Determination Of Yarn Blended Ratio By Image Analysis

It's of great practical value to determine textile yarn blended ratio by image analysis on their cross sections. This article deals with the details in the four aspects—image preprocessing, individual detection, characteristics extraction and statistic analysis—which compose the wholeness of this research.On image preprocessing, a general solution based on mathematical morphology is advanced to make possible batch operations on images.On individual detection, a new model, Facula Diffusion, is put forward. Compared with the traditional Contour Tracing arithmetic, the new model works on gray-scale images rather than binary images, thus making the image binarization which always tends to cause considerable signal loss no longer necessary. Moreover, the new model has an intelligence that is far beyond the capability of the old arithmetic, especially on treating with fusional individuals.On characteristics extraction, three new shape indices are advanced from the angle of"span". They are the Acreage index, which describes the acreage that the individual contour closes in, the Abnormity index, which describes the extent of departure from circularity, and the Fluctuation index, which describes the roughness of the contour. How to select shape indices to form feature vectors is also discussed.On statistic analysis, all feature vectors are clustered and classified by means of cluster analysis. The yarn blended ratio in volume is computed by accumulating individual acreage in each category, and the blended ratio in quality can be figured out given blended ratio in volume and the densities of each category of fiber. Finally, a computer software called Blend Yarn Analyzer is developed. It is a complete realization of the theories discussed in this article, with extensions in some aspects for the purpose of practicability.