计算机应用 ›› 2012, Vol. 32 ›› Issue (10): 2821-2823.DOI: 10.3724/SP.J.1087.2012.02821

• 图形图像处理 • 上一篇    下一篇

基于梯度的快速图像插值算法

杨鹤猛1,2,黄战华1,2   

  1. 1. 光电信息技术教育部重点实验室,天津 300072
    2. 天津大学 精密仪器与光电子工程学院,天津 300072
  • 收稿日期:2012-04-09 修回日期:2012-05-23 发布日期:2012-10-23 出版日期:2012-10-01
  • 通讯作者: 黄战华
  • 作者简介:杨鹤猛(1985-),男,河北辛集人,博士研究生,主要研究方向:红外视频成像系统及其处理算法;黄战华(1965-),男,湖北孝感人,教授,博士生导师,博士,主要研究方向:光电信息检测、图像处理。
  • 基金资助:
    国家863计划项目(2008AA01Z407)

Fast image interpolation algorithm based on gradient

YANG He-meng1,2,HUANG Zhan-hua1,2   

  1. 1. Key Laboratory of Optic-Electronics Information Technical Science of Education Ministry,Tianjin 300072, China
    2. School of Precision Instrument and Opto-electronics Engineering,Tianjin University,Tianjin 300072, China
  • Received:2012-04-09 Revised:2012-05-23 Online:2012-10-23 Published:2012-10-01
  • Contact: HUANG Zhan-hua

摘要: 为解决传统插值算法效果不佳而新发展的边缘方向插值算法计算量大的问题,提出了一种基于梯度检测边缘的快速图像插值算法。该算法通过在原图像每一个3×3邻域内计算二阶梯度,检测边缘是否存在:若存在边缘,则计算一阶梯度判断边缘强弱并自适应得到权值因子,然后进行带权的线性插值;若不存在边缘,则在此邻域内进行双线性插值。实验结果表明,此算法不仅使插值后图像具备整体清晰性和边缘平滑性,有效保持了图像的主观视觉质量,而且相比现有同等效果的插值算法,运算时间至少能降低20%,尤其适用于嵌入式设备图像放大的应用。

关键词: 图像放大, 快速插值, 带权插值, 边缘检测, 二阶梯度

Abstract: Traditional image interpolation algorithms have poor performance and most of the recently developed edge-directed interpolation algorithms are of high computation. To solve those problems, a fast image interpolation algorithm based on gradient distribution detection was proposed. The basic idea was to detect the existence of edges in each 3×3 neighbourhood of the original low-resolution image by calculating its two-step gradients. If an edge existed, adaptive weighted factor up to edge strength was obtained by calculating its one-step gradient, and then the weighted linear interpolation was implemented. Otherwise, bilinear interpolation was carried through in that neighbourhood. The experiments show that the proposed algorithm is able to not only keep the subjective visual quality of the interpolated high-resolution image with whole clarity and edge smoothness, but also reduce the operation time by at least 20% compared with the equivalent algorithms now available. It contributes to an image enlargement application in embedded systems.

Key words: image enlargement, fast interpolation, weighted interpolation, edge detection, two-step gradient