The Design Of Cashmere Heterogeneous Fiber Detecting System Based On Image Processing Technology

With the continuous improvement of the quality of clothing for people,cashmere has become one of the most economical and popular raw materials in the textile industry due to its warmth,softness and dignity.However,China's cashmere sorting has long been done by manual sorting.It is a time-consuming and laborious work,and the workers are easily fatigued,resulting in low efficiency.In order to improve the present situation,it is urgent for people to replace the artificial machines,and the cashmere fiber sorting needs to be done on the basis of computer image processing and machine vision technology.Therefore,this paper designed a set of cashmere fiber detection system,the main contents are as follows:The first is preprocessing the image.The process of image preprocessing includes binarizaiton,image enhancement and image denoising.The binarizaiton of the fixed thresholds simple and can improve the speed of operation effectively.The automatic threshold binarizaiton is easy to use and more widely for the application of the crowd.In this paper,a variety of image enhancement and image denoising methods are compared.According to the experimental results,the piecewise linear transformation enhancement method and the median filter denoising method are selectedAnd the second is the identification of different fibers.Because of the irregularity of different fibers,it is impossible to find a specific template to compare and recognize,and a discriminant method based on continuous function is designed by using its pixel characteristics,but it was abandoned for its high failure rate.Then,according to Depth-First-Search Algorithm,a cashmere fiber recognition method is designed to test the continuous number of pixels.If the set value is reached,it will be judged as a different fiber.At the same time,due to industrial camera,vertical will always appear in the image interfering results.According to the samples,the average number of vertical pixel values is obtained,and when the number of pixels in a column is greater than the value,it is determined as a vertical line.The last is the location and removal of different fibers.Accurate coordinates can be obtained according to the identified different fiber position,the specific time of the nozzle start and the starting nozzle serial number can be obtained by the distance from the fibers to the nozzle and the speed of the conveyor.When the fibers pass through the nozzle queue,the console sends instructions,so that the different fibers can be sent out accurately.The cycle can eventually achieve the purpose of detecting and removing different fibers.In this paper,the image enhancement and image denoising method are compared by MATLAB,and the development of cashmere fiber detection system is completed by C#and C language.The final detection speed is 655mm/s and the recognition rate is over90%.The experimental results show that the proposed algorithm is effective,and the cashmere fiber detection system basically achieves the expected effect.