Research of Image Segmentation Methods

The process of clustering images, that is, of searching for homogeneous regions in them, is called segmentation. It is considered to be the first stage of image analysis, it is the basic procedure for almost all image processing tasks using computer vision systems. Such systems are used in industrial robots of new generations, television tracking systems, transport robots, graphics systems, etc. As with other tasks in this field, segmentation can not be completely formalized, it includes elements for filtering noise and highlighting images.In work the main features of intellectual methods of processing of digital images are considered, examples of processing of such images by means of neural networks are given. Classification of the methods of segmentation is described in various papers devoted to this issue. Often they are subdivided into those that allocate areas of uniform brightness or color, and those that determine the homogeneity of other properties, most often textures. The methods of the first type and their capabilities are well known. In publications of recent years, mainly methods of the second type are considered.