Walsh变换特征用于回声信号的识别

Recognition of echo signal using Walsh transformation

随着水下目标降噪技术的进展,利用回声信号来探测和识别水下目标的越来越重要,水下目标识别是水声领域中存在的难题之一.以大量的水下目标回声信号资料为依托,借助Walsh变换的分析法,研究了脉冲声的回声识别问题,给出了回声信号的时频特性及其相应的特征向量,提取了43维的Walsh谱特征.结合ART神经网络,对一定范围对真假两类目标进行分类,可取得80%以上的正确识别率.该方法具有较好的实用性,是一条解决实际回声信号识别的有效途径.

With decreased noise levels for underwater targets, the detection and recognition of underwater target becomes more and more important and yet underwater target recognition is one of the most difficult problems in processing underwater acoustic signals. In the paper, recognition problems of a pulse echo signal is investigated and time-frequency characters as well as character vector are given. The 43 dimensional Walsh spectrum features from different kinds of targets are extracted via Walsh transformation analysis based on echo signal data. By applying this in echo signal combined with ART networks, the correct recognition ratio can arrive 80% to true and false targets. This method possesses better practicability and is an effective way to solve real recognition problems of echo signal.