In this paper a novel method for detecting and recognizing weapons like different types of handguns in different scenarios through webcam video is proposed. There is significant need of prevention of terrorist attack and the earliest detection of such threats is a major concern to ensure human safety. So there is need of deploying these smart surveillance cameras and security systems around the globe. With the high increasing demand for intelligent security surveillance cameras, various

more »

... es have been undergoing in a detection system for pre-processing, manipulation and interpretation of video frames and images. The detection process contains analysis of both static and dynamic images to continuous monitor the area for security. Different tracking techniques have been developed with a capability to detect as well as recognize the object in real time scenarios. Tracking of an object and its various associated issues such as extraction of object features, analyzing shape of the object and analyzing object position are the active areas of research. The researchers are continuously proposing new efficient algorithms to enhance the system capability for tracking the suspicious different types of object in motion and extracting the attributes to analyze the object properties if there is a weapon present in the scene or not. The proposed hybrid neural network system combines the learning capabilities of neural networks for nonlinear function approximation. The accuracy and efficiency of the system is detected by different objective as well as subjective parameters for evaluation purposes.