AUTOMATION OF RESEARCH AREA RECOGNITION UNDER WEAK- CONTRAST BORDERS ON A PICTURE OF TRANSPARENT OBJECTS

Abstract

Recently, due to the active development of computer technologies, in the field of science dealing with visual objects, there are problems of creating systems for automatic image recognition. This class of problems does not have a universal way of solving and this leads to the fact that, depending on the needs of a particular branch of  science and engineering, the developers of such systems are forced to solve problems of a private nature.

The purpose of this work is to create a software product by the authors that would allow automatic recognition of the study area of objects having a weakly expressed boundary between the separation of object and background to further determine the values of the optical characteristics of transparent objects, as common features in identification trace evidence analysis the whole by parts.

One of the main problems of detecting the boundaries of the separation of objects in an image is the poor contrast of the transition region between adjacent areas of the divided space, which leads to the need for image processing to segment and correct the contrast. To develop an effective way to limit the recognition area and identify the boundaries of the studied object under the condition of low contrast between the light background and the space occupied by a transparent medium, which differs from the background in optical properties, there is a need to analyze existing image enhancement algorithms and their further practical implementation using mathematical software products of numerical analysis.

Thus, the paper reviews modern algorithms for image recognition methods and algorithms for changing image quality by segmentation and contrast correction. The proposed algorithms are implemented in a computer program developed by the authors of the Sumy Research Institute of the Ministry of Internal Affairs of Ukraine on the basis of the mathematical software MATLAB, which allows to carry out automatic comparative analysis of the studied samples on the basis of their digital images. The questions posed to the authors for further development of this topic are outlined.