Efficient Histogram based Background Subtraction Approach
R. S. Amshavalli, Shail Mohammed Shakeel, Seva Vijay Vardhan, K Naveenraj, N. Jeenath Shafana · 2025
The widespread use of video surveillance has resulted from the increased need for security in many aspects of life. Since IoT era emerges, it demands for smarter video surveillance system that can analyze the video footage and detect for the unusual occurrences instantly to provide the security at its fullest level. To bring smartness in the system, proper pre-processing of video data must be done and preprocessed data will be injected into knowledgeable framework. And now it is the part of the modern technologies like deep learning, artificial intelligence, semantic processing etc to perform the abnormal motion detection. The study mainly focuses on the pre-processing step of a smart video surveillance system. Owing to the growth of modern computing paradigm, there are potential of handling continuously generated large volumes of surveillance data across IoT devices. To handle and analyze such level of ever-increasing input data, the need of the hour is stronger version of pre-processing techniques. In pre-processing layer of a surveillance application, background subtraction or background elimination is the crucial step in extracting foreground information which then can be fed to the later stages. The main focus of the work is to employ histogram-based background subtraction algorithm, which is used for removing the distracting backgrounds. Experiments are conducted on a real-time dataset to evaluate the suggested technique. Data set considered is obtained from the education institution. It is also validated against the video surveillance benchmark dataset called VIRAT. The computed results demonstrate the superiority of proposed algorithm with an accuracy of 94% on real-time dataset and accuracy of 87% on VIRAT dataset.