A Computer Vision-Based Classification Method for Pearl Quality Assessment

Pearl´s color is an important feature to assess its value, including the hue and its color depth. A method for pearl color classification was investigated in this paper. Computer Vision is used to process the pearl image after transforming it from RGB to HSV color model, which can show the hue and color depth information of pearl. According to the histogram of V (Value) weight, the bright area is extracted by Ostu Segmentation and the average value of H (Hue) and S (saturation) are obtained. Aiming at the standards of hue classification, the artificial neural network method based on RPROP Algorithm is adopted; Aiming at the color depth`s difference, Fuzzy C-means Clustering Algorithm is adopted to classify the average value of S. The proposed method can be used for the first classification according to the surface color of pearl and further classification according to the saturation of pearl in the same color series and realizing the standard classification of pearl quality.