基于K-means颜色聚类分割与边缘检测的文字提取

Based on K-means Color Clustering Segmentation and Edge Detection of Text Extraction

针对自然场景中文字提取受复杂环境因素的影响,如光照不均匀、自然场景背景颜色多样等因素影响,采用任何单一的图像分割技术都无法进行有效地进行文字区域分割和文字提取,提出一种两种方法相结合的自然环境场景中的文字提取方法。首先,采用实现颜色聚类的K-means算法对文本区域与有颜色背景分割,然后在文本区域内对文字进行二值化处理后运用边缘检测的方法提取文字。通过VC＋＋编程环境及OpenCV技术作为该方法的验证平台,结果显示基于Kmeans算法实现颜色聚类与边缘检测方法相结合能有效进行自然环境中文字的提取。

According to the natural scene text extraction is influenced by complex environmental factors,such as uneven illumination,natural scene background colors and other factors,any single method is unable to effectively carry out the text segmentation and text extraction,a method of combining two methods of regional segmentation and text extraction is proposed.Firstly,the Kmeans algorithm is used to segment the text area and the colored background,then the text is processed to white and black value in the text area,and then the edge detection method is used to extract the text.Through the VC＋＋ programming environment and OpenCV technology as the verification platform of the method,the results show that the color clustering and edge detection Based on K-means algorithm can effectively extract the text in the natural environment.