Image Localization Models in Computer Image Analysis Systems

Abstract

The purpose of the certification work is to create a modern information system for effective contour segmentation of digital images, whose models can adapt to the features of the images being processed. In the course of the certification work, the existing models and algorithms for contour segmentation of a digital image were analysed, a computer system for adaptive boundary segmentation in C# was developed and implemented, and the quality of image processing was evaluated. A comparative analysis of the proposed adaptive contour segmentation algorithm with its analogues has shown significant advantages of the proposed algorithm, which are due to higher computational efficiency and quality due to the use of invariant masks and detectors.