The Research Of Fast Recognition Algorithm On Fabric Textile Based On Machine Vision

With the automation technique applying in production, detection and management increasingly, the auto inspection of fabric is becoming more and more important. The high cost, low accuracy and very slow performance of human visual inspection has enforced the development of on-line machine vision-based systems capable of executing effective inspection tasks. The problem of fabric inspection is very important and complex and the research is widely open in this field.Chapter 1 mainly introduces to the research work existed, and put forward still needing a further research to reach agreement. Chapter 2 prepares the processing method to carry on the research and analysis, the fabric image previous processing is divided into two steps. The experiment enunciation Gabor filter is better, But the Gauss filter is put forward because it can be optimized easily. Chapter 3 introduced the district recognition. It mainly includes a few aspects, such as the edge detection, the threshold partition and the blobs processing etc. The experiment expresses Sobel operator can calculate the defects'edge better. The binary image need to be connect or clean otiose smaller blobs, Therefore still need an after handle to resolve these problems. Chapter 4 put forward how frame the whole system, it mainly introduces to use of the hardware and structure, the processing structure of the software etc.The paper studies some problems that how to detect the defects on the fabric textile fast and precisely. These problems include the fast filter and textile processing, the fast edge detection etc. Put forward a new method that can detect the defects in real-time. The experiment expresses that the method is valid and viable.The problem of fabric inspection is very important and complex, and the research is widely open in this field.