Motion Tracking of Intelligent Surveillance based on Machine Vision and Infrared Spectroscopy

Circumferential point method has been adopted in this paper for calibration. Binocular vision dynamic tracking technique is used to track the independent motion axis of each joint of the robot, and then construct and fit the axis of joint to attain kinematic model parameters. This method does not need previous complicated steps of numerous robot kinematics models iterating solution equation; for the independent-developed robot control system, the linkage model parameters attained from above method can offer feedback to the robot timely, execute parameter amendment and compensation, and then improve the positioning accuracy of robot. This method is with advantages of generality, momentarily and easy operation etc. Experiment has shown that through parameter feedback compensation, the positioning error of the independent-developed robot has been decreased significantly. This method is easy to realize, which provides reliable evidence for robot accuracy research. Introduction With the development of industrial automation, the robot has been applied in production more and more extensively. Generally speaking, the repeatability accuracy of robot is high, but its absolute accuracy is relatively low, which will cause great influence on the using performance of the robot. For the robot without calibration, the accuracy error can be few millimeters. Therefore, it has to make accurate calibration for robot in a lot of applications. Calibration is the integration process of modeling, measurement, parameter identification and error compensation [1]. The purpose of calibration is to decrease the geometric parameter error of robot. Geometric parameter error of robot is the error deviation between nominal value and true value of geometric parameter in kinematics model. The kinematics calibration of geometric parameters of robot connecting rod involves in the conversion problem between measurement coordinate system and robot base coordinate. Some calibration processes establish the coordinate system of measuring instrument on the base coordinate system of robot, but this will cause complicated coordinate system conversion, make the calculation process complicated and even difficult to solve [2~4]. Binocular Vision Dynamic Tracking Technique The main function of binocular vision dynamic tracking technique is to detect moving target automatically and realize real-time tracking and three-dimensional ranging. The basic skills are as shown in figure 1. Target detecting Binocular matching Target positioning Camera 1 inputs image Camera 2 inputs image Fig.1. Binocular vision dynamic tracking steps Basic principle of binocular vision positioning system is as shown in figure 2. With the adoption of parallax formed by the same feature point P at left and right images, it can measure the depth information of feature point. 402 This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Copyright © 2017, the Authors. Published by Atlantis Press. 3rd Workshop on Advanced Research and Technology in Industry Applications (WARTIA 2017) Advances in Engineering Research (AER), volume 148