Motion Detection in Real-Time Surveillance Using Two Frame Differencing

This research article delves into the creation of a real-time system for detecting motion, particularly for surveillance purposes. The proposed approach relies on computer vision algorithms to analyze live video streams captured by a camera. By scrutinizing these feeds, the algorithms can discern alterations within the scene, specifically identifying objects in motion. The primary aim is to ensure that this system functions in real time, thereby furnishing instantaneous insights to security personnel. The paper expounds on the different steps integral to motion detection, encompassing procedures like background subtraction, foreground segmentation, and object tracking. To validate the system’s effectiveness, an evaluation is conducted using a collection of actual surveillance videos. The findings underscore the system’s capability to accurately detect moving objects across diverse scenarios. The article concludes by contemplating the system’s potential applications, which span security setups, traffic supervision, and ecological monitoring. The focal technique utilized in this study is Frame Differencing, a means of detecting object motion. This approach is adept at distinguishing moving objects within a given environment. In conjunction with this method, the paper emphasizes the viability of background subtraction to further enhance Frame Differencing, thereby augmenting its precision and efficacy. Upon applying this method to a camera, it is discerned that luminosity has a significant impact on the minimal value. Notably, a minimal value of 35 is identified as the most favorable choice. Here, threshold value 35 is selected because we have seen in our experiment that 35 gives the optimal result with our algorithm.