Practical method of shadow detection and removal for high spatial resolution remote sensing image

摘要

High spatial resolution remote sensing image (HSRRSI) has received a warm welcome in many fields. However, building shadows of large area on HSRRSI (up to 30% in some cases) are one of the biggest hindrances for further applications in many fields. To keep a balance between precision and efficiency required by applications during shadow removal, this paper introduces a creative and practical strategy based on the theory of the pulse coupled neural network (PCNN). By applying the simplified model of PCNN, shadows on HSRRSI had been detected and removed respectively. When applied to HSRRSI, the method could not only remove the shadows, but also keep the contrast between removed areas with shadows and other areas without shadows from being too big, which might distort the image. Therefore the satisfactory result is gained.