基于动态色彩补偿的足球机器人目标识别方法

摘要

Color properties which contain a lot of information and are geometry - invariant are widely used in object detection in football robot vision systems. But the problem is the guarantee of both the adaptiveness and fastness. In order to tackle the problem of color - deviation aroused by light intensity changing as a result of time and the appearance of shadow, an object detection algorithm based on dynamic color compensation was proposed. Firstly the history information of the objects was used to predict their locations and the searching area was limited to small scopes, thus reducing a lot of searching time. Secondly the paper adopted a method based on dynamic color compensation and performed the adaptive detection of the objects successfully in order to tackle the problem of light intensity changing with the time in the contest field.%在足球机器人视觉系统中,颜色特征由于其信息量大,具有几何不变性并广泛应用于目标识别,难点在于识别的鲁棒性和实时性的保证.针对半自主微型足球机器人比赛中,由于场上光强随时间变化引起的彩色空间漂移现象,提出一种基于动态彩色补偿的目标识别算法.首先利用场上机器人小车的历史运动信息,预测目标的位置,并将目标搜索区域限定在一个较小的范围内,从而大大减少了搜索时间,其次在搜索范围内,针对场上光强随时间变化而引发的色彩漂移现象,实时计算照明变化率,并对颜色知识库进行补偿,实现了目标的鲁棒识别.