Wool Measurement Based On Digital Image Processing And Classification System

Measurement of the diameter and curvature of wool plays a very important role in wool measurement, classification and sheep breeding. Technicians usually measured wool's diameter and curvature by hand, which is a waste of time and energy,and the error result is produced by even professional workers. As a result,it's stringent to make a wool measure and classifying system based on digital image processing technology.In this paper, an automatic wool measure and classifying system is designed and implemented that includes optic system, image disposal system, and a method to diameter measurement and classification based on image disposal. In this method, the image processing and analysis is fully used. At first, this method enhances the captured images by image gray modification, image flatness and image sharpening. We modify images by vertical square chart and flat images by K-neighbor median filter in detail. Secondly, Canny operator is used to detect the edges of wool. And then, the binary images are got by clustering based on Gaussian distribution and a session of processes are operated to the binary images so as to extract the wools'skeletons conveniently. Next, the value of diameter by Euclid distance transformation. At last, we use the Bayesian Classifier to classify wool images based on both diameter and curvature characters.Digital image processing technology can greatly ease the intensity of manual measurement, and improve accuracy and efficiency. If we had used the method based on the linear-fitting, we didn't only encounter a great amount of computation, but also introduced additional handling errors, including fitting error. The more errors happen, the more fibers are included in an image. In this paper, we use a new method based on Euclidean distance transform and 8-neighborhood searching algorithm, and the result is more accurate and faster. With the two characters, we classify wools by the Bayesian Classifier that is based on minimum error rate. The system is more convenient, rapid and accurate to measure the diameter and classify a batch of wool images.