Map Difference Indication Using CNN Based Deep Learning Algorithm

There exist several image change detection model but these approaches are not efficient in terms of change detection due to its accuracy,performance and latency.The time complexity of such approaches are too high and the accuracy of prediction is too low.In order to provide solution in change detection in a map at different dates,so we proposed a novel hybrid CNN model based the combination of LSTM with CNN.The model takes images as input and transforms to grayscale in order to extract the feature,create a matrix based on the pixel value.Moreover,the proposed framework uses Euclidian distance based on the image pixels to provide the image change detection in a map,the proposed model trains the with dataset available publicly especially we used Argentina historical map from1994 and 2014 and other public maps as well in order to assess our model.Moreover,the proposed model is based an efficient map change detection framework in identifying places that have changed on a map of the same area when viewed at two distinct times.In order to justify the proposed framework,we carried out comparative analysis through simulations and provide justification that the proposed framework outperform the benchmark models in terms of change detection accuracy as well as overall performance.The benchmark model was considering as polygon and the object-based detection integrated with CNN.