基于图像处理和UKF算法的小球位置预测

Forecast Tracking of Ball Based on Image Processing and UKF

针对球的非线性运动和系统延时等因素对球的位置预测精度的影响,采用帧差法和改进的霍夫圆检测方法得到球的位置信息,并应用无迹卡尔曼滤波(UKF)算法对球的运动轨迹进行预测。通过摄像机采集球在自由滚动和发生碰撞时的几组图像,在上位机VC6.0软件上实现对球位置信息采集和处理,得到球准确的预测位置信息,满足决策系统的对球位置信息的要求,从而验证了算法的有效性。

In small size soccer robot competition,it is required that the decision system should exactly extract and predict the moving position information of the soccer.The non-linear movements of the soccer,system delay and other factors have influences on the prediction precision of the position of the soccer,according to which,the frame difference method and the improved Hough circle detection algorithm were introduced to acquire the position of the soccer while the UKF（Unscented Kalman Filter） algorithm was adopted to predict the trajectory of the soccer.Groups of images on conditions of ordinary rolling and continuous collision of the soccer were captured using camera.All the algorithms were realized on VC6.0 platform to acquire and process the position of the soccer,and the predicted position of the soccer was then obtained,which could satisfy the requirements of decision system on the position of the soccer.The validity of the algorithms was testified.