Remote sensing change detection is a hot issue in recent years. However, most methods originate from statistical pattern recognition. Parameter resolution and time-consuming are main disadvantages of these methods. Hence, in this paper we propose a novel remote sensing change detection method which originates from neural network pattern recognition. The method is based on Growing Hierarchical Self Organization Map(GHSOM). GHSOM has flexible network architecture to adjust remote sensing scene complexity. Theoretically speaking, GHSOM is able to extract change areas well. In experiment, we select three pairs of remote sensing image. We compare the results with Gaussian Mixture Model result and traditional SOFM result. The experiment shows the proposed method is advantageous in efficiency and detection accuracy. It can be expected that the method will be applied in GIS data updating, land use cover surveying, and natural disaster evaluation. Keywords-remote snesing; chang detection; Growing Hierarchical Self-organization Mapping; Differential Image
[1]
Andreas Rauber,et al.
The growing hierarchical self-organizing map: exploratory analysis of high-dimensional data
,
2002,
IEEE Trans. Neural Networks.
[2]
C. Ji,et al.
Land-Use Classification of Remotely Sensed Data Using Kohonen Self-organizing Feature Map Neural Networks
,
2006
.
[3]
Paul L. Rosin.
Thresholding for change detection
,
1998,
Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271).
[4]
Lorenzo Bruzzone,et al.
An unsupervised approach based on the generalized Gaussian model to automatic change detection in multitemporal SAR images
,
2005,
IEEE Transactions on Geoscience and Remote Sensing.
[5]
C. Ji.
Land-use classification of remotely sensed data using Kohonen Self-Organizing Feature Map neural networks.
,
2000
.
[6]
Diego Andina,et al.
Application of Neural Networks
,
2007
.