A new method for grading of silk yarn using electronic vision

The color of Tasar silk yarns is determined by a number of production factors, any slight variation in any one of these factors lead to variation in color of the yarn produced. At the present production technology, it is difficult to produce yarns of uniform color at the producers´ level, but once produced, those yarns can be sorted based on its color. The important characteristic of tasar silk yarn is its lustrous nature, it reflects light, thus difficult to ascertain the exact color manually. Slight variation in color is difficult to detect manually but the market demands lots with perfectly uniformly colored yarns within the lot though inter-lot variation in color is encouraged. So, Yarn separation based on the color is highly subjective and the process of manually separation of color is tedious and monotonous also. Also, it requires expert manpower, which may not be available in the remote villages in all cases. So, there is a need to develop an instrument, which can easily grade the yarns based on the color. This paper proposes automation of the silk yarn grading process by capturing images and classifying the silk yarns using digital image processing based color analysis technique thereby improving productivity and accuracy of this process. CIELCh color scale has been used for color analysis. Principle Component Analysis (PCA) shows the formation of inherent clusters in the image dataset. Color feature parameter based hierarchical grouping has been introduced here for silk yarn color grading. More than 2000 images have been analyzed using developed solution & the results have been validated with the human experts. Laboratory experiments found the overall accuracy of system in the tune of 91%.