由于无人机影像具有数据量大、分辨率高等特点，在进行影像拼接处理时会涉及多张影像，因此其拼接误差不能忽略，且随着拼接影像数量的增多误差会不断积累。在分析了无人机影像拼接主要误差来源之后，较深入地研究了影像重叠度及误差累积问题。确定了无人机影像拼接的最佳重叠度，并依据最佳重叠度对原始影像进行“抽稀”处理，最后提出了一种适用于无人机影像的分块拼接方法。研究结果表明，根据最佳重叠度对原始影像进行“抽稀”能减少参与拼接的影像数量，对“抽稀”后的影像采用分块拼接方法能够有效地减小拼接误差、改善影像拼接视觉效果，同时在一定程度上提高了影像处理效率。

As the unmanned aerial vehicle (UAV) image has large data amounts and high resolution, the stitching error cannot be ignored during the image mosaic process which involves multiple images. Furthermore, with the number of mosaic images increasing the errors will continue to accumulate. The main error sources of UAV image mosaics are analyzed. A deep study is made on the image overlap and the error accumulation problem. The optimal overlap of UAV image mosaic is determined, and the original image is sparsed according to the optimal overlap. Aiming at the error accumulation problem during splicing process, the block splicing method is proposed. The experimental results show that the number of images can be reduced according to the optimal overlap, and the stitching error can be reduced effectively as the block splicing method is adopted to splice sparsed image. The vision effect of image splicing can be improved, with the efficiency of image processing enhanced in a certain extent.