Human body tracking system based on DWT and Mean-shift algorithm on ARM-Linux platform

A modified human body tracking system based on Discrete Wavelet Transform (DWT) and Mean-shift algorithm is proposed. Most of the traditional object tracking systems have many disadvantages like complexity, high computation power and large size. Here, whole system is implemented on ARM-Linux platform with camera mounted on rotary platform. DWT divides a frame into four different frequency bands without losing spatial information. So it neglects most of the fake motions in background as they are decomposed into high frequency wavelet sub-band. Color and spatial information are used as tracking parameters. Mean-shift algorithm takes less number of calculations while converging to new object search window. Ultimate aim of this project is to implement single human body tracking system on ARM-Linux platform, for which minimum computations should be performed. Combination of DWT and Mean-shift algorithm significantly decreases computation power. As shown in results, human body tracking system is successfully implemented.