A STUDY OF IMAGE SEGMENTATION BY PARALLEL COMPUTATION

Researchers spend part of their time doing microscopic analysis of muscle tissue samples in order to classify different fiber types. This analysis lets them identify some metabolic and contractile muscle properties, allowing the diagno- sis of muscular and neuromuscular diseases. Digital image processing can reduce the time spent on this tedious laboratory work. In particular, image segmentation identifies regions or contours of the objects in an image and is frequently a nec- essary, and computationally intensive, stage in the process of digitally classifying images. The aim of this paper is to show the parallel implementation of three image segmentation methods and their application to biomedical images. The methods are compared by considering segmentation quality and its potential for fast execution on parallel computing machines. Key Words: Digital Image processing, image segmentation processes, sample mean and sample variance, parallel computing.