一种边缘定向平滑图像插值算法

An Edge-directed Smoothness Algorithm for Image Interpolation

针对传统图像放大算法边缘处理效果较差,自适应图像插值方法存在高计算复杂度的问题,该文提出一种有效增强图像边缘轮廓的插值放大算法。该算法结合边缘定向平滑滤波器和双线性插值的特点,使得图像在平坦和非平坦区域均能取得理想效果。仿真测试结果表明,与基于统计特征的自适应插值算法相比,该文提出算法能显著提高插值速度,平均运行时间降低8.33 s;与双三次插值算法相比,图像峰值信噪比平均增加0.30 dB。

A novel effective image interpolation mechanism to enhance the enlarged image edges is proposed for the visual artifacts problem in traditional magnification methods and the high computational cost in some adaptive image interpolation methods. The proposed algorithm combines the edge-directed smoothness filter with the bilinear interpolation algorithm to obtain the higher quality of an interpolated image both in the edges and smooth areas. The simulation results show that the proposed algorithm gets a better visual effect. Compared with the edge-adaptive interpolation algorithm using covariance, the proposed algorithm is less complex with running time reduced 8.33 s on average and compared with the bi-cubic algorithm has higher quality with the PSNR is raised 0.30 dB on average.