845

  对电路板的图像进行分割，可以提取电路板中的目标物，以对电路板进行检测。文章使用K均值聚类算法完成对电路板图像的分割，针对传统的K均值聚类算法的不足，提出了使用直方图波形的有效波峰个数来确定K值的大小，并通过使用一种比传统的绝对误差的表示更简洁的表达式，达到了快速分割的目的。对一些电路板图像分割的实验结果表明，文章的方法能够根据目标物的数目有效的确定K值的大小，且比传统的K均值算法减少了运算量及计算时间。 (On the circuit board image segmentation, to extract the target board to the circuit board for testing. Article using the K means clustering algorithm to complete the circuit board image segmentation, the traditional K-means clustering algorithm, the paper proposes the effective use of histogram peak wave number of K values to determine the size, and by using a ratio of absolute error of the traditional expression that is more concise, to the fast segmentation. For some circuit board image segmentation results show that the method can be according to the article the number of valid target value to determine the size K, and K-means algorithm than the traditional reduced computation and computing time.)