Research on Defect Detection System for Print Based on Machine Vision

Defect inspection based on machine vision for print plays an important role in print industry. Numerous techniques have been applied in it to effectively reduce labor cost and avoid human error, improving products quality. This study aims at discussing a new defective detection algorithm for print, and the systematic design process is presented in the paper. To obtain high accuracy and real-time performance, the novel coarse-to-fine image registration method combined with preprocessing is proposed. For further decreasing error rate, the error corrected function combined with inverse mapping mechanism is introduced to finally identify real defects position and label them. Experiments show that the improved system has better performance with strong reliability and adaptability.