Study of methods for extracting contours of objects on raster images of amber samples

Abstract

The article discusses the software and algorithmic methods of selecting the contours of objects used in technical vision systems (TVS). The effective operation of the TVS requires the development of methods and algorithms for processing raster images of objects, taking into account their features for the applied field of the TVS application. For example, TVSs are widely used in industry, agricultural sector, medicine, cosmonautics, automotive industry, etc. Effective methods and algorithms for detecting objects and their contours on raster images significantly improve the performance of TVS in applied tasks. Also, these methods exclude subjective factors influencing decision-making. TVS can determine fine particles, various inclusions, cracks, and texture on the examined surface much better, using various methods and algorithms for contour detection and further processing of the obtained results. Thanks to this, we can get an accurate representation of the condition and properties of samples, raw materials and future products. It is also possible to determine the amount of material that will be lost during the processing of the material of the samples, and to localize in the three-dimensional volume various inclusions that can affect the further quality of the processed product. Therefore, the article provides a detailed analysis of existing methods for detecting contours on digital images of amber for its further classification and quality assessment. The best methods among the existing ones are determined, namely the Kenny, Sobel, and Laplace operators. The use of these operators in TVS provides the most complete and reliable information for building a three-dimensional model, classifying and evaluating the quality of amber samples. The open source computer vision library was taken as the algorithmic and software basis of the conducted research. This library contains most of the basic methods and algorithms for determining contours of objects in images.

Keywords

• methods of selecting contours of objects
• technical vision system
• machine vision