Wavelet-based Image Segmentation Feature Recognition in Noise-Distorted Biomedical Images

INTRODUCTION Wavelets are a mathematical tool for hierarchically decomposing functions in the frequency domain by preserving the spatial domain. Since their introduction [1], wavelets have found more and more applications in computer graphics, such as image compression, digital image processing, and feature detection [2-3]. Using wavelets, an image pyramid can be produced which represents the entropy levels for each frequency. In this case study, we demonstrate how this property can be exploited to segment objects in noisy images based on their frequency response in various frequency bands, separating them from the background and from other objects. We compare our noise-robust Haar wavelet-based technique to other standard image processing methods.