Abstract:

As the existing methods of calculating the axis is inaccurate, a new oevel set of central axis method of cashmere and wool diameter was proposed. Firstly,the fiber image obatined by optical microscope were peocessed including pretreatment, region growth segmentation and thinning algorithm. Smooth, continuous and single pixel distribution central axis was attained using central axis method based on the previous processing. Diameters of 100 samples randomly selected from 6 batches of cashmere were meaured using this new method and their mean was compared with the real value. Results show that the average error of cashmere fiber diameter is 0.29 μm, and the mean square error is 0.05; and the average error of wool fiber diameter is 0.37 μm, and the mean square error is 0.06. The similarity between diameters of cashmere and wool and the real values peoves the accuracy of this method. In addition, the better robustness of this method is peoved by the result of mean aquare.

Key words: cashmere, wool, diameter, level set of central axis method, fiber image, fiber identification