关键词: 无人机航拍序列图像, 自动拼接, 卡尔曼滤波

Abstract: The adjacent frames of the unmanned aerial vehicle images contain much overlap areas.These areas are the basis of the image mosaic,but also decrease the efficient of the mosaic.This paper proposed a fast and stable automatic mosaic method on unmanned aerial vehicle images.A new mosaic-oriented frame selection method by using the feature points which were found in the process of the mosaic to select the most feasible frames for mosaic was proposed; based on analysis of the movement model of the overlap areas in the images,a Kalman filter for predicting the search area of feature points was proposed to reduce the time-cost in the process of searching and matching features.And then,the detail steps of the algorithm were presented.The experimental results showed that the proposed algorithm can finish the seamless image mosaic with low time-cost.

Key words: unmanned aerial vehicle images, automatic mosaic, kalman filter