The Research Of Least-Square Algorithm For Deconvolution In Two Channels

The research of deconvolution algorithm is a hot point in digital signal processing field. It has been widely applied in science research and engineering practice areas, such as seism signal processing, communication, radar signal processing and medical signal processing. And it is an important tool for signal recover and model recognition.Deconvolution is the inverse algorithm of convolution. In mathematics, we call the process that known the convolution and one serial to calculate the other serial deconvolution. There have been several different deconvolution algorithms in local and foreign research papers till now. In general, there are two kinds of basic method to calculate deconvolution. One is directly using time-region convolution equations to calculate, the other is frequency-domain method. The more important deconvolution algorithms among these are time-domain circle convolution method, inverse filter, Z transform, DFT transform method, alternative method, least-square method and so on.In application process, for the known serial is discrete limited signal, and its value may be zero or fairly small in magnitude, the result of deconvolution we calculated will appear fairly great or infinite. At this moment, it appears "ill-pose" problem. To resolve the problem, many researchers presented some good solutions.We described the convolution theory as matting in our paper, presenting the concept of deconvolution. We introduce the research background and application field of deconvolution, then concluding several typical deconvolution methods, especially the least-square deconvolution algorithm.