Computer Aided Drafting, Design and Manufacturing
Volume 25, Number 3, September 2015, Page 55
CADDM
Project Item: Supported by National Natural Science Foundation of China (Nos. 61373080, 61402261, 61303088, U1201258), Promotive Research Fund for
Excellent Young and Middle-aged Scientists of Shandong Province (Nos. BS2013DX039, BS2013DX048).
Corresponding author: ZHANG Yun-feng, Male, Ph.D., Professor, E-mail: yfzhang@sdufe.edu.cn.
Analysis of Multi-Scale Fractal Dimension for Image Interpolation
YAO Xun-xiang
1,2
, ZHANG Yun-feng
1,2
, LIU Geng
1,2
, BAO Fang-xun
3
, ZHANG Cai-ming
4
1. School of Computer Science & Technology, Shandong University of Finance and Economics, Jinan 250014, China;
2. Shandong Provincial Key Laboratories of Digital Media Technology, Jinan 250014, China;
3. School of Mathematics, Shandong University, Jinan 250100, China;
4. School of Computer Science and Technology, Shandong University, Jinan 250100, China.
Abstract: This article presents a novel image interpolation based on rational fractal function. The rational function has a simple and
explicit expression. At the same time, the fractal interpolation surface can be defined by proper parameters. In this paper, we used the
method of 'covering blanket' combined with multi-scale analysis; the threshold is selected based on the multi-scale analysis. Selecting
different parameters in the rational function model, the texture regions and smooth regions are interpolated by rational fractal
interpolation and rational interpolation respectively. Experimental results on benchmark test images demonstrate that the proposed
method achieves very competitive performance compared with the state-of-the-art interpolation algorithms, especially in image
details and texture features.
Key words: multi-scale analysis; fractal dimension; rational fractal interpolation; gradient
1 Introduction
Image interpolation is one of the key technologies
of image processing. It puts emphasis on the problem
of obtaining a high-resolution image from low-
resolution image. The nature of image interpolation is
estimating the unknown pixels form the known
adjacent pixels. In given conditions, image
interpolation can maintain sharp edge and vivid
texture, the high-resolution image obtained without
annoying artifacts and blurred edge. The image
interpolation is widely used in military, medical,
remote sensing, animation production, communica-
tions, aerospace, etc.
With the development of science and technology, it
is essential to monitor changes of the earth. Remote
sensing images are one of the major data resources for
obtaining changes of the earth. High-resolution remote
sensing images have been provided recently for
capturing detailed information, such as GF-1, GF-2,
etc. However, images are captured by satellite need to
be processed in detail for practical applications. Image
interpolation holds a peculiar position and a vital step
for features analysis to get useful information.
In recent years, many interpolation methods have
been developed for different purposes. The visual
quality of the remote sensing images is improved by
effective interpolation method. However, the
conventional interpolation methods failed to change
the situation, nearest neighbor interpolation, bicubic
interpolation, and bilinear interpolation included. In
reference [1], using a priori edge model, Carrato and
Tenze proposed an edge model to obtain optimization
the parameters of the interpolation operator. In
reference [2], by applying soft decision, an interpola-
tion that could compute a group of pixels at a time is
introduced. Hu and Tan
[3]
introduced the adaptive
osculatory rational interpolation for image processing.
In reference [4], the authors presented the comprehen-
sive update of changes in sensors and literatures.
In this paper, we propose a rational fractal model
for the remote sensing image interpolation. The
multi-scale fractal feature is used to decide the area
whether in texture area or not. At the same time,
nonlocal gradient information is in consideration. At
the end of the paper, our experimental results on test
images clearly demonstrate that the proposed method
outperforms the classical bi-cubic interpolator
[5-6]
, and
the recently developed image interpolation methods
[7-9]