Crankshaft defect detection based on machine vision

Abstract As a core component of the Marine engine, crankshaft crank is required to have high surface quality ensure that ship does not fail during voyage. After analysing defect characteristics, aiming at problems such as inaccurate detection single feature in bend defect, which leads poor effect, classification method based on machine vision was proposed. The recognition accuracy can reach 97.8%. Compared with identification by texture and geometric features alone, be improved, requirements industrial better met.