Frame Rate Fusion and Upsampling of EO/LIDAR Data for Multiple Platforms

T. Nathan Mundhenk, Kyungnam Kim, Yuri Owechko; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2014, pp. 748-755

Abstract


We propose a method for fusing a LIDAR point cloud to camera data in real time, which will also backfill the myriad of data holes LIDAR creates. This is done in a way that also leverages the images features to weight how point clouds are filled. Multithreaded programing and GP-GPU methods allow us to obtain 10 fps with a Velodyne 64E LIDAR completely fused in 360° using a Ladybug panoramic camera. The method also generalizes to other kinds of point clouds such as those obtained by aerial vehicles. The primary advantage of our approach is it combines 360° fusion with upsampling in real time without mode smoothing.

Related Material


[pdf]
[bibtex]
@InProceedings{Mundhenk_2014_CVPR_Workshops,
author = {Nathan Mundhenk, T. and Kim, Kyungnam and Owechko, Yuri},
title = {Frame Rate Fusion and Upsampling of EO/LIDAR Data for Multiple Platforms},
booktitle = {Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2014}
}