Color Regression and Sorting System of Solid Wood Floor

Solid wood floors are very common in interior decoration, and their color is an important indicator of product quality, selected order to achieve the overall aesthetic needed ensure consistency. In realize sorting solid based on depth, so that colors could be freely graded, one image acquisition system was built collect 108 floor images a set fast methods for depth developed. Among these, 10 were used as test therefore not sorted, 98 sorted by depth. 80 original expanded 13 times 1040, use training set, 18 validation set. The characteristics RGB, HSV Lab space extracted, LightGBM floors. At same time, two deep learning algorithms, Vision Transformer well Densenet121, improved means adaptive pooling layer, different sizes. final ranking results showed method using regress features exhibited most harmonious results.