Automatic Detection of Minimal Repeated Pattern in Printing Fabric Images

Abstract

Detection of minimal repeated pattern (MRP) is an indispensable ingredient of the color dividing and plate-making system in the IoT-based printing fabric industry. We present a minimal repeated pattern detection problem and then design an effective and efficient neighbor similarity point location (NSPL) method for automatic detection of PRP. We theoretically reveal the effective working mechanism of NSPL and provide a specific implementation by transforming it into a sub-pattern retrieval problem. Evaluation on the printing fabric image set and the experimental results demonstrate that the NSPL is very suitable for the detection of the minimal repeated pattern of printed images and achieves greater performance than the manual and other existing automatic detection methods in detection accuracy, efficiency, and robustness.