Estudo de técnicas de segmentação de imagens da mucosa intestinal

Abstract

It is estimated that the diabetes will affect 592 million people around the world by the year 2035 and by the year of 2014 half of the diabetics do not know they have the disorder. At Medicine is known that there is relation between the type II of this disease and the hypertrophy of intestinal mucosa, but it is not exactly known what binds them. In the present paper, techniques of preprocessing, edge detection/segmentation and post-processing of Image Processing are used in order to segment the villus – being the villus the structure responsible for increasing the area available for absorption on the intestinal walls. The techniques used are: Histogram, Median, Contrast-Stretching, Logarithmic Transformation, Laplacian Filter, Mathematical Morphology, Marr-Hildreth, Canny and Watershed. These techniques are used in three different approaches: firstly, it is used only segmentation. Secondly, preprocessing and segmentation are combined. Finally, preprocessing, segmentation and post-processing are combined in order to compare the result in each approach. The results show that the third approach have better results, although the image is not completely segmented.