Multispectral remote sensing image change detection based on Markovian fusion

This paper presents a novel multispectral remote sensing image change detection (CD) algorithm based on Markovian fusion. This new method intends to obtain the optimal change map (change detection result) by fusing information contained in each band. The optimal change map are modeled as Markov Random Fields (MRF) which takes into account not only the spectral information of multiple bands but also the contextual information of both the pixels in the optimal change map and the relationship between the optimal change map and change maps of each band respectively, and thus, leads to a more accurate and robust change detection result. In the analysis of difference image, an unsupervised threshold selection algorithm based on Bayesian decision theory is introduced, which aims at extracting the changed information from the images. The finding of optimal change map is equivalent to minimizing the total Gibbs potential function by using simulated annealing algorithm. The experimental result of the proposed algorithm compared with the change map of each band is presented, which indicates that the proposed method improves the result effectively and is superior to any band's change map.

[1]  Wei Wang,et al.  Image Change Detection Based on the Minimum Mean Square Error , 2012, 2012 Fifth International Joint Conference on Computational Sciences and Optimization.

[2]  Chein-I Chang,et al.  Constrained band selection for hyperspectral imagery , 2006, IEEE Transactions on Geoscience and Remote Sensing.

[3]  Pramod K. Varshney,et al.  An image change detection algorithm based on Markov random field models , 2002, IEEE Trans. Geosci. Remote. Sens..

[4]  Donald Geman,et al.  Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images , 1984 .

[5]  P. V. Oort.  Interpreting the change detection error matrix , 2007 .

[6]  Timothy A. Warner,et al.  Optimal band selection strategies for hyperspectral data sets , 1998, IGARSS '98. Sensing and Managing the Environment. 1998 IEEE International Geoscience and Remote Sensing. Symposium Proceedings. (Cat. No.98CH36174).

[7]  Russell G. Congalton,et al.  A review of assessing the accuracy of classifications of remotely sensed data , 1991 .

[8]  Adolfo Martínez Usó,et al.  Clustering-based multispectral band selection using mutual information , 2006, 18th International Conference on Pattern Recognition (ICPR'06).

[9]  Boli Xiong,et al.  A Threshold Selection Method Using Two SAR Change Detection Measures Based on the Markov Random Field Model , 2012, IEEE Geoscience and Remote Sensing Letters.

[10]  L. Bruzzone,et al.  A Bayesian approach to automatic change detection , 1999, IEEE 1999 International Geoscience and Remote Sensing Symposium. IGARSS'99 (Cat. No.99CH36293).

[11]  Farid Melgani,et al.  Markovian Fusion Approach to Robust Unsupervised Change Detection in Remotely Sensed Imagery , 2006, IEEE Geoscience and Remote Sensing Letters.

[12]  Paul L. Rosin.  Thresholding for change detection , 1998, Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271).

[13]  Wei Zhang,et al.  Change detection based on iterative invariant area histogram matching , 2011, 2011 19th International Conference on Geoinformatics.