Search

the article discusses the features of the synthesis of images of artificial objects with the aim of their search and recognition in the vision system of an underwater robot. The results of research and development to given: a database - a classifier of features that allows you to store information about the reference underwater objects; software module for generating images of reference objects and their insertion in a sonar image file. In the modeling program for the synthesis of underwater objects in the HBO image, the algorithms for introducing objects used by taking into account the peculiarities of the formation of HBO images by sonar. The program allows implementing several objects of spherical, cylindrical and cubic shapes with specified parameters, which allows simulating the situation of search for an artificial object according to given geometric features against the background of other objects. The developed software allows simulating underwater objects according to the characteristics close to real objects with the help of their preliminary processing and subsequent implementation in real HBO images

The article discusses the segmentation algorithm of digital image based on the measurement contrast of fine detail in the normalized CIELAB system. Using normalized CIELAB system allows you to segment fine detail in accordance with the contrast sensitivity. Focusing the camera is carried out on the basis of the search of function maximum that determines the dependence of the number of segmented fine detail from the focal length. Proposed software as part of an interactive digital camera control system implements this function. The results of experiments and comparative analysis with other methods focus are given. It is shown that the proposed method improves the accuracy of the focus when photographing images in low light.

In this paper we present a novel assessment method of fine-details level (FDL) of photorealistic images. The method is based on the search algorithm and recognitions of fine-details in the normalized color space. The novel image metric allows evaluating quantity of fine-detail fragments which are noticeable an eye. Results of an assessment of method stability to distortion factors are given and applications in the multimedia system are discussed.