Publications

Shen, YL; Liu, XG; Yuan, XH (2017). Fractal Dimension of Irregular Region of Interest Application to Corn Phenology Characterization. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 10(4), 1402-1412.

Abstract
Analysis of multitemporal remote sensing imagery offers a reliable and cost-effective means for monitoring crops on a broad-scale and provides consistent temporal measurements. Fractal geometry has been used as a quantitative description of spatial complexity of remote sensing images. Yet, corn field of spatial irregularity alters the fractal dimension of the landscape, which shall be suppressed in the estimation. In this paper, we propose a method for computing fractal dimension from irregular region of interests that minimizes the contribution from 2-D spatial irregularity. Our method was evaluated with normalized difference vegetation index products derived from moderate resolution imaging spectroradiometer and satellite pour l'observation de la terre VEGETATION sensors from three states in the U.S. The experimental results using the time series demonstrated that our proposed fractal dimension estimation method exhibited great consistency and invariance to the change of image spectral characteristics, spatial resolution, and the degree of pixel mixing. In contrast to entropy and variance, the spectral characteristics of different imaging devices exhibited lower impact to the fractal dimension, which also implies its scale invariance. With respect to the detection rate of the first peak, fractal dimension achieved the best consistency. The proposed method for computing fractal dimension provides a critical and reliable measure for studying phenological patterns.

DOI:
10.1109/JSTARS.2016.2645880

ISSN:
1939-1404