在分区域的图像测量中,皮带轮图像呈现同心弧特征,一般可采用例如霍夫变换等算法进行分析和检测,也可采用开窗式快速检测算法。为了解决开窗式算法的不足并保证检测速度,考虑检测中皮带轮位置确定,设计了一种快速检测算法。首先描述了检测中曲线转换水平直线的算法优势；然后分析了皮带轮圆心坐标偏移对检测的影响；接着设计曲直变换式的尺寸检测算法步骤。对比实验证明,此算法在保证检测速度的同时,能够提高检测精度,不仅适用于皮带轮尺寸检测,也适用于确定位置的圆形物体尺寸检测。

The pulley image has the characteristics of concentric arcs in the regional image measurement. Hough Transform is always used to analyze and detect the image,also a window clipping method can be used in detecting the image. In order to solve the problem of window clipping algorithm and guarantee the dectection speed,a fast detection algorithm is designed in consideration of location determination. The advantages of converting curves to lines are discribed,and the influence of the pulley center coordinate offset is analyzed. Then the steps of algorithms are designed to detect the dimension using curve to line transform method. Finally,the experimental results prove this algorithm can increase the dectection accuracy while insuring the computer velocity. And this method can be used to detect not only the pulley dimension,but also round objects in the fixed location.