| Wavelet Transform is an internationally recognized up to date tool for analyzing time-frequency. It is chiefly due to the "adaptive feature" and "mathematical microscope feature", wavelet transform is becoming a focus of many sciences. One of the basic techniques to construct wavelets is to use "cardinal B-spline functions", which are maybe the simplest and most effective functions having least compact for software and hardware implementation.In this paper, we studied the multiresolution analysis theory of B-spline at the first beginning, and constructed B-spline orthogonal wavelet, which is known as Battle-Lemarie wavelet. Two types of nonorthogonal wavelets based on edge detection are constructed, corresponding to the modulus maxima and zero-crossings. The corresponding scale functions, the general expressions of the low pass and high pass filter banks, and the formulae of decomposition and reconstruction algorithms are all derived for above two detection technique, and the complexity of these algorithms are also studied. Then, we studied the multiscale edge detection of infrared image. An algorithm for the multiscale edge detection of infrared image is presented. The experimental results show this algorithm can fulfill the task of edge detection of infrared image nicely, which founded a good basis for registering infrared targets. In the following, we studied noise filtration of infrared image. A spatially selective noise filtration technique in wavelet domain is introduced. The experimental results show it can reduce noise contents in infrared images greatly and maintain most edges sharp, so it is superior to Wiener filter under these circumstances. Last, we studied the contrast enhancement of infrared images. A new contrast enhancement technique is presented, which can not only extend the dynamic scope of the image gray scale, but also increase the local image contrast. The experimental results show that this technique can perform quite well. |
PDF Full Text Request