近岸4种鱼的声散射特征提取及融合实验研究*

为了有效地提取表征鱼类间差异的声散射特征参数，该文通过绳系法实验研究了近岸4种经济鱼类的声散射信号特征提取及融合方法。首先，通过自研双频鱼探仪采集花鲈、许氏平鲉、黑鲷和斑石鲷的个体鱼声散射信号；然后，分别测定200 kHz和450 kHz换能器下鱼体的目标强度，同时提取鱼声散射信号的时频域统计特征；最后，将降维后的时频特征与频差特征融合组成新的特征向量。该文通过实验验证了该方法的有效性，基于组合特征的支持向量机识别准确率达93%。结果表明，鱼的频率响应特性和鱼声散射信号的时频域统计特征能一定程度上反映鱼的固有属性，有效地增加判别依据能显著提高以上4种鱼类的识别准确率。

Experimental study on acoustic scattering feature extraction and fusion of four inshore fishes

In order to effectively extract acoustic scattering characteristic parameters that can characterize the differences between fish species, this paper studied the acoustic scattering signal feature extraction and fusion method of four economic fish species near shore by tether experiment. Firstly, the acoustic scattering signal of individual fish (Lateolabrax maculatus, Sebastes schlegelii, Acanthopagrus schlegelii, Oplegnathus punctatus) were collected by self-developed dual-frequency fish detector. Then, the target strength of fish under 200 kHz and 450 kHz transducers was measured respectively. Meanwhile, the time and frequency domain statistical characteristics of fish acoustic scattering signals were extracted. Finally, a new feature vector is composed of the frequency difference response features and the time-frequency domain statistical features after dimensionality reduction. The effectiveness of the proposed method is verified by experiments, and the recognition accuracy of the support vector machine classifier based on combinatorial features reaches 93%. The results show that, the frequency response characteristics of fish and the time and frequency domain statistical characteristics of fish acoustic scattering signal can reflect the inherent attributes of fish to a certain extent, effective addition of discrimination basis can significantly improve the identification accuracy of the above four species of fish.