The construction of Latin based European OCR system is studied in this paper. Compared with English, other Latin based European languages use more characters, which is called European special characters in this paper to be distinct from English letters. To construct a European system with high performance, the key is the recognition of the European special characters. In this paper, the European special characters are automatically divided into three subsets by the different handwritten position. And two solutions are proposed, one solution in which is used to recognize "i", "j" and the European special characters in subset 1, while another solution is used to recognize other English characters, digits and the European special character in other subsets. Experiment shows, the new system is more effective than the old one, which provides an experimental support for our research work.