EDGE ENHANCEMENT OF IMAGES BY MULTIRESOLUTION VECTOR QUANTIZATION

Edge enhancement is employed to improve visual perception of images th at are unclear because of blur. In this paper, the point spread functi on is considered to be known, and a method of edge enhancement is prop osed using wavelet expansion and vector quantization for multiresoluti on representation. With the proposed method, sharp images are prepared along with corresponding blurred images, and the local frequency char acteristics are taught to a vector quantizer for subsequent edge enhan cement. The local frequency characteristics are extracted from images by means of wavelet expansion and then classified and processed throug h the vector quantizer. Experiments With blurred images, including tho se with noise superposed, showed the proposed method to be efficient i n edge enhancement and, at the same time, in noise suppression. (C) 19 98 Scripta Technica.