Research On The Key Technology Of Defect Detection For The Glass Fiber Cloth Base On Machine Vision

With the development of economy,chinese textile has entered a stage of rapid development.However,the labor intensive of most domestic textile enterprises is very high,and the defect detection of textile still depends on manual inspection.This method has many disadvantages,such as strong subjectivity,low precision and high working strength.Machine vision technology is becoming more and more mature,so it is more and more widely used in the industrial production process,on-line defect detection of textile products based on machine vision has become an important development direction of textile quality control.The development of the on-line detection technology is relatively early foreign countries.However,it is expensive and high cost to introduce on-line defect detection equipment.The domestic research is mainly aimed at a certain kind of algorithm research and only applies to a kind of fabric defect detection,and it can not be directly applied to the defect detection of glass fiber cloth.Therefore,it is very important to study the key technology of glass fiber cloth defect detection,which can promote the automatic production of glass fiber fabric and the rapid classification of fabric quality.This paper focuses on the key technologies of glass fiber fabric defect detection system,The main contents include:Based on the texture characteristics,testing requirements and production environment of glass fiber cloth,the overall scheme of glass fiber cloth defect inspection system is designed,a machine vision inspection platform for glass fabric is constructed.According to the defect characteristics of glass fiber cloth,the light source configuration scheme of back lighting and the multi camera detection scheme based on GIGE are determined.The high contrast fabric image is obtained,and the difficulty of defect recognition is reduced.A multi camera synchronous acquisition image is proposed to solve the problems that the glass fiber cloth is wide and small field of view industrial CCD.Then,the collected images are processed with image stitching technology based on the template matching method and Harris feature points,we compare and analyze the registration accuracy and stitching speed of two methods.Starting from the real-time and reliability,this paper chooses the stitching method based on template matching to mosaic of the glass fiber cloth image.In order to solve the problems of low efficiency and poor real-time performance,a fabric defect detection method based on Blob analysis is proposed.Firstly,the fabric image is smoothed by mean filter so as to weaken the noise and the interference of fabric texture,then iterative method is used to find the best threshold,and the image is divided into Blob and background pixels,morphological segmentation is used to adjust the shape of the Blob after segmentation.Finally,the image is connected and the feature is extracted.Through the minimum circumscribed rectangle fitting of Blob area,the number and size of the defect characteristics are obtained,and the common defects such as splits,skips,holes,stains,etc.of glass fiber cloth are recognized.The experimental results show that the method is simple,robust and reliable.It is an effective on-line detection method for fabric defects.And it is developed through writing C# and Halcon codes and connecting Visual Studio 2010 with SQL Sever databases.It includes four parts: image acquisition module,human-computer interaction module,image processing module and defect data statistics module,the defect detection of glass fiber cloth is realized.After debugging and experimental verification of the software system,it proves that the designed system has stable and real-time performance.