FPGA-based real-time visual tracking system using adaptive color histograms

摘要: Visual tracking using a color feature is based on pattern matching algorithms where the appearance of target compared with reference model in successive images and position estimated. The major drawback these methods that such operations are usually considered at top level image processing both due to data's intrinsic complexity high computational cost associated solution real time. probabilistic have been shown be robust versatile for modest cost. However, break down easily when object moves very fast because search only regions interest (ROIs) probability density function (pdf) estimate moving object. In this paper, we propose real-time visual circuit adaptive histograms. We window- structure improve speed circuit. searches all perform operation order main results our work designed implemented physically feasible hardware required tracking. Therefore, has resulted development system employing an FPGA (field programmable gate array) by VHDL (the VHSIC description language). Its performance measured compare equivalent software implementation.