基于定向拉普拉斯模型的图像放大方法

摘要

针对传统插值及PDE模型放大方法的缺点,提出一种具有边缘增强作用的基于定向拉普拉斯模型的图像放大方法.该方法以传统插值放大方法为初值,然后经边缘增强的定向拉普拉斯PDE方程反复迭代的扩散作用实现图像放大.实验结果表明,该算法是一种有效的保持边缘和细节的图像放大方法,在客观指标和主观效果上均优于传统双线性插值和P-M方程的方法,是一种有效的图像放大算法.%To overcome the shortcomings of conventional interpolation and partial differential equation (PDE) methods,an image zooming method with edge-enhancing function based on oriented Laplacian operator is introduced. Starting from the interpolated image, the proposed PDE equation is resolved iteratively, and the zoomed image is obtained until the process is terminated. Experiments demonstrate that the proposed zooming approach can achieve better results such as edge-preserving and details-preserving than traditional bilinear interpolation method and P-M equation.