Contourlet域分形编码的图像插值算法  

Image Interpolation Based on Fractal Encoding in Contourlet Domain

作　　者：应毓海[1] 

机构地区：[1]安徽广播影视职业技术学院信息工程系,合肥230011

出　　处：《合肥学院学报（自然科学版）》2016年第1期35-40,共6页Journal of Hefei University :Natural Sciences

基　　金：安徽省教育厅科学技术研究重点项目(kj2014A033)资助

摘　　要：自然图像具有分形局部自相似结构,不同区域不同尺寸的图像块之间存在相似关系。对图像作contourlet变换,根据图像的contourlet域分形编码确定子树与父树之间的变换参数,建立相邻尺度不同区域子带系数间的变换关系,由已知子带对未知的高频子带进行恢复,经contourlet逆变换得到高分辨率插值图像。实验表明,该算法能够对图像的结构细节实现准确有效的恢复,具有较高的插值精度和图像质量。Local self-similarities exist in nature images widely. Image blocks with different sizes in dif- ferent areas are similar to each other. In this paper, such main research contents are conducted：Decompose the image using contourlet transform, calculate the parameters involves in the transform relation between child tree and parent tree based on the fractal encoding in contourlet domain, and establish the transform relation connecting different areas in adjacent subbands. With the help of relations between adjacent subbands, a restoration algorithm of the unknown high frequency subband using known subbands is studied and the high-resolution image is obtained by performing the inverse contourlet transform. Experimental resuhs show that the proposed interpolation algorithm can effectively recover the structural details of images, and as a result high interpolation accuracy and image quality are achieved.

关 键 词：图像插值 分形 局部自相似 CONTOURLET变换 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]