Towards Optimized Placement of Cameras for Gait Pattern Recognition

Locomotion of an individual i.e., gait is proven to be unique. Recent past has seen a paradigm shift while considering gait as a trusted biometric trait even though it is a behavioral biometric. The study of gait includes body mechanics, changes in muscular action, and uniqueness in body movements. For extraction of features from gait, its proper acquisition becomes an important issue. This makes placement of cameras and their localization as an important domain of research for gait pattern analysis. When gait biometric is used for identification in surveillance purpose, it works in unconstrained manner since there is no predefined path or ramp for recording the motion of a subject. The model proposed in this article approaches for determining best possible placement of optimal number of cameras in a given coverage area. The model also updates/modifies the placement of cameras as the active walking region (path-band) in that area changes temporally. Moreover the model also provides the camera system to work in master-slave mode efficiently utilize the cameras to minimize the computational complexity