In view of the draw backs of jujube grade identification in China, which still relies on photoelectric sorting and manual separation, this paper presents a processing method on the basis of the technology of computer vision and digital image. Utilizing image processing technology, the researcher extracted the red mean (R), green mean (G), blue mean (B) and their mean square error R G B s s s 、 、 and a total of 6 kinds of color characteristic variables from jujube image after pretreatment. Further made the image color spatial switching from RGB to HIS, then for the HIS color space, extracting the hue mean (H), intensity mean (I), saturation mean (S) and their mean square error and the total of 6 kinds of color characteristic variables, the total of 12 color characteristic variables, as the key characteristics of the BP input of network to build a network and identify the level of jujube through analysis of the external characteristics of jujube. The optimum structure parameters of the BP neural network which had 9 hidden layer neurons were determined by RP training algorithm. Results showed that average accuracy for fruit classification can reach 92.5% by using this model，and the executing time of microcomputer for grading of one jujube is 9.3 ms．This method has the characteristics of high accuracy and good real-time performance．