基于特征的文档图像高速插值算法

Efficient interpolation based on document image texture

提出了一种基于文档图像特征的混合插值算法。该算法通过分析文档图像的像素邻域特征,使用邻近插值、双线性插值和双三次样条插值方法估计插值像素灰度值。其中,三次样条插值方法使用了优化算法,即三次样条卷积模板的整数化运算,其时间复杂度接近双线性插值。给出了算法的实现过程。实验结果表明,该算法输出的图像质量接近双三次样条插值,运行效率高于双线性插值算法,在文档图像扫描过程取得了良好的应用效果。

In this paper, the mixed interpolation method based on the document image texture is proposed. Through the analysis of document image pixel neighborhood characteristies, this algorithm uses neighboring-interpolation, bilinear interpolation and bi-eubic interpolation to estimate the interpolation pixel gray value. The bi-cubic spine interpolation method uses the optimization algorithm is that is the integer computing of bi-eubie spine convolution template. The time complexity of bi-eubie spine interpolation algorithm is close to the bilinear interpolation. In this paper, the implementation process of algorithm has been given. The experimental results show that the output image quality of the algorithm is close to the bi-eubic spine interpolation, and the operational efficiency is higher than that of the bilinear interpolation algorithm. In document image scanning process, the algorithm obtains a good application.