印刷体维吾尔文字符切分自适应算法

摘要

To research and develop the webcam taking and translating Uighur word system and to solve the difficulties of Uighur word letter segmentation in the system ,a new algorithm based on the adaptive threshold was proposed .Aiming at the features of the Uighur word pictures taken by the webcam ,the word was exacted accurately first .Then the main part of the word was used to make the projection histogram of pixels ,the threshold of segmentation was extracted automatically .The threshold was used in segmentation at last .Experiments verify the segmentation accuracy rate of this method reaches over 96% .At the same time ,it has better adaptability for different images ,and it can promote the study of Uighur word camera translation system .%为研究开发维吾尔文摄像头取词翻译系统，解决其中维吾尔文字单词图像切分难题，提出一种印刷体维吾尔文字符自适应切分算法。针对摄像头取词图像特点，准确提取目标单词；利用维吾尔文单词基线以上的主体部分做像素积分投影，从投影结果中自动提取切分阈值；利用该阈值完成字符切分，达到自适应的效果。经过实验验证，该方法切分正确率达到了96％以上，针对不同图像具有较好的适应性，对维吾尔文摄像头取词翻译系统的研究具有促进作用。