摘要
鱼种的快速识别是渔业资源评估乃至海洋生态系统监测重要组成部分。声学方法是主流识别方法中的重要组成部分,目前常用的声学识别方法主要基于鱼类的回波信号。传统的回波包络或能量特征很难全面的表述鱼体回波信号信息,因此鱼类识别效果一般。本文提出一种基于Walsh变换的鱼类回波识别方法。试验获取鲫鱼、嘎鱼、武昌鱼的回波信号,处理过程中分别提取三种鱼类回波包络信号的Walsh谱作为识别特征量,并利用BP神经网络分类器对其进行了分类。结果表明利用回波的Walsh谱可以成功识别不同形状的鱼类,其中对武昌鱼的识别正确率达90%以上。
The fast and efficient fish identification is the composition of fishery survey and marine ecosystem monitoring. The identification method based on active acoustics is the most important one. The fish identifi- cation is limited, due to the traditional echo envelop and energy way cannot describe the fish backscattering properly. A method of fish identification based on Walsh transform is proposed in this paper. Firstly, an ex situ experiment has been performed with three kinds of fish: Crucian carp (Carassius auratus), Yellow-headed catfish (Pelteobagrus fulvidraco) and Bluntnose black bream (Megalobrama amblycephale). The backscatter- ing signals of these fishes are obtained to verify this method. Then, the Walsh spectrum of backscattering is extracted as the indicator to describe these three kinds of fish species. Finally, three kinds of fish are success- fully identified by using a BP neural network. The result shows that it's possible to identify fish with different shape using Walsh transform.
出处
《应用声学》
CSCD
北大核心
2015年第5期465-470,共6页
Journal of Applied Acoustics
基金
山东省科技发展计划项目(2013GHY11517)