利用改进的GLA算法快速检测球场区域

Fast Playfield Detection Method Based on Improved GLA

球场区域检测作为球和球员检测的预处理,是体育视频语义分析和球场三维重建的关键步骤。该算法首先引入了一种分裂和合并机制对GLA进行了改进。然后,利用改进的GLA将每帧的图像象素值进行量化分类,找到主色。最后,结合找到的主色,融合多个非连通的主色区域,利用球场区域为凸多边形的性质,采用Sklansky算法确定球场区域的边界。文中所示算法全部采用基于缩略图的计算方式,实验结果证明该算法具有很高的准确性,并且计算非常快,可以满足球场区域检测的要求。

Playfield detection plays the fundamental role in sports video semantic analysis. It＇s an important first step in player or ball tracking and detecting the location of them on the playfield. An improved GLA （Generalized Lloyd Algorithm） based method was proposed to detect the playfield. Firstly, a splitting and merging method was used to improve the GLA＇s effectiveness, Then, the dominant color was found by the improved GLA. Sklansky＇s algorithm was employed to find the convex boundary of the playfield, because the playfield is often a normal polygon, Finally, the playfield area could be filled through the detected boundary. The algorithm is based on contractible image, because it will improve the computing efficiency. The experimental results verifies the effectiveness of the given method.