A new algorithm for detecting and separating marine objects from the background in surveillance systems

The main problems in the construction of artificial vision systems in the field of maritime transport are the development and implementation of effective algorithms for the semantic analysis of events unfolding at the scene of events, which allow taking into account the influence of external and internal factors in the process of recognizing the silhouette of an object of study. In this case, the external factors include external illumination, the movement of a marine object, its various external features and properties, and the internal factors include the speed of image signal processing and the computational complexity of the observation system. The main task in the field of detection and recognition of images of marine objects is to identify and separate the object of interest from the background in images under conditions of noise and interference. The paper proposes a new method for detecting and separating objects of maritime transport from the background based on a structured representation of the image pyramid obtained as a result of polygonal-recursive partitioning into polygons of various shapes and sizes. The article presents the results of modeling the proposed method, tested using various video streams, and the estimation of the speed of detection of objects of maritime transport with its help, as well as the results of its work in comparison with already known approaches.