基于纹理方向的运动图像插值算法

为解决插值过程中图像纹理方向的马赛克效应和结构信息丢失问题，设计纹理结构驱动的运动图像插值算法。通过有限Radon变换提取图像的纹理方向信息，对随机纹理的区域采用传统插值方法；对规则纹理的区域，根据纹理方向与待插值点的位置分别设计不同的插值滤波器。实验结果表明，与常规方法相比，采用该插值方法可以得到更高的图像质量。

Moving Picture Interpolation Algorithm Based on Texture Direction

To conquer structure information loss and mosaic effect at picture texture direction caused during the process of interpolation, a texture structure driven moving picture interpolation algorithm is proposed. It analyzes picture edges and texture information by performing Finite Radon Transform(FRAT), applies traditional interpolation methods to random texture areas, and designs different interpolation filter for areas with identified texture based on pixel position and texture direction derived before. Experimental results show that this strategy can increase image quality compared to usual ways.