An inspection approach for casting defects detection using image segmentation

Abstract

Casting defects are significant factors to overall quality of foundry manufactures. The detection and recognition of these defects can provide effective information for production optimization and efficient product life cycle support. The paper suggests an innovative approach for casting defects detection using their classification for a futuristic automated optical inspection. This research proposes a novel technique that utilizes image segmentation to detect casting surface defects. The defects under scrutiny include cracks, blowholes and pinholes that are initially filtered to remove noise and clutter, subsequently the target image is segmented using K-Means partitioning. Hence, the proposed technique shows promising results to classify casting defects using the aforementioned image segmentation technique.