Information System for Automating Processes of Biological Objects Detection, Recognition, and Measurement

Abstract

The article briefly describes the existing and proposes novel approaches to the detection, estimation of the number and measurement of physical parameters of biological objects in the frames of computer vision systems. The object detection method is based on a two-stage algorithm: at the first stage, the approximate position of the object is determined using edge detection algorithms; at the second stage, graphic models of the object are built by creating graphical primitives. An algorithm for constructing graphical primitives is developed using the color of a biological object as a parameter. The presence of the second stage can significantly reduce both false positive and false negative results. To implement the presented model theoretical studies and practical experiments are carried out. During experiments, it was revealed that the proposed approach allows increasing the accuracy of determining biological objects, since it works with a color image and, accordingly, uses the ranges of digital color channels. In addition, the construction of graphical primitives, taking into account color ranges, makes it possible to estimate the physical characteristics of a biological object. The proposed two-stage technique for the automatic determination of biological objects makes it possible to solve the problems of positioning ultraviolet and low-frequency laser devices for point irradiation. The presence in the algorithm of the stage of constructing graphical primitives on biological objects makes it possible to estimate the degree of fatness of the animal, which is one of the components of determining its state of health, and can be part of an automatic system for contactless determination of the animal’s weight.