基于颜色学习的移动机器人目标跟踪研究

Object Tracking of Mobile Robot Based on Studying Color

针对目标的颜色特征较难描述的特点,并且不同光线下不同环境中相同目标的颜色信息也会不一样,本文提出了基于颜色学习的目标识别的方法。该方法可以在不知道目标颜色特征的情况下,通过摄像头采集目标的颜色自主地进行分析,然后生成目标的颜色特征信息,从而实现对目标的颜色特征的自主学习。在滤波环节中本文将噪声也进行图像分割,然后进行像素统计,将像素个数小于设定值的区域去除,从而达到滤波的效果。针对目标大小可以获取的情况,本文采用单目视觉对目标进行测距,结合从图像中获取的目标相对机器人的方位角信息,在极坐标中对其方位进行描述,并在定位的基础上对目标进行定位及跟踪实验。

For the difficult to describe the color characteristics of target. Under different hght conditions the same goals have different color information. This paper proposes a method of target recognition based on studying color. In the case the target color characteristics is unknown, this method collect the color of target by camera and analyze independently and generate the color feature information of target. In this paper, the noise fdter is also part of the image segmentation. To counting pixel, ff the number of pixels of the area less than the settings, the area will be removed. For the case target size can be get, this paper get the target distance based on monocular vision. Com- bined with images obtained the azimuth information of the target relative robot, and describe target position in polar coordinates. Making experiment of location and tracking based on the positioning of the target.