-
15 things to do in every city
Here is a list things I want to do in every city I have lived:
-
Estimate focal length from Homography
This post discusses estimating focal length from Homography. This technique tries to estimate focal lengths from the given homography under the assumption that the camera undergoes rotations around its centre only. The original paper is Construction of Panoramic Image Mosaics with Global and Local Alignment. The OpenCV implementation can be found here.
-
Estimate relative pose from Essential matrix
This post discusses estimating relative camera positions and orientations from Essential matrix.
-
Bundle adjustment of Structure from Motion in Ceres Solver
Bundle adjustment is used ubiquitously as the last step of most feature based estimation problems. The goal is to simultaneously optimize the motion and data (estimated 2D feature or 3D point position). Some popular open source libraries are: sparse bundle adjustment (sba), ceres solver, Multicore Bundle Adjustment, Simple Sparse Bundle Adjustment (SSBA). In this post, I focus only on ceres solver. I’ll add support for sba in future release of open3DCV.
-
Feature matching
Feature matching is a critical part in many vision problems, such as estimation of various matrices (projection, homography, essential, fundamental, relative post, etc), tracking, SfM and SLAM, just to name a few. I will discuss here in this post implementation of the feature matching algorithms implemented in open3DCV.
-
Robust Estimation of Fundamental Matrix
This post is the results of two previous post: Estimation of fundamental matrix and A framework for RANSAC.
-
Point cloud animation in Blender
This is a tutorial on how to render the point cloud animation in Blender. We use the Blender Render engine instead of Cycles.
-
Essential matrix
- 5-point algorithm
- estimate from Fundamental matrix
-
A framework for RANSAC
RANSAC is short for Random SAmple Consensus, which is an iterative method to estimate parameters of a mathematical model from a set of observed data that contains outliers. It can be incorporated to estimate many key matrices in vision, such as homography, fundamental/essential matrix, etc. This post aims to develop a general framework so that RANSAC could be easier applied to existing estimation algorithms. The code is modified from the code originally written by Ziv Yaniv.
-
Estimation of camera pose
There are multiple ways of estimating camera position and orientation, typical approaches include
- camera pose from point correspondences
- camera pose from Fundamental matrix