Novel Approach of Video Tracking System Using Learning-Based Mechanism over Crowded Environment

Video tracking is an essential operation in the area of the surveillance system. The model must possess an efficient object detection and recognition capability to facilitate a better video tracking form. Existing video tracking approaches show that they emphasize accuracy without considering the larger proportion of dynamics in mobile objects’ movement patterns. Therefore, this work presents a novel learning-based approach that facilitates video tracking by differentiating mobile objects specific dynamics from other mobile objects. The approach extracts significant cells from the video feed. It subjects them to a learning-based approach in contextual information of all moving objects and dictionaries to extract dominant features. A study outcome shows higher accuracy and faster tracking in contrast to the frequently used approach.