. In order to implement the accuracy and robust classification of Chinese dates according to size and shape based on machine vision techniques on line, the method of classification according to size and shape for Chinese dates was studied. The Chinese date is segmented from the background in RGB color space by analyzing respectively the histogram of R, G and B channel to make comparing among them and find an optimal one, resulting in good contrast between Chinese date and background in G channel. Three different profile image of a Chinese date were gained to calculate the diameter of the Chinese date and the maximum diameter value was used to be the character value for size classification. The ratio of maximum radius and the minimum radius was used to be the character value for shape classification. Experiment results proved the methods is effective to classify Chinese date by size and shape.