Efficient monocular pose estimation for complex 3D models

We propose a robust and efficient method to estimate the pose of a camera with respect to complex 3D textured models of the environment that can potentially contain more than 100; 000 points. To tackle this problem we follow a top down approach where we combine high-level deep network classifiers with low level geometric approaches to come up with a solution that is fast, robust and accurate. Given an input image, we initially use a pre-trained deep network to compute a rough estimation of the camera pose. This initial estimate constrains the number of 3D model points that can be seen from the camera viewpoint. We then establish 3D-to-2D correspondences between these potentially visible points of the model and the 2D detected image features. Accurate pose estimation is finally obtained from the 2D-to-3D correspondences using a novel PnP algorithm that rejects outliers without the need to use a RANSAC strategy, and which is between 10 and 100 times faster than other methods that use it. Two real experiments dealing with very large and complex 3D models demonstrate the effectiveness of the approach.

Authors

• Antonio Rubio 0001
• Michael Villamizar
• Luis Ferraz
• Adrián Peñate Sánchez
• Arnau Ramisa
• Edgar Simo-Serra
• Alberto Sanfeliu
• Francesc Moreno-Noguer

Keywords

• Three-dimensional displays
• Solid modeling
• Estimation
• Computational modeling
• Cameras
• Training
• Feature extraction
• Monocular
• Pose Estimation
• Complex 3D Models
• Deep Network
• Image Features
• Input Image
• Top-down Approach
• 3D Point
• Geometric Approach
• Camera Pose
• Accuracy Of Pose Estimation
• Training Set
• Deep Learning
• Computation Time
• Convolutional Neural Network
• Local Features
• Training Images
• Similar Images
• Image Retrieval
• Bag-of-words
• Global Descriptors
• Bag Of Visual Words
• Outlier Rejection
• Geometric Method
• Geometric Approximation
• Translation Error
• Image Descriptors
• Reprojection