Defects Detection in PCB with Ground Plane by Clustering and Flood-fill Algorithms

Abstract

The known K-means clustering and flood-fill algorithms are implemented. To reduce the color number multilevel segmentation by clustering algorithm is realized. A number of levels are taken from user. The etalon PCB image can be used to determine coordinate of pixels forming contacts. The flood-fill algorithm is used to mark contacts, traces and ground plane. The flood-fill algorithm also is used to control contacts of chains. The model of marked chains is being used to detect defects of shot and open in the manufactured PCB.