A Pan-sharpening method for multispectral image with back propagation neural network and its parallel optimization based on Spark

Abstract

The Pan-sharpening method, which is used to address the fusion problem of multispectral (MS) images and panchromatic (PAN) images, has continuously been a hot spot in image fusion technology. In order to improve the quality and accuracy of the fused image, this paper proposes a Pan-sharpening method for MS image based on Back Propagation (BP) neural network, and further uses the Spark platform and the TensorFlowOnSpark (TFOS) framework to optimize the BP neural network. The experimental results show that the proposed method effectively enhances the quality of the fused image, and the parallel optimization method for BP neural network based on Spark improves the computational efficiency while ensuring the fusion accuracy.