In this paper, an online detection platform for jujube surface defects based on machine vision is designed to realize the automatic real-time measurement of full-surface information of jujubes. According to the characteristics of jujubes and their surface defects, the blob analysis algorithm in the color space was used to separate the jujubes from the background and identification of the surface defects of jujube model was proposed. The color space model and the segmentation threshold of different defect characteristics were given,the broken fruit, mold fruit, pulp head fruit, insect fruit and other typical jujubes surface defects identification was realized speedily and accurately . It shows that the results of this study are robust and reliable in this experiment, and the accuracy of defective identification is over 90%.