摘要
在图像自动报靶系统中,运用广义Hough变换(GHT)快速地获取靶纸图像子像素级圆心坐标是其关键。但GHT存在着过多无效边界点参与运算,参数空间较大且均等细化等缺点,导致速度太慢而不能实用。在先验知识的导引下,本文对GHT求取圆心坐标的算法进行了有效改进,图像空间引入边界追踪算法,消除了无效边界点参与运算。参数空间则有效缩小并采取由粗到细两级量化结构。实践证明,改进后的GHT显著提高了求取圆心坐标的算法速度且获得了子像素级圆心坐标。
It is important to acquire the subpixel coordinates of center of circle of target image fast by using Generalized Hough Transform(GHT) in automatic scoring system,but there are some disadvantages in GHT,such as much more computation intermixed with null boundary points,large parameter space and the space is equally divided,etc,which result in so slow speed that GHT can't be used in applications.Guided with the former knowledge of target image,this paper puts forward a revised GHT algorithm,the algorithm introduces tracing boundary in image space which can avoid the computation intermixed with null boundary points,on the other hand,reduces parameter space rapidly and takes a two dividing structure from rough to fine.A lot of experiments show that revised GHT improves the speed dramatically and acquires the subpixel coordinates of center of circle.
出处
《计算机与现代化》
2003年第10期43-45,50,共4页
Computer and Modernization
