A Comparative Study of Background Segmentation Approaches in Detection of Person with Gun under Adverse Weather Conditions

The ability to detect gun held in hand or other body parts is an ordinary human skill. The same detection problem presents an exceptional challenge for machine vision system. Very few works has been done in the area of automatic detection of moving persons carrying gun in hand under adverse weather condition, although it has several implication in the area of video surveillance and security. The quality of outdoor video scenes suffered from poor visibility and loss it's contrast due to adverse weather atmospheric conditions by scattering of aerosols. In this article, we present a structured comparison carried on between the state-of-art object detection algorithm. Different quality assessment matrices has been used for the evaluation of the performances of state-of-the-art methods. Incidentally relevant public datasets handling such a problem is scanty, if not absent, so far as our knowledge goes. As a result, the present article provides a newly collected several real time crime scene based video data clips from different web sources. The dataset consists of seven sets of data clips, such as, clear day, night, blur, disguise, dusty, foggy, and rainy. The proposed dataset will facilitate the research community to assess the performance of algorithms.