Abstract
UAV autonomous landing refers to the UAV lands on only depending on airborne navigation equipment and flight control system, and ultimately achieves a safe landing. To achieve self-landing, UAV must have the ability to self-navigation and positioning, so that the high-precision visual navigation positioning technology is the key to achieve UAV self-landing technology.This paper concentrating on the noise effects on the pictures obtained during visual landing process of UAV, it has introduced the gravity of classical physics to image pixel, come up with a mathematical expression for the strength of gravity between pixels, then conformed the adaptive window by the gravity between pixels and performed corresponding filtering processing. It is shown by the experimental results that this algorithm has a great improvement on image denoising and detail preserving when compared with the traditional median filtering and switching median filtering algorithms.