View Printable Version

Jinhui Qin, September 2000--October 2001, "A New Wavelet-based Method For Contrast/Edge Enhancement", degree conferred on Spring 2002 convocation, Computer Science Department, University of Western Ontario, Canada.

M.Sc. Thesis Abstract

Digital images are often deteriorated by noises due to various sources of interference and other phenomena that may affect the quality of these images. Image enhancement is a mathematical technique that is aimed at realizing improvement in the quality of a given image. In this research we focus only on certain image enhancement issues, namely, contrast enhancement and edge enhancement.

Contrast enhancement is usually achieved by histogram equalization in the spatial domain to redistribute gray levels uniformly. However, it has a drawback, which is some information might be lost. Meanwhile, edge enhancement attempts to emphasize the fine details in the original image. But in spatial domain it is hard to selectively enhance details at different scales. Moreover, in the spatial domain, applying contrast and edge enhancement techniques in different orders may yield different enhancement results.

Wavelet-based image analysis provides multiple representations of a single image. It decomposes an image into approximated-coefficients and multi-resolution detailed-coefficients. To overcome the above spatial domain enhancement issues, a new wavelet-based image enhancement method is proposed. The proposed method histogram-equalizes the approximated-coefficients. At the same time, it high-boost filters the detailed-coefficients at selected resolution levels separately. Since contrast and edge enhancements are applied on different components in the wavelet domain, the contrast and edge enhancements should not affect each other. Moreover, the order of applying both of them becomes irrelevant.

The proposed method is implemented in C++ on UNIX platform. Various parameters in the method are adjusted and discussed based on experiments. These parameters include number of transformation levels, different wavelet filter set selections, the stretching factor of coefficients in a component over which the histogram equalization is performed and the A value of the high-boost filtering function. Final experiments show that utilizing the proposed method can achieve robust contrast and edge enhancement.