A Method Of Raw Silk Electronic Detection Based On Machine Vision

Raw silk is the primary product in silk industry. The quality of raw silk is the key factor to quality assurance in silk production. Currently, the detection of raw silk is usually performed by experienced human inspectors. In recent decades, computer vision, pattern recognition technology and image processing is developing rapidly. It’s inevitable that industrial automation detection instead of human inspection. Recent researches on various detection of raw silk were analyzed and compared in the paper. On the base of the existing technologies, a method of raw silk electronic detection base on machine vision is introduced, which has a high accuracy. And a raw silk rating standard which is applied to the method is proposed in this paper.Firstly, we should build a test platform of raw silk detection. When the raw silk is moving at a certain speed by a double- roller drive device, the raw silk images are continuously captured by image acquisition system including a Charge Coupled Device(CCD) line sensor, a LED light source and a telecentric lens in the Halcon.The purpose of the acquisition of raw silk images is for image analysis. The extraction of the raw silk trunk was realized by threshold segmentation and morphological opening operation. Then we can easily get the diameter and coefficient of variation(CV) of the raw silk. In order to verify the calculation results of this method, we put our equipment on Uster ME100 tester for synchronous detection. We can find that these two results have a significant correlation relationship when comparing the test results. For the raw silk defects part, we put forward the method of area to divide the raw silk defects into five categories based on the national standard. In addition, we adopted Support Vector Machine(SVM) to recognize the raw silk defects in this paper. At the result, we got the high recognition efficiency.At last, refereed to the indicators of the seriplane inspection, we realized them based on our method. In the case of the raw silk of 20/22 D, we proposed the raw silk rating standard which was applied to this method by analysing the raw silk of 6A, 5A and 2A. Experimentresults show that the method of raw silk electronic detection based on machine vision can detect the raw silk effectively and accurately.