摘要
侧扫声呐图像在获取的过程中由于其成像机理和复杂的海底环境会产生斑点噪声,其中由于不同底质类型形成的海底混响是造成这些斑点噪声的一个主要因素.首先利用5种典型的概率分布对海底混响的统计特性进行拟合得到了最优的拟合分布.再通过分析灰度直方图相关的特征参数,提出了基于两个特征参数对最优分布参数的估计.最终,通过多元回归分析实现了侧扫声呐图像噪声模型的建立和不同底质类型的图像分类.实验结果表明:伽马分布模型具有拟合特性准确和概率计算便捷的优势;根据模型参数与图像特征所得到的模型可以有效地模拟不同底质类型混响所导致的噪声,有利于底质的分类和噪声的消除.
Side scan sonar image contain speckle noises due to their characteristics of imaging mechanism and complicated seabed en vironment in the acquisition pprocess.Seabed reverberations of different types of substrates often exert different effects on lateral scan sonar imaging.Based on the analysis of the seabed reverberation statistical model, the optimal fitting distribution model is obtained under five typical probability distribution models and the evaluation based on two eigenvalues is proposed after analyzing the charac teristic parameters of gray histogram.Finally,the noise model of side scan sonar image and image classification of different sediment types wcrc achieved by multiple regrcssion model analyses. Experimental results show that Gamma distribution model enjoys the ad- vantages of accurate fitting characteristics and convenient calculation of probability. According to model parameters and image lea tures,the model can effectively sinmlate the noise caused by different sediment types of reverberation,and is conducive to the classifi cation of the sediment and the elimination of noise.
作者
张楷涵
袁飞
程恩
ZHANG Kaihan;YUAN Fei;CHENG En(Technology Ministry of Education,Xiamen University,Xiamen 361005,China)
出处
《厦门大学学报（自然科学版）》
CAS
CSCD
北大核心
2018年第3期390-395,共6页
Journal of Xiamen University：Natural Science
基金
国家自然科学基金(61471308,61571377)
关键词
海底混响
概率分布
声呐图像
底质分类
多元回归
seabed reverberation
probability distributions
sonar image
sediment classification
multiple regression
