超分辨率图像的小特征增强方法研究与仿真

Study on the Method of Super-Resolution Image Little Feature Enhancement and Simulation

对超分辨率图像的小特征进行增强,可增强图像目标的边缘信息,改善图像的清晰度和分辨率。进行小特征增强时,需要根据图像帧之间的像素差设定图像增强增量,而传统多小波域算法在超高分辨率下无法获取图像帧之间的像素差,不能获取相关的增强增量,导致小特征增强效果差。提出改进小波变换的超分辨率图像小特征增强方法,先融合于Daubechies 8紧支撑正交小波对高分辨率图像的边缘信息进行去噪,利用张量扩散算法组建图像的各项异性张量扩散模型,将图像信息精细划分为光滑区域、边缘、和孤立噪声点,针对不同的分类设定相应的扩散张量的特征值,有效的完成了超分辨率图像的小特征增强。仿真结果证明,可以更有效增强图像的边缘和细节特征。

In the paper,we proposed a small feature enhancement method to improve the super resolution image of wavelet transform. First,we fused the method with Daubechies 8 compact support orthogonal wavelet to denoise high resolution image edge information. Then we used the tensor diffusion algorithm to establish an anisotropic tensor diffusion model,and meticulously divided the image information into smooth area,edge,and isolated noise points.According to different classifications,we set the corresponding diffusion tensor of characteristic values,and effectively completed the small feature enhancement of. super resolution images The simulation results show that the edge and detail features of image can be enhanced more effectively.