Skip to Main Content
In order to solve the problem that current license plate recognition methods, such as template matching and neural network computing, which need a large number of samples and large amount of computation, this paper proposed a sub-image fast independent component analysis (SI-FastICA) method for plate recognition. It can obtain the local feature of the image with a small amount of computation. In order to obtain better recognition results, in the stage of character segmentation, this paper carried segmentation based on the proposed relative coordinate dichotomy. Then, the feature of characters was extracted by SI-FastICA. The experiments show that SI-FastICA can reflect the local characteristics of the character very well. At last, this paper put the collected actual license plate images into experiment, and achieved good recognition results.