Background subtraction and human detection in outdoor videos using fuzzy logic

This paper presents a scale and view invariant approach for human detection in the presence of various other objects like animals, vehicles, etc. Human detection is one of the essential steps in applications like activity recognition, gait recognition, human centric surveillance etc. Inaccurate detection of humans in such applications may increase the number of false alarms. In the proposed work, fuzzy logic has been used to model a robust background for object detection. Three different features are extracted from the contours of the detected objects. These features are aggregated using fuzzy inference system. Then human contour is identified using template matching. The proposed method consists of four main steps; Moving Object Detection, Feature Extraction, Feature Aggregation, and Human Contour Detection.