Crankshaft defect detection based on machine vision

Abstract

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