Concealed weapon detection from images using SIFT and SURF

Nowadays the global attacks of terrorism continues to escalate, so it is exigent to find the different techniques to ensure the safety of individuals at the public places like airports, railways and sensitive buildings etc. This paper presents proposed method for feature extraction from images to detect weapons especially different shapes of guns. Most effective algorithms SIFT and SURF used for feature extraction applied on database of images of luggage. Both algorithms give results that are invariant to scale, rotation, occlusion, change in illumination and noise. In the proposed method, input image is divided into parts in order to detect multiple guns of same or different type. Experimentally proposed method shows 90% efficiency and SURF is faster than SIFT. Algorithms are tested on the database of images containing weapons.