Automatic measurement based on stereo vision system using a single PTZ camera

A parallel binocular stereo vision system and its application to dimensional measurement is introduced in this paper. The proposed system is established by controlling a single PTZ (pan-Tilt-zoom) camera linear motion on a one dimensional precision displacement platform. To achieve accurate measurement results, an improved matching algorithm called as DC-SURF (Distortion Compensated-Speeded Up Robust Feature) by using adaptative filtering in terms of the image distortion during generating scale space is presented. Furthermore, the classic eight-point algorithm and the designed zoom strategy for PTZ camera are adopted for the measurement. The experimental results show that the proposed matching algorithm has better performance than other two classic matching algorithms over distorted images. Furthermore, the experiments demonstrate that without any reference materials the stereo vision dimensional measurement system proposed in this paper can be successfully applied to human height measurement with high precision. The results indicate a potential possibility of our approach to be used in other computer vision applications.