针对半导体封装等工业应用对图像模板匹配的速度要求很高的特点,进行了快速模板匹配算法的研究.同心圆窗口矩匹配是一种速度较快的算法,是将模板分成一系列的同心圆环,然后通过比较模板和待检测图像上对应圆环的零阶矩确定可能匹配点.但由于确定各像素点属于哪一个圆环需要进行复杂的乘积运算,耗费了大量时间.为此,提出了一种生成距离图的改进算法,即匹配前对模板进行预处理,生成距离图,即各像元存储的是距离图像中心（圆心）的半径值.这样,确定像素属于哪一个圆环,只需要经过对各像素的索引值的简单比较,就可以确定.经过实验验证,改进算法可以大大提高模板匹配的速度,满足半导体封装等自动化作业的需求.

The template matching algorithm of digital image is studied to improve the matching speed that is critical for the semiconductor packaging industry. The moment matching of circular windows is a fast algorithm, which divides the template into a series of circular ring and computes the moment of corresponding circular ring between template and source image to find the matching point. But the arithmetic product operation is needed to determine a pixel belonging to which circular ring, which consumes much time. For this reason, an improved algorithm by using a distance map before matching is proposed. The radius is stored in the distance map to record the distance between each pixel and the center of circular rings. Thus, simple comparison operation on the indexes of a pixel can determine a pixel belonging to which circular ring, which can save much time. Experiments show that the improved algorithm can improve the matching speed greatly and meet the needs of precision electronic packaging.