基于聚类与边缘检测的自然场景文本提取方法  

Based on clustering and edge detection of natural scenes text extraction method

作　　者：常莹[1] 何东健[2] 李宗儒[1] 

机构地区：[1]西北农林科技大学信息工程学院,陕西杨凌712100 [2]西北农林科技大学机械与电子工程学院,陕西杨凌712100

出　　处：《计算机工程与设计》2010年第18期4040-4043,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60975007)

摘　　要：为了解决复杂自然场景、光照不均匀及背景纹理丰富图像中文本的有效提取,提出一种基于K-means聚类与边缘检测结合的自然场景文本提取方法。该方法通过改进K-means聚类算法,实现文本区域的分割;然后对分割后的图像进行二值子图分解,将分解后的各子图像的连通区域进行标记与分析,得到候选的字符区域;最后利用文本区域的边缘特征对候选字符区域过滤,实现文本字符的提取。实验结果表明,该方法能有效提取出复杂背景、光照影响及背景纹理丰富图像中的文本字符区域。In order to efficiently extract text from complex natural scenes, uneven illumination and rich background textures, we proposed a natural scene text extraction method, which based on K-means clustering and edge detection. This method improved the K- means clustering algorithm, to achieve text regional segmentation, and then divided the value of the results into some black and white sub-images, marked the decomposition of the connectivity of the various sub-image regions and analyzed the characters to be the candidate regions, f＇mally, we used the text edge characteristics to filter the candidate character regions and achieved the extraction of text characters. The experimental results showed that this method can effectively extract the text regions from natural scenes which with complex back- ground, lighting effects and rich background textures.

关 键 词：自然场景 文本提取 聚类 连通区域标记与分析 边缘检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]