VLSI architectures for mean-shift based object tracking

Abstract

The demand for real-time video surveillance systems is increasing rapidly. The purpose of these systems includes surveillance as well as monitoring and controlling the events. Today there are several real-time computer vision applications based on image understanding which emulate the human vision and intelligence. These machines include object tracking as their primary task. Object tracking refers to estimating the trajectory of an object of interest in a video. A tracking system works on the principle of video processing algorithms. Video processing includes a huge amount of data to be processed and this fact dictates while implementing the algorithms on any hardware. An efficient video processing algorithm is adopted here for estimating the trajectory of moving objects in a video. The tracking algorithm is based on mean-shift iteration technique. This method tracks accurately the target object in a sequence of video frames. The key objective is to implement the algorithm on an FPGA platform with less computational complexity and hardware utilization for real-time applications. Two VLSI architectures for the mean-shift based object tracking system are implemented and verified. The FPGA target device used here is XILINX xc5vlx110t. The architectures consist of many divider modules which plays a significant role in the performance of the system. Divider includes shifting and addition operations repeatedly to get a particular result. Hence emphasis should be given for the design of an optimized divider unit. Here a serial divider using non-restoring algorithm is implemented in 90 nm technology using CADENCE tool.