Research On Infrared Image Characteristics And Applications Based On Generalized Wavelet Transform

Recently, infrared image has been widely used in military and civil areas, but the intrinsic characterisitics of infrared imaging lead to edge bluring and image noise, the quality of the images are not suitable for appilication. Here, extended wavelet is introduced to solve those problems. As the extended wavelet can be use as sparse representation of straight line and curve singularity signal, it has been widely used in image processing. As one kind of extended wavelets, Curvelet transform combines anisotropic and multiscale characterisitics of Ridgelet and Wavelet, use the edge as the fundamental element, so it has completeness. In this thesis, the characteristics of infrared images are studied, then edge detection, denoising and background suppression based on Curvelet transform are resreached.Firstly, the statistic characteristics of infrared image histogram is analysised base on the infrared imaging theory. Detailed descriptions of infrared imaging noise characteristics are given, the target and background characteristics are fully analysised, so the basement for application has been set up.Secondly, this paper is base on the extended wavelets, it implement the basic algorithm in image processing. The superiority of Curvelet transform shows that it can be used in infrared image processing.Lastly, Curvelet transform is used in three areas of infrared image processing. In denoising, considering the Gaussian noise of infrared imaging, a new synthesis thresholding method is proposed based on hard-thresholding and soft-thresholding, the experimental results shows the new method greatly improves the subjective visual effects. In edge detection, Curvelet transform and Roberts operator are combined to detect edges clearly and continuously . In small target detection, the background clutter are supressed, the small targets in the infrared image are enhanced, the extensive experiments shows that Curvelet has better result than other algorithms.This thesis also introduces objective and subjective evaluation standard for image quality, which provides effective evaluating method for the infrared images processing.