Prediction of Tie-Dye Pattern Characteristics

We do a set of tie-dye experiments to study predictive method of tie-dye image pattern. As known to all, pattern information can be showed by color and texture, we choose the average value of HSV (hue, saturation, value) tri-component of the valid tie-dye area, the proportion of tie-dye white area and coarseness as image feature. According to linear regression analysis between two technological parameters (rotate speed and dye liquor concentration) and image feature, we found that there is obvious correlation between them, and the residual error of predicted models is quite low; the linear regression models can be used to predict tie-dyed image pattern effects.