PCAFlow-Sparse feature matching

最新推荐文章于 2024-10-04 22:29:20 发布

原创最新推荐文章于 2024-10-04 22:29:20 发布 · 940 阅读

0 ·

CC 4.0 BY-SA版权

Feature 专栏收录该内容

7 篇文章

订阅专栏

本文介绍了一种视频帧间特征匹配的方法，首先利用CLAHE进行图像对比度归一化以增强细节，然后采用专为视觉里程计设计的特征进行匹配。实践表明，这种方法相较于SURF或SIFT等传统图像匹配特征，在视频相邻帧间有更好的匹配效果。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

http://files.is.tue.mpg.de/black/papers/cvpr2015_pcaflow.pdf

First, we normalize the image contrast using CLAHE [55] to increase detail that can be capturedby the feature detectors. Then, we use the featuresfrom [17], which are designed for visual odometry applications.We found that to match features across video frames,these features work much better than image matchingfeatures such as SURF or SIFT. The latter are invariantagainst a wide range of geometric deformations whichrarely occur in adjacent frames of a video, and hence returna large number of mismatches.

[55] K. Zuiderveld. Contrast limited adaptive histogramequalization. In P. S. Heckbert, editor, Graphics GemsIV, pages 474–485. Academic Press Professional, Inc.,San Diego, CA, USA, 1994.

code : http://www.cnblogs.com/Imageshop/archive/2013/04/07/3006334.html

https://github.com/joshdoe/opencv-clahe

[17] A. Geiger, J. Ziegler, and C. Stiller. StereoScan:Dense 3D reconstruction in real-time. In IntelligentVehicles Symposium (IV), 2011 IEEE, pages 963–968,June 2011.

code : http://www.cvlibs.net/software/libviso/