Parallelization Of A Fabric Defect Segmentation Method Using OpenMP

With international market increasing competition, product quality is an important sign,which determine whether the enterprise has the competitive ability, and it is an importantfoundation of survival and development for an enterprise. Fabric defect detection is a crucialstep to guarantee product quality and improve qualitative control level in the process oftextile producing. In domestic textile enterprises, fabric defect inspection is performed bymanual detection, whose results are easily affected by the subjective factors and have pooraccuracy and low efficiency, and labor costs lead to substantial increases of production costs.Defect detection technology based on machine vision has provided an important means forautomatic defect detection. Considering detection efficiency in a fabric defect inspection, aparallelization of a fabric defect segmentation method using OpenMP has been studied in thispaper. The research subject has supported by “13115” Major Innovation Project of ShaanxiProvince.After reading lots of literature at home and abroad, a defect segmentation algorithmbased on Pulse Coupled Neural Networks (PCNN) has been studied and its parallelization hasbeen realized using OpenMP on the multi-core processor. The main research works include:(1) a model of improved PCNN for fabric defect segmentation;(2) an improved FVDalgorithm for fabric defect feature extraction;(3) a parallelization method using OpenMP forfabric defect detection. Experiment results show the method reaches expected requirementsof performance for a online fabric defect detection. These research findings have providedcertain reference for online fabric defect detection based on machine vision.