On problems of intelligent monitoring for traffic

Abstract

The article considers the natural-scientific and practical approaches to processing algorithms of virtual detectors and applications for pattern recognition problems. The recovery problem of object with smooth boundary is discussed. This method is generalized for object with piece-smooth boundary. The method of the estimation of ‘image depth’ is presented and the measure on a digital matrix is introduced for object-detecting problem on a homogeneous background. Methods of image recognition on the video sequences offered in this article use the concept of space depth and are based on the modelling of human vision and bio-inspired methods. The exact problem formalizations and mathematical solution methods are used where possible as it is an important element of the qualitative computer vision models. There are the formalizations and conceptual solutions of intelligent monitoring problems for traffic flow controls based on the method of virtual detectors. These problems have important practical value for control of traffic flows on megapolis network.