Detection Of Tiny Defects Of Stripe Fabric

China has been a large textile country,the textile industry is a huge and stable pillar industrial sector in the national economy of China.In the past and present,the quality of textiles directly determines the grade and price of textiles.The testing of quality is very important in the production of textiles.According to statistics,fabric defects will lead to a 45% to60% drop in its price.Therefore,defects in the fabric seriously affect the economic income of the textile industry.In the production of stripe fabric,the enterprise need to conduct real-time monitoring and to identify the defects area of stripe fabric.Comparing with manual detection,automatic detection reduces subjective interference in the actual production and application process,greatly improve the detection efficiency,accuracy,and reduce the cost of detection.Automatic detection has the following outstanding advantages:low cost,high efficiency,high algorithm stability and detection accuracy,and the structure is simple and easy to operate and maintain.The main requirements of the automatic detection algorithm are: designing a real-time monitoring system,and reducing the false detection rate and the missed detection rate.This paper mainly deals with the two problems of the preliminary algorithm for the defects detection of the striped fabric(1)Mistaken inspection caused by oil pollution(2)The undetected phenomenon caused by tiny defects.For the problem of false detection caused by oil pollution,we distinguish between the essential characteristics of oil pollution and defects(i.e.the defect have the characteristic of a neat boundary,similar to a straight line.However,the boundary of the oil pollution is spreading,and it is not similar to a straight line),and propose the edge detection and least square method to describe the edge information.Not only separate oil pollution and defects successfully,but also save the cost of time greatly.The experiment proves that this method can complete removal of oil pollution successfully,and its accuracy is up to 98%.The problem of error inspection caused by tiny defects(It is mainly manifested in the narrow width of the defects,and the difference between the gray value of the defects and the other parts of the fabric surface is very small),we first use the image processing method to enhance and highlight this tiny defects.Then we propose a method for finding the image column sum by angle integral,using image column sum to make difference and determining whether there is a defect.Finally,using the experience of multimodal ideology through a comprehensive analysis of the multiple features of a tiny defect and fully combining with factoryfabric defect detection requirements and standards,making a final confirmation of fabric defect.The experiment proves that this method can complete the detection of tiny defect successfully,and its accuracy is up to 99%.In this paper,we solve the two problems(That is,the error detection caused by oil pollution and the leak detection caused by tiny defects)through preliminary algorithm of detection of stripe fabric defects.Finally,a set of more maturity and efficient methods for detecting the defects of the stripe fabric is proposed,realizing the automatic detection of stripe fabric defects,and receiving a better detection result.It has more performance in efficiency,accuracy,stability,noise resistance and so on.At present,it has been applied to the defects detection of the stripe fabric surface in the factory and can satisfy the actual demand of factory production,help factories save labor costs,guarantee the quality of the produced fabric and the economic benefit of the factory is improved at the same time.