Research On Face Detection And Tracking System Based On OpenCV In Embedded Environment

Today’s society has maintained a rapid development of science and technology. Theembedded system’s storage capacity and computing power has been greatly improved. Ithas small size, low cost, portability, low power consumption and good crop. This is alsoselected embedded platforms for real-time face detection and tracking of importantreasons. The face tracking is one of the hot areas in the field of computer vision. In thispaper, using open source visual function library OpenCV to achieve face detection. Itcontains more than500interface functions. It can provide a very friendly interface onmachine vision functions resulting in reduced product development cycle.This paper describes a method of human face detection and tracking of researchprogress and principle. This article focuses on continuous Adaptive CamShift(Continuously Adaptive Mean Shift) algorithm. This algorithm is not affected by the sizeand shape of tracking target. The algorithm has the better robustness. And the CPU usageis very low. Embedded real-time human face detection and tracking specific functionalrequirements to build: version Ubuntu10.4-Linux operating system, the kernelcompile-tools, Linux-2.6.38, GCC-4.3.2, arm-Linux-GCC-4.3.2, cross-compilation tools.Cameras use a Logitech C270USB based on CMOS PC camera. Porting OpenCVcomputer vision library to set up ARM platforms, and real time face detection usingOpenCV to tracking purposes, and are able to develop stable operation on the Board, onthe basis of theoretical research and experimental validate the face detection functionmodule analyses.CamShift algorithm must be manually selected tracks on face. This introduces theAdaBoost algorithm for fast human face in front of the face tracking. This algorithm canautomatically enable real-time human face tracking. Experimental results show that theproposed improvements to the face tracking algorithm for fast, accurate real-time facetarget tracking.