基于二值化聚类的图像文字提取算法

为解决渐变色给文字提取聚类算法带来的问题，研究与实现了基于二值化聚类的图像文字提取算法。图像通过一系列预处理后，得到了利于聚类的二值图像，根据背景图像区域特征，对图像进行聚类分块，再利用文字图像区域特征，聚类识别出文字区域。实验表明，该算法在各类图像上取得了理想的效果。

Text extraction algorithm based on binary clustering

To deal with the gradient problem in the clustering process of text extraction, an algorithm based on binary clustering was proposed. The original image was converted to binary bitmap after preprocessing. The background blocks of the image were clustered by the region features, and then text blocks were recognized by the distribution features. The experiment shows this method achieves satisfactory result on various kinds of images.