Extraction of Heterogeneous Region from Image Data

Abstract

Recently, wavelet analysis is applied in various fields, in place of traditional transient analysis methods such as short time Fourier transform and Gabor transform. In this paper, we investigate the discriminant accuracy of the heterogenous region from the other normal area. In order to estimate the capability of discriminant, using the non-stationary random field as the image data, we carry out the simulation. The experiment results show the shift within a few pixel from the true boundary. To proceed these realized situations to the practical uses, some important problems to be settled are pointed out.