Mixed Kalman/H? Filter for Multi-Object Tracking in Video Frames

Abstract

In this paper, the mixed H-infinity and Kalman filter is proposed for multiple target tracking in the video arrangements. Here, the proposed system will be the joined execution of Kalman filter and the H-infinity filter. The Kalman filter is the best filter that is a linear combination of the measurements. That is why; it is widely used in tracking systems. The H∞ filter, also called the mini-max filter. The H∞ filter does not make any assumptions about the noise and it required only last time step and current state estimation for object tracking. Consequently, there would be no necessity for a high limit of computational stockpiling. This mixed filter uses a lower gain in order to obtain better performance, where as the pure H-infinity filter uses a higher gain because it does not take Kalman filter performance into account. The mixed Hinfinity and the Kalman filter, used to find the location and speed of the objects when objects are moves with a certain motion law. The Kalman filter doesn't limit the mean square error. In this way, the H-infinity filter limits the mean square error and also utilized to limit the impact of unexpected noise whose insights are obscure. Usage of the proposed system was implemented in MATLAB and the execution of this system has better execution.