| Abstract [eng] |
Computer vision and image analysis is getting more popular in science and industry areas. In order to solve new challenges it is needed to understand computer vision theories and methods. In this research one of computer vision and image analysis areas, edge detection, is examined. Edge detection is used to identify area and perimeter of metal plate used in research. Gradient and second order derivative methods are generally used for edge detection. In this research new methods for computer vision analysis are discussed. Artificial intelligence methods for edge detection are applied: artificial neural networks, support vector machine, k nearest neighbor. In theory part standart Roberts, Prewit, Sobel, Zero-Cross and Canny edge detection methods are analyzed. Artificial intelligence methods and their principles of operation are examined. The most common RGB and HSV color codes are presented. In experiment part influence of internal parameters of artificial intelligence methods and image resolution to edge detection quality and image processing duration are examined. Edge detection quality is determined by comparing received and physical geometric parameters of metal plate used in research. |
