Changes detection of remote sensing images using two-stream two-stream deep neural network

Identification of changes in remote sensing images plays an important role in many applications, such as monitoring the growth of urbanization, land use changes, and assessing disasters and natural damages. This process aims to assign the label "changed" or "not changed" to the pixels of two images taken from the same place but at two different times. On the other hand, in the last decade, deep learning methods have attracted the attention of many researchers in this field due to their proper performance in interpreting and processing remote sensing data and the ability to remove feature engineering and extract high-level features from images. In this regard, in this article, an optimal deep learning model has been designed, which increases the accuracy of identifying changes in two-time images due to its hierarchical structure, appropriate efficiency of multiscale features, and effective design of feature transfer. Due to the optimal structure and architecture, the proposed model has higher speed and accuracy of results compared to some popular models such as BIT. Applying the proposed model to the two data sets under investigation indicates an average accuracy of 96% and less complex calculations.