Algorithm for Detecting Biological Objects and Assessment of their Parameters for Automated Bioassay

Detecting biological objects (BO) at the micro-and meso-levels and observing their vital functions are important for biodiagnostics of the environment. It can also be the basis for biotesting. In this article we propose an approach for detecting BOs moving in an aquatic environment based on the representation of the background by a mixture of Gaussian distributions, followed by applying a mathematical model of a two-dimensional Markov chain to detect BO contours. To assess the vital functions of BOs, we calculate the following parameters: the number of BOs, their size, movement speed, acceleration and movement trajectory. We present the results of the algorithm on the example of detecting cladocerans Daphnia magna and ciliates Paramecium caudatum in laboratory conditions.

I. Introduction

Biodiagnostics of the environment is relevant due to the increasing anthropogenic pollution. Bioassay is currently a method of great importance. It determines the quality of the environment by the responses of laboratory organisms. Micro- and mesobiological objects (BO) living in natural and artificial environments have a lot of advantages for biotesting, including using a set of computer vision algorithms [1]–[4]. Very informative indicators for bioassay and subsequent assessment of the quality of the environment are the presence of certain types of BOs, the dynamics of their development and mortality in various environments, the nature of their movement. Visual assessment of many characteristics of BO during biotesting is very laborious (direct microscopy). It is possible to increase the productivity and accuracy of evaluating these indicators using methods and algorithms of computer vision.