BM3D与PCNN结合的海底管道侧扫声呐图像处理方法  

Method on Processing Side-Scan Sonar Image of Submarine Pipeline Combined BM3D and PCNN

作　　者：陈林 白兴兰 胡轲 CHEN Lin;BAI Xinglan;HU Ke(School of Naval Architecture and Maritime of Zhejiang Ocean University,Key Laboratory of Offshore Engineering Technology of Zhejiang Province,Zhoushan 316022,China)

机构地区：[1]浙江海洋大学船舶与海运学院,浙江省近海海洋工程技术重点实验室,浙江舟山316022

出　　处：《浙江海洋大学学报（自然科学版）》2023年第2期180-188,共9页Journal of Zhejiang Ocean University：Natural Science

基　　金：国家自然科学基金(52171279);舟山市科技计划项目(2021C21002)。

摘　　要：海底管道作为油气输送的重要通道,波流冲刷作用下易发生悬跨现象,易发生疲劳破坏。侧扫声呐利用回波强度成像,是对海底管道状态检测的重要设备。复杂的海洋环境使得侧扫声呐图像不可避免地产生混合噪声而影响图像质量,因此如何有效去除噪声对侧扫声呐图像的判别与分析非常重要。针对图像的去噪问题,运用三维块匹配算法自编程序对实测的海底管道侧扫声呐图像进行去噪处理,将不同去噪方法处理结果从主观评价和客观评价指标上比较,结果表明三维块匹配算法要明显优于其他去噪方法。同时针对海底管道侧扫声呐图像分割困难的问题,提出了一种基于三维块匹配算法与脉冲耦合神经网络相结合的图像分割方法,对实测图像进行分割处理,并从图像处理效果的主观和客观评价指标进行分析,结果表明该方法较传统脉冲耦合神经网络图像分割方法有更好的分割效果,图像目标分割更加准确。Submarine pipeline is an important channel for offshore oil&gas transportation,and is prone to be suspended and fatigue damage will occur under the combined action of wave and current.The condition of the submarine pipeline is usually obtained by the side-scan sonar using echo intensity imaging.Nevertheless,the image collected by side-scan sonar will carry mixed noise and its quality is not good under complex environment.It is important to effectively remove noise for the identification and analysis of side-scan sonar image.Aiming at the mixed complex noise problem of the submarine pipeline side scan sonar image,the 3D block matching algorithm is used for denoise the sonar image,and the results of each de-noising method are compared by subjective and objective evaluation indexes.It shows that the 3D block matching algorithm is obviously better than other de-noising methods.In order to solve the segmentation problem of side scan sonar image,3D block matching combined with pulse coupled neural network is proposed to segment submarine pipeline side scan sonar image.The results of segmenting the actual image show that the proposed approach is feasible and more accurate than traditional pulse coupled neural network through objective and subjective evaluations.

关 键 词：侧扫声呐图像去噪 三维块匹配算法 侧扫声呐图像分割 海底管道 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]