基于Sobel算子的混合有理插值图像算法

Hybrid Rational Interpolation Image Scaling Based on Sobel Operator

应用传统的插值算法,在水平或垂直方向上不加区别地插入丢失的像素,会造成图像模糊、阻塞、振铃等伪像问题。为了解决此问题,提出基于Sobel算子的混合有理图像插值算法。运用该算法适应图像的变化边缘结构,在此基础上按照三次卷积插值流程建立插值曲面,再对图像进行下采样,实现图像的缩放。经实例验证,可以利用Sobel算子提取出图像的边缘特征,同时有效地对数字图像进行缩放,其峰值信噪比和结构相似性比传统算法有了较大提升。同时,运用该插值算法能有效地抑制图像缩放中的伪像问题,减少插值带来的噪声,基本满足图像插值过程中的噪声小、鲁棒性强、稳定可靠等要求。

In order to solve the problem of image blur,blockage,ringing or other artifacts caused by the traditional interpolation algorithm in the horizontal or vertical direction without inserting missing pixels,in this paper an edge-oriented rational interpolation based on sobel operator is proposed.The scheme can adapt to the changing edge structure of the image.On this basis,the interpolation convolution surface is established by the process of cubic convolution interpolation,and then the image is down sampled to realize image scaling.In this paper,the proposed algorithm performs actual image scaling and experimental verification.The experimental results show that the proposed algorithm can extract the edge features of the image by using the sobel operator,and effectively scale the digital image.Compared with the traditional algorithm,PSNR and SSIM are greatly improved.Experiments show that the interpolation algorithm can effectively suppress the artifacts in image scaling and reduce the noise caused by interpolation,which basically meets the requirements of small noise,robustness,stability and reliability in image interpolation.