Edge enhancement using compactly supported orthogonal wavelets

This thesis examines the use of compactly supported orthogonal wavelets for edge enhancement, within a multiresolution framework. Two methods for generating the modulus of the wavelet transform (MWT) image of an input image at the scale of the input image are developed. The first method involves simple linear interpolation of the MWT image output from each step of the wavelet pyramid. The second method involves expanding the resolution images output from each step of the wavelet pyramid to four times the size of the input image using an allpass filter. The wavelet transform of the expanded image is then used to generate the MWT image. Edge enhancement using the two methods was performed on a set of test images using various order compactly supported orthogonal wavelets and at various resolutions. Based on the results from the set of test images, it was concluded that the MWT methods performed relatively well for images characterized by large contrast between objects within the image and small variation of gray level within each object and poorly for images characterized by small contrast between objects and large variation in gray level within each object. In addition, it was found that, in general, the lower order analyzing wavelets produced better edge enhancement results in the first method and the higher order analyzing wavelets produced better edge enhancement results in the second method.