多源异质遥感影像的分形特征分析

分形理论已经在影像压缩、分割、去噪等方面表现出了较大的优势,其特点是能够准确地描述影像的空间结构信息和变化规律。分别采用分线法和三角棱柱法对同一实验地区的多种分辨率不同波段的遥感卫星影像进行分析,分别计算了图像中不同地形区域(如城区、湖泊、田地等)的分形维数,并根据两种算法的特性对图像进行了特征分析。多源遥感影像的实验结果证明,分形维数确实可以作为影像特征分析的一个重要工具,并可以指导分析遥感影像的应用情况。此外,在计算分形维数的过程中,采用三折线拟合法自适应地确定影像的无标度区间,提高了分形维数计算结果的可靠性。

Fractal Characteristic Analysis for Multi-resources Heterogeneous Remote Sensing Images

The excellent characteristic of fractal theory was the ability to record spatial structure and change rule of images accurately,and it has great advantages in image compression,image segmentation and image denoising.In this paper the line-divider and the triangular prism algorithms were used to analyze the different spatial and different spectral resolution remote sensing images for the same region.Fractal dimensions values of different objects(e.g.city,lake,farmland) were computed and used to analyze characteristic of the images based on advantages of two algorithms.Experiment results with multi-resource remote sensing images proved that fractal dimensions was a superior tool for analyzing characteristic of the images and applications of various remote sensing images.Furthermore the three lines fitting algorithm was used to auto-ascertain the no-scale part during computing fractal dimensions,and it was demonstrated to improve the reliability of fractal dimensions.