Textile Research Journal · 2020 · 12 citations · 8 references
ComputingComputer ScienceComputer VisionMachine LearningMachine VisionFracture DetectionImage ClassificationImage ProcessingIndustrial InspectionPattern RecognitionSupport Vector MachinesClassification AccuracyComputer Vision For InspectionFabric Defect ClassificationMachine Vision SystemsNeedle DetectionSupport Vector ClassificationSupport Vector Machine ClassificationVision-based Inspection
Problems with knitting needles are one of the main causes of production loss of fabric. In order to detect problems with knitting needles quickly and accurately, this paper proposes a hosiery knitting needle detection system based on machine vision. Meanwhile, according to the working condition of real hosiery knitting needles, a simulated needle cylinder rotary platform is built. The system can detect knitting needle problems, and the needle causing the issue can be identified as the beginning of the fabric defect appears. The production losses caused by the bending and fracture of knitting needles are reduced at the source. In the image processing, the vertical projection algorithm is introduced in the system, and the defect of image binarization is improved. At the same time, the multi-classification support vector machine based on the decision tree is used to realize the image classification. The experimental results show that the classification accuracy of the two levels classifiers in the system can reach 100% and 98%, and the best time of system detection is 0.348 s.
8
Fabric defect detection methods for circular knitting machines
Kazım Hanbay, Muhammed Fatih Talu, Ömer Faruk Özgüven et al. · 2015 · 21 citations