Shape defect detection for product quality inspection and monitoring system

This paper presents an automated computer vision system of shape defect detection for product quality inspection and monitoring system. Soft drink bottle is used as a tested product for the proposed system. The analysis framework includes data collection, pre-processing, morphological operation, feature extraction, and classification. Morphological operation technique is used to segment the image of the bottle via erosion and dilation process. Through this technique, the defect in the bottle structure is described from the feature set such as area, perimeter, major axis length and extend. Then, the bottle is classified either it is pass or rejects from the estimated parameters using Naïve Bayes classifier. The results have proven that the proposed system can be applied to differentiate bottle according to shape with 100% accuracy using 100 samples.