The Design of the Manchu Character Recognition System

Character recognition is an important part in pattern recognition and artificial intelligence area. In order to recognize Manchu character, it is a must to combine image processing techniques and pattern classification together. Based on the particularity of the Manchu, we raise an arithmetic using K classifier. First of all, the original Manchu images are obtained by scanning the texts through the camera or the scanner, and then preprocess them by binarization, denoising and skew emendation. In the segmentation, the paper presents an arithmetic taking the projection of the character into consideration. In this system, we take the Manchu character's outline as its feature and calculate the Euclidean distance between the test image and all of the training images and classify its class by the main category the k nearest neighbors belong to.