,and the view range obtained from a single underwater image is limited when AUV is applied to check regularly and to take pictures of cracks in the outlet tunnel.Aiming to solve these problems,the AUV underwater image mosaic method based on SURF algorithm is proposed.The distortion of the underwater image is eliminated and contrast adaptive histogram equalization limited in the pro-processing stage in order to solve the problem of distortion,low contrast and serious noise in underwater image.The feature point detection algorithm based on SURF is applied to underwater image registration,and the feature points are precisely matched with the RANSAC algorithm.The linear gradient fusion algorithm is applied to achieve underwater image fusion,which remove splicing gap effectively and complete underwater image mosaic.Finally,the correctness of the proposed method is verified through the mosaic image during the tank test and the field test.
作者
陈舟
唐松奇
石磊
孙玉山
盛明伟
CHEN Zhou;TANG Songqi;SHI Lei;SUN Yushan;SHENG Mingwei(Zhejiang Design Institute of Water Conservancy&Hydro-electric Power,Hangzhou 310000,China;Science and Technology on Underwater Vehicle Laboratory,Harbin Engineering University,Harbin 150001,China)
出处
《船舶工程》
CSCD
北大核心
2019年第3期115-121,共7页
Ship Engineering
基金
国家自然科学基金(51609050
61633009
51509057)
中央高校基本科研业务费(HEUCFM170105)
