Research of Analysis Filters of Wavelet Transformation of the One-Dimensional Signal Carrying Information on Contours of the Image

Abstract

The article presents the results of research for analysis filters of wavelet transformation of one-dimensional signal of masked image. Masked image is the one-dimensional signal carrying information about the contours of the processed image. The article shows that the contour image data are very specific: in contrast to the whole image data, they have a much higher frequency in nature, since they are the values of brightness of contour and nearcontour pixels. It was found that it is more effective to use the biorthogonal wavelet 8-4 for this kind of data when transformational changes, as it concentrates the energy of such data better than other: it is better than compared to biortogonal wavelet 5-3 by 3 % in average, and better than the biorthogonal wavelet 16-4 by 1,8 % in average.