Study And Application Of Wavelet Theory In Optical Correlation Detection

Optical correlation detection realizes correlation operation optically, and then discriminates target detected from cluttered scene. It primarily has two kinds of systems, matched filter and joint transform correlator. This paper used joint transform correlator as detection system. After inputting target image and reference template into joint transform correlator simultaneously, correlation result emerged in form of two points, which were called correlation points or correlation peaks. Existence of correlation peaks indicates successful recognition of target, and the relative position between peaks is consistent with that between target and template. It could conclude that quality of correlation peaks right influences the detection result of target.Enhancing intensity of correlation peaks obviously makes for improving the power of detection for target with joint transform correlator. How to increase intensity is always the issue which researchers interested in. Because joint transform correlator has developed from purely optical system to hybrid optoelectronic system, enhancement method of correlation peaks is gradually closely related to digital image processing technology. Among several way of edge extraction, wavelet is applied to preprocess target joint image for its characteristic of multiresolution analysis. Joint image is transformed by spline wavelet function firstly, then edge at j=1 is extracted through the method of modulus maximum. Simulation and optical experiment results show that the remarkable edge realizes detection of target in cluttered scene, infrared target, small target and low contrast target. However, edge at j=1 did not exploit fully the multi-scale property of wavelet. Therefore, edge extraction based on wavelet multi-scale product is presented. The edge contains product relation of wavelet coefficients between adjacent scales. It considers detail and contour of image at different scales. Simulation and optical experiment results show that edge extracted by wavelet multi-scale product could greatly improve energy of correlation peaks, and it provides an ideal tool for target recognition.The wavelet processing methods mentioned above just researched in terms of scalar wavelet, a part of content in powerful system of wavelet theory. Based on scalar wavelet, multiwavelet and hyper-wavelet are regarded as subjects investigated for researching applications of wavelet theory in optical correlation detection.Multiwavelet, extension of scalar wavelet, expands space of multiresolution analysis from scalar to vector, and it has much freedom of construction. Consequently, multiwavelet has orthogonality, symmetry, short support and high vanishing moment at the same time, which cannot satisfy by wavelet. GHM multiwavelet is adopted to transform target joint image, and edge feature is extracted by processing multiwavelet coefficients using method of modulus maximum. Experimental results show that edge extraction based on multiwavelet could realize detection of infrared target and low contrast target.Appearance of hyper-wavelet is made up for weak orientation of wavelet. It is a general concept of orientation wavelets, such as Wedgelet, Contourlet, Curvelet. Among of them, Curvelet is used to preprocess joint image because it has excellent power to represent edge. According to the practical application of low contrast target recognition, image enhancement based on Curvelet transform is employed to preprocess joint image. This method heightens the edge feature of image as well as contrast of gray level. Experimental results show that image enhancement based on Curvelet transform supplies a new solution for successful detection of low contrast target.In a word, applying image preprocessing scheme based on wavelet theory to optical correlation target recognition has a good prospect.