AN EFFICIENT QUALITY CONTROL SYSTEM BY MACHINE LEARNING FOR SURFACE DEFECTS

Quality control plays a crucial role to meet the high and accurate quality of production in many manufacturing industries. The high quality products may become unreliable due to surface defects. Generally, quality control is more important in automotive industry such as in the field of car-body parts manufacturing. The exterior appearance of the car body should be smooth surfaces and edges with flawless nature. In order to build such flawless body parts, surface defect detection system is taken into account. In this paper, an Automated Defect Detection (ADD) system is presented. The design of the ADD system consists of two steps. The first step is considered as a classification system where the given image is classified into defected or non-defected using Gabor expansion with Principal Component Analysis (PCA). The next step is segmentation where the region of defect is identified using local thresholding. The evaluation is performed on raw alloy steel surface and machined surfaces of steel and cast iron. Results prove that the ADD system classify the input image into defect/no defect with 100% accuracy by a simple nearest neighbor classifier and with 94.5% detection accuracy for the segmentation system.