Paper Info
Title
Scene Coordinate Regression Forests for Camera Relocalization in RGB-D Images
Abstract
We address the problem of inferring the pose of an RGB-D camera relative to a known 3D scene, given only a single acquired image. Our approach employs a regression forest that is capable of inferring an estimate of each pixel's correspondence to 3D points in the scene's world coordinate frame. The forest uses only simple depth and RGB pixel comparison features, and does not require the computation of feature descriptors. The forest is trained to be capable of predicting correspondences at any pixel, so no interest point detectors are required. The camera pose is inferred using a robust optimization scheme. This starts with an initial set of hypothesized camera poses, constructed by applying the forest at a small fraction of image pixels. Preemptive RANSAC then iterates sampling more pixels at which to evaluate the forest, counting inliers, and refining the hypothesized poses. We evaluate on several varied scenes captured with an RGB-D camera and observe that the proposed technique achieves highly accurate relocalization and substantially out-performs two state of the art baselines.
Year
DOI
Venue
2013
10.1109/CVPR.2013.377
Computer Vision and Pattern Recognition
Keywords
Field
DocType
cameras,image processing,optimisation,regression analysis,3D points,RANSAC,RGB-D camera,RGB-D images,acquired image,camera relocalization,hypothesized camera poses,image pixels,robust optimization scheme,scene coordinate regression forests
Computer vision,Pattern recognition,Computer science,RANSAC,Camera auto-calibration,Image processing,Camera resectioning,RGB color model,Artificial intelligence,Pixel,Sampling (statistics),Computation
Conference
Volume
Issue
ISSN
2013
1
1063-6919
Citations 
PageRank 
References 
112
3.10
30
Authors
4
Search Limit
100112
Name
Order
Citations
PageRank
Jamie Shotton17571324.72
Ben Glocker22157119.81
Christopher Zach3145784.01
Shahram Izadi45573285.39