摘要: In order to improve the purity of tobacco materials, machine vision technology combined with Bayesian classification algorithm were applied to online detection and separation of greenish tobacco strips. The characteristic distributions of chrominance indicator in Lab color space were obtained for greenish and normal cured tobacco strips. Bayesian classification and decision algorithm were applied in the recognition of greenish tobacco strips. Verification experiments were performed by the designed machine vision platform. The results showed that the recognition rate for greenish tobacco strips was more than 93% when pure greenish tobacco strips were used, and the average elimination rate reached 90.7%. When greenish tobacco strips were mixed with normal cured tobacco strips, the average elimination rate was still up to 90.5%. The present work indicated the feasibility of this method for the detection and separation of greenish tobacco strips during tobacco processing.