Abstract: A recognition method of container number based on machine vision is studied. For the gray level method in the process of color image preprocessing of containers, the traditional grays cale algorithm can not effectively make up for the defilement or other missing information in the image. So, a hybrid method combining principal component analysis (PCA) with the gray change rate of Bayes threshold estimation is proposed to optimize the gray level of the image. It can make up for the missing information and effectively determine the edge features after judging the change rate of the gray value of a certain point in the image and the gray value of the neighboring pixel points, thus greatly improving the character recognition accuracy of the subsequent sequence. Finally, a set of intelligent detection system for port container is designed and implemented by using the algorithm model. Through the Matlab test for the identification of 50 port container number images, compared with the ordinary mean method and the weighted average method, the accuracy is better by using the mixed gray method proposed in this paper, and the accuracy rate of the single character can reach 96%, the accuracy rate of the container number recognition can reach 92%.

Key words: container, gray scale, principal component analysis(PCA), Bayes threshold estimation