A method for assessing norms of Chinese character based on K-means clustering and hough transformation

In this paper, we process the image of Chinese characters, and assess the regular of Chinese characters in the image. Images of Chinese handwritten characters is processed to remove the irrelevant information such as background color, background noise. After processing and Hough transforming the image, we can get the information of strokes and angle through the outline of the image. After Hough transforming, the lengths and angles of the Hough Lines can be obtained and they are the approximate equivalent to the length and angle of strokes of Chinese characters themselves. Further analyzing the length and angles of the Hough Lines, the structure of Hough Lines can be equivalent to the structure of Chinese character strokes. Using the K-means algorithm of Clustering analysis, we grouped the degree of angles into different clusters. Because of Chinese characters strokes have vertical and horizontal features, we just need analyze the clusters near 0°, 90°, 180°, 45°, 135°. Finally,we use the length of the Hough lines as weights to assess norms of the Chinese characters.