Video Background Subtraction Algorithms for Object Tracking
Sourabh Pandey, Prashant Jain, Prabhat Patel · 2022 Third International Conference on Intelligent Computing Instrumentation and Control Technologies (ICICICT) · 2022
Surveillance applications may requires tracking of multiple moving objects in the videos. The Background Subtraction (BS) is the prime stage of tracking methods used for separation of foreground and background. Paper addressed various design issues of BS for the object detection. As the varying illumination, shadows and large interfere motions may affect the efficiency of the BS for objet detection. Performance may also vary in the case of stationary and dynamic background of video frames. Therefore in the paper three different approaches of the BS including pixel differencing, morphological opening, and the moving average methods are evaluated. Initially paper classifies the various BS methods and presets the brief review of recent works. The results of BS in the tri color RGB space are presented using the adaptive thresolding techniques. Paper considers the basic kalman filter for tracking. The efficiency of the background substation is evaluated under the presence of the multiple fast moving objects in the scene. The proposed moving average based BS method out performs in all cases