DIGITAL IMAGE PROCESSING BY IMPLEMENTING UNSUPERVISED LEARNING ALGORITHMS

Maria CRISTEI Universitatea de Stat din Moldova

Abstract

This paper presents the development of an image processing application using color modeling algorithms with the implementation of the unattended K-means automated learning method and the Canny contour detection algorithm for the improvement of high-resolution images and extracting data relevant to their subsequent analysis or display interpretation. The use of the application developed in biomedical imaging processing and analysis (computational tomography, radiography, magnetic resonance, ultrasound, etc.) provides contrast enhancement, encoding the intensity (in gray scale) of monochrome images in color, contours determination and recognition of specific forms, contributing to the easier determination of anomalies. They can also be used to monitor patients and discover / identify diseases and tumors, thus improving medical performance. Keywords: segmentation, machine learning, unsupervised learning, algorithm, color model.