Розробка алгоритму виділення стаціонарних об’єктів

Abstract

This work presents an algorithm for detecting stationary objects in video frames based on contrast analysis. The method segments the frame into small blocks and calculates pixel intensity to differentiate objects from the background. Absolute and relative contrast values determine object presence, forming a binary mask. Connected component analysis refines object boundaries, filtering out noise. The algorithm considers real-world conditions by introducing noise to evaluate signal-to-noise ratios. Experimental results demonstrate the algorithm’s effectiveness in identifying objects with sufficient contrast while highlighting its limitations in detecting low-contrast or small objects.