Research On The Real-Time Detection Algorithm For Defects And Hardware Realization

Fabric defect detection is one of the most important procedures effecting the manufacturing efficiency and quality in fabric industry. So far, most domestic factories accomplish this work by human vision, which makes the quality of fabric devoid of consistency and reliability. It was incompatibility with the development of textile industry. Inspection Automation is the best resolution. As a prolonged functionality of human vision, computer vision has been collecting more and more favor in the application of industrial surface detection such as fabric defect detection with the advances of digital integration and digital image processing techniques.In this thesis, the main elements which consist of the vision system are firstly introduced. Then the performance requirement is analysis as to emphasize the key futures of research——effective and real-time. The appropriate arithmetic can insure the efficiency. The real-time performance requires not only appropriate arithmetic but also advanced hardware. The fabric defect detection algorithms in the researches and applications are summarized and are conclude to two fields of method, space and frequency. Typical examples in the two fields are given respectively. They both have their own advantages and disadvantages in on-line inspection. Then a new thought that combining them is put forward.The first reaction of human eyes for image defect is that images containing defects are less homogenous than those that are defect-free. It is the start point of the thesis, too. By choosing the local roughness of images as the characters, the global homogenous of images can be measured to indicate whether the images contain defects. The wavelet transform module in the proposed segmentation algorithm constitutes a preprocessing step with the objectives of attenuating the background texture and accentuating the defects. The image fusion and enhancement is the father steps to complete the task that was initiated by wavelet transform, that is background attenuation and defect accentuation. The simulation from Matlab proves the efficiency of the algorithm.Because of owning the MAC resources and architecture that is adapt to run parallel algorithm, FPGA is chose to be the core device of the hardware system. According to algorithm, FPGA can increase the throughout through increasing pipeline levels. The hardware emulation using Altera's software, QuartusII 5.1, verifies that the system is real-time.