基于特征匹配和卡尔曼滤波的机器人视觉稳像.pdf

网络出版时间：2010-12-27 13:29 网络出版地址：/kcms/detail/31.1289.tp1329.008.html 37 5 2011 3 Vol.37 No.5 Computer Engineering March 20 11 2011)05000 A TP242 310018 6 KLT Kalman KLT Kalman Video Stabilization Algorithm for Robot Vision Based on Feature Matching and Kalman Filter XU Yin, WANG Bin-rui, JIN Ying-lian (College of Mechanical and Electrical Engineering, China Jiliang University, Hangzhou Zhejiang, 310018, China) AbstractVideo stabilization is the key of robot vision. Aiming at mobile robots docking operations, image affine kinematics model with 6 parameters and recurrence relations were established. KLT feature matching method was adopted and sub -regional computing procedure was designed. Optimization of sum of absolute difference was used to match feature points. Through analysis of image kinematics model, observation model of intended motion parameters was derived, and the least squares algorithm was used to solving over -determined kinematics equations. Unintended motion parameters were removed using Kalman method to filter synthetic parameters including jitter. Through reverse computing of kinematics model using filtered parameters, jitter was compensated and stabilized images were achieved. On autonomous mobile robot test -bed, experiments were finished. The results show method proposed is effective, and feature points were