The information of position and attitude of unmanned aerial vehicle (UAV) during autonomous landing is calculated based on runway images by UAV visual navigation system, which provides a basis for flight control systems to perform the function of autonomous landing control. The calculations of image processing, position and attitude estimating in visual navigation system are analyzed and researched. Template matching mode is used to recognize the runway. Image characteristics of runway are extracted by Hough transform. The least square method is used to merger the detected edge lines of runway and their equations in the image coordinate system are achieved. Based on coordination relations in space, the algorithm is derived to calculate the position and attitude of UAV. Real images are used to verify all the algorithms in experiments.