改进模型估计的无人机侦察视频快速拼接方法

Fast mosaic method of unmanned aerial vehicle reconnaissance video based on improvement model fitting

摘要: 针对提高无人机侦察视频的拼接速度与效果的问题，提出一种改进模型估计的无人机侦察视频快速拼接方法。首先，基于自适应鲁棒性尺度不变的特征检测子对视频各帧进行基于点的特征匹配。其次，提出改进的随机抽样一致性算法进行模型估计，并去除误匹配点。最后，提出侦察影像快速拼接算法，计算各影像变换到正射拼接图的单应性矩阵，完成视频序列拼接。实验结果表明：改进的随机抽样一致性算法在保证鲁棒性的同时，提高了执行速度；侦察影像快速拼接算法提高了拼接速度，同时改善了拼接效果。

Abstract: In order to improve the speed and quality of video mosaic, Fast Mosaic of Unmanned Aerial Vehicle Video based on Improvement Model Fitting (FMUAVRVIMF) was proposed. Firstly, Fast Adaptive Robust Invariant Scalable Feature Detector (FARISFD) was used for video frames registration. Then, the improved Random Sample Consensus(RANSAC) algorithm was proposed for model fitting, and false matching points were removed. Finally, Fast Mosaic of Reconnaissance Frames algorithm was proposed to calculate homography matrix which transformed frames to orthophoto mosaic image, the video sequence mosaic was compeleted. The experiment results show that Improved Random Sample Consensus ensures the robustness, and the mosaic speed was increased, and the mosaic effect was improved.