Abstract: A fast seamless mosaic way to splice image of measurement area is crucial, which provides a guarantee for UAV to give full play to its flexible advantages. Under the circumstances of no other GCP（Ground Control Point）, it splices the image only through the images and GPS data recorded by the UAV system itself. The whole mission area is divided into several blocks with the aid of GPS. GPU-SIFT algorithm is used to match the images that have been blocked and eliminate the mismatch. A fast image fusion seam elimination method is presented, which exhibits smooth transition effect. The result shows that this method can significantly enhance the mosaic efficiency on the premise that certain mosaic precision can be guaranteed and provide the emergency data when sudden natural disasters happen.

Key words: Unmanned Aerial Vehicle（UAV） sequence images, image fusion, seamless mosaic

摘要： 以一种快速无缝的方式对测区影像进行拼接是保障无人机充分发挥机动灵活优势的关键。在无其他控制点的情况下，仅通过利用无人机序列影像和自身记录GPS信息对整个任务区影像进行快速无缝拼接。根据GPS记录的坐标信息对测区影像进行分块。使用GPU-SIFT算法对分块后影像进行匹配与误匹配剔除。提出一种具有较小时间开销的拼接缝消除算法。实验结果表明，该拼接方法在保证一定拼接精度的前提下，大大提高了拼接效率，可为突然自燃灾害等紧急情况下提供应急数据源。

关键词: 无人机序列影像, 图像融合, 无缝拼接