基于K-means的图像文字识别与提取研究  

Research on the recognition and extraction of image characters based on K-means

作　　者：岳建杰 赵旦峰[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001

出　　处：《应用科技》2015年第2期17-20,共4页Applied Science and Technology

基　　金：黑龙江省科技攻关计划资助项目(GC12A305)

摘　　要：针对当前图像文字识别与提取的最新发展状况,提出一种基于K-means的图像文字识别与提取算法,其主要处理步骤包括图像的预处理、像素点聚类处理、图层的选择与优化以及最终的文字切分等。经过上机对算法进行测试,该算法能够有效提高图像文字识别与提取的准确率与执行效率,并可以针对不同环境下的图片文字进行有效识别。Based on the latest developments in the recognition and extraction of image characters, this paper propo- ses a K-means-based algorithm for the recognition and extraction of image text. Its main processing steps include im- age preprocessing, pixel clustering process, selection and optimization of the layer and the final text segmentation. By computer testing, it was found that the algorithm can effectively improve the accuracy and efficiency of recogni- tion and extraction of image characters. In addition, it can effectively identify image characters under different envi- ronments.

关 键 词：K—means 图像文字提取 文字识别 二值化处理 

分 类 号：TN971.1[电子电信—信号与信息处理]