The Study Of Computer Image Recognition Method Of Cotton And Ramie Fiber

Textile exports need to go through a very strict the detection. As the largestcountry of textile production and export, the textiles fiber types identify is mainly torely on artificial conduct detection. It has low work efficiency but also easily affectedby subjective factors influence and cause false identification. So the use of computerautomatic identification of the fiber is an inevitable trend.The main content of this paper is the computer automatic identification of cottonand ramie fiber image. Propose the overall design of the system. To amplifier andcollected fiber by using a microscope. Research the fiber processing and recognitionalgorithms.The mainly jobs in this paper are the angle of the fiber image correction, theouter contour extraction and repair fiber characteristic parameter extraction andanalysis, the ramie fiber cross section, and the mention of the crack patternrecognition theory.In the processing of fiber image angle correction, the paper attempts to two anglecorrection methods. There are one step in the overall correction and graduallycorrected. Compared the advantages and disadvantages of the two algorithms andcontact the actual requirements of real-time operation of the line system to select themore suitable for this algorithm.A creative method of extracting the outline of an image is extracted contourfracture repair work. And we also do something to contour fracture repair work.In the part of characteristic parameter extraction, we first extract distortion、fiber diameter and so on seven characteristic parameters, and then successively dosestudy correlation and excluded wherein a low correlation parameters.Finally, after a comprehensive comparison, the neural network theorysuccessfully applied to the fiber pattern recognition in this paper, we electedparameters input to the neural network and training it, then we tested the net byactual sample. Eventually taken by the front this method is feasible. And ultimatelyadopted the method in this paper is feasible.