A novel remote sensing image change detection algorithm based on self-organizing feature map neural network model

Image change detection is based on the analysis in different time from the same area of two or more images, detect the feature in the region information changes over time. A self-organizing map integrated with a two layer neural network is implemented in this paper where the two input SAR images obtained at two different time instants are subjected to differencing and thresholding and weights are updated to converge the neural learning process to a minimum error value. Observed results from experimentations conducted on two sets of SAR images report a good accuracy in event detection with satisfactory image visual quality. The input images utilized in this paper and the event change recorded in this work could be applied to urban and vegetated land registration to indicate the change of terrain over a period of time. This might be utilized in urban planning applications. The work has been compared with fuzzy based techniques and a reduced computation time is also reported in this paper.