The proposed Content-Adaptive Non-Local Convolution (CANConv) module simultaneously incorporates spatial adaptability and non-local self-similarity to enhance the performance of remote sensing pansharpening.
