Defect Classification Method of PCB Solder Joint by Color Features and Region Segmentation

Abstract

This paper proposes a solder joint inspection method based on region segmentation and color feature extraction. This paper classifies various types of solder joint by an inspection procedure that is divided into a feature extraction stage and classification stage. In the feature extraction stage, a solder joint image is divided into several sub-regions, and then various color features are extracted from every sub-region. In the classification stage, classifiers are generated by training color features extracted from several samples. Finally, the inspection of solder joint defects is performed using the color features of the test samples. The performance of the proposed method is evaluated by inspecting ten types of solder joint defect. The experimental results have verified this method’s effectiveness in terms of recognition rate and speed.