Optical Image Change Detection Method Based On Depth Convolution Difference Map Fusion

Abstract Aiming at the problem of low change detection accuracy caused by difference in brightness optical remote sensing images, an This method borrows idea convolution learning features deep learning, performs operations on two and obtains convolutional map fusion. logarithmic ratio map; then multi-layer maps respectively. The feature after each layer are combined to obtain a fused fusion finally maximum article uses LANDSAT satellite images verify compares variance automatic threshold between classes is used for segmentation result graph. experimental results show that enhances detailed information In terms effect, anti-noise performance enhanced. improved.