Research And Application On Image Processing Algorithms Of Paper Defects Based On Machine Vision

Today, with the increasingly fierce competition in paper industry, ton-paper profit is getting lower and lower and the cost of raw material is rising higher and higher. Improving products'quality and reducing the labor cost has become a very important factor of influencing each paper mill. With the great improvement of the machine speed, relying on manual visual inspection into the paper defects has been unrealistic and it also faces that the paper appear more defects. Once the factory's products are found paper defects by users, paper mills will not only be subject to complaints and compensation for economic loss, but also lose customers. In the market filled with increasingly stringent quality requirements and more competition, the paper defects automatic detection system, combining the use of machine vision technology, motion control with network communication, brings paper manufacturers not only a technical change. Web Inspection System (WIS) is online. It can effectively monitor various paper defects in practice in order to make operators have a fast response to the cause of paper defects, decreasing the rate of the broke paper, increasing yield and reducing operators'workload.This paper focuses on image processing algorithms of paper defects and the experimental system design of web inspection, including system architecture design, image pretreatment, wavelet de-noising, on-line detection system for programming and so on. Research on paper defects image pretreatment algorithms includes three kinds of de-noising algorithm: histogram equalization of digital image processing, linear filtering, median filtering, adaptive filtering, frequency domain image processing, mathematical morphology and edge detection.Based on the algorithms study and according to the practical application desire and the laboratory of existing equipment, this paper designs an experimental system of paper defects online inspection, combining the function of image collecting, image pretreatment, image feature extraction with identification through using the platform of software Delphi, and gets the system tested and analyzed. Although having some achievement in experimental research, the method mentioned in this paper has a certain gap with the practice, due to some poor environmental conditions of the practical production field, such as high temperature, humidity, dust, vibration, accidental impact and so on, which would cause serious harm to the system. Finally, some suggestions to the deficiencies existed have been afforded for the next work.