RoboCup足球赛彩色视觉系统及其滤波算法

为了满足机器人足球赛对视觉识别子系统高精确度的要求 ,介绍了一种参加RoboCup小型组比赛的计算机视觉子系统和滤波算法。为了快速地识别场上信息，本文采用了一种基于HSI色彩空间利用阈值向量分割图象的方法，这种方法先用阈值法将HSI色彩空间分成若干个子集，通过象素点与子集的位 “与”运算确定象素点的颜色。由于噪声的影响，总伴随目标识别不出来的情况，本文基于航迹预测的原理，提出了一种快速的动态窗口图像滤波方法。实验结果表明该方法可以有效地提高视觉系统的识别精度。

Color Vision System and Filtering Algorithm for RoboCup

To meet the high control accuracy of the vision subsystem of RoboCup robot soccer competition,a computer vision subsystem and image filter algorithm are described.In order to recognize the different object on the ground quickly,a threshold segmentation method used in HSI color space is presented in this paper,first the authors have used threshold method to classify HSI color space into several subsets,for each subset,by applying "bit and" operation on the pixel and its subset,the color of pixel can be defined.It is always hard to recognize the destination because of the affection of noises,so,based on the principle of tracks of target,a fast and dynamic windows image filter method is put forward.Experiments show that the new approach can improve considerably the precision of recognition.