Integration of GMRF Model and HTNN Network for Change Detection

In this work, a spatiocontextual unsupervised change detection technique for multitemporal, multispectral remote sensing images is proposed. The technique uses a Gibbs Markov random field (GMRF) to model the spatial regularity between the neighboring pixels of the multitemporal difference image. The difference image is generated by change vector analysis applied to images acquired on the same geographical area at different times. The change detection problem is solved using the maximum a posteriori probability (MAP) estimation principle. The MAP estimator of the GMRF used to model the difference image is exponential in nature, thus a modified Hopfield type neural network (HTNN) is exploited for estimating the MAP. In the considered Hopfield type network, a single neuron is assigned to each pixel of the difference image and is assumed to be connected only to its neighbors. Initial values of the neurons are set by histogram thresholding. An expectation–maximization algorithm is used to estimate the GMRF model parameters.

Publication:

1. A. Ghosh, B. N. Subudhi and L. Bruzzone, "Integration of Gibbs Markov Random Field and Hopfield-Type Neural Networks for Unsupervised Change Detection in Remotely Sensed Multitemporal Images", IEEE Transactions on Image Processing, vol. 22, no. 8, pp. 3087-3095, 2013. (pdf)

2. B. N. Subudhi, S. Ghosh and A. Ghosh, “Spatial Constraint Hopfield-Type Neural Networks for Detecting Changes in Remotely Sensed Multitemporal Images” Proceedings of 20th International Conference on Image Processing (ICIP-2013) (Published by IEEE Computer Society), pp. 3815-3819, Melbourne, Australia, 2013. (pdf)

Integration of Fuzzy Incorporated MRF Model for Change Detection

In this work, a novel spatio-contextual fuzzy clustering algorithm for unsupervised change detection from multispectral and multitemporal remote sensing images is proposed. The proposed technique uses fuzzy Gibbs Markov Random Field (GMRF) to model the spatial gray level attributes of the multispectral difference image. The change detection problem is solved using the Maximum a posteriori probability (MAP) estimation principle. The MAP estimator of the fuzzy GMRF modeled difference image is found to be exponential in nature. A convergence by conventional fuzzy clustering based search criterion is more likely to lead the clustering solutions to stuck to a local minimum. Hence we adhered to the variable neighborhood searching (VNS) based global convergence criterion for iterative estimation of the fuzzy GMRF parameters. Experiments are carried out on different multispectral and multitemporal remote sensing images. Results confirm the effectiveness of the proposed technique. It is also noticed that the proposed scheme provides better results with less misclassification error as compared to the existing techniques.

1. B. N. Subudhi, F. Bovolo, A. Ghosh and L. Bruzzone, "Spatio-contextual Fuzzy Clustering with Markov Random Field Model for Change Detection in Remotely Sensed Images", Optics & Laser Technology, vol. 57, pp. 284-292, 2014. (pdf)