基于距离平方模板的快速质心搜索算法

智能识别领域,严格要求系统能快速准确地搜索跟踪目标质心,一个基于欧式模板的快速质心搜索算法得到了广泛应用。但该方法仅能求得近似解,难以求取目标质心的精确坐标。提出一种基于距离平方模板的快速质心搜索算法,并将该算法从二值图像推广到灰度图像的质心搜索。两种算法对同一图像质心搜索结果进行对比,欧式模板算法误差在2%～6%范围内,而距离平方模板算法精度为100%。结果表明:在精确度要求较高的场合,特别是小目标追踪识别时,必须使用距离平方模板进行目标质心的计算。

A fast centroid search algorithm based on the square of distance template

In the field of intelligent recognition,demanding the system can search and track the target centroid quickly and accurately,a fast centroid search algorithm based on Euclidean distance template is widely used.However,this method can only obtain an approximate solution while difficult to obtain precise coordinates of the target centroid.A fast centroid search algorithm based on the square of the distance template is proposed and the algorithm is extended to the gray value image centroid search from the binary image.Results from two algorithms on the same image were compared,error of algorithm that based on Euclidean distance templates range from 2%to 6%,and accuracy from our algorithm is 100%.The results show that in situations requiring high accuracy,especially in small target tracking identification,the template used to calculate target centroid must be square of the distance template.