Research On Target Detection Algorithm For Hyperspectral Imagery Based On Optical Correlation Recognition

This paper deals with target detection algorithm for hyperspectral imagery based on optical correlation recognition.The major research are as follows:Based on analysis of optical correlation recognition and JTC(joint transform correlation)-based hyperspectral target detection algorithm,a MFC(matched filtering correlation)-based algorithm is proposed in this paper;The discriminations of MFC-based spectral domain filter and JTC-based spectral domain filter are discussed here;A technique based on 2-dimension transformation for MFC and JTC algorithm is proposed; MFC-based and JTC-based algorithms are implemented in a HYDICE hyperspectral image.four mitrics are introduced to quantify the correlation output,and the results of different mitrics and different algorithms are discussed then.To overcome the influence of environmental noise on hyperspectral data,based on discussion of methods in derivative analysis of hyperspectral data,a technique based on derivative filter is introduced to MFC-based algorithm to enhance the objective feature and improve correlation intensity;Different salcs of mexico wavelet are selected for simultion,results show that the introduction of mexico wavelet can significantly improve detection performance of MFC-based algorithm.