|
Online Camera Pose Optimization for the Surround-view System
Xiao Liu, Lin Zhang*, Ying
Shen, Shaoming Zhang, and Shengjie Zhao |
Introduction
This is the website of our paper "Online Camera Pose Optimization for the
Surround-view System", published in ACM Multimedia 2019.
Surround view system is an important information media for drivers to monitor
the driving environment. A typical surround view system consists of four to six
fsheye cameras arranged around the vehicle. From these camera inputs, a top-down
image of ground around the vehicle, namely the surround view image can be
generated with well calibrated camera poses. Although existing surround view
system solutions can estimate camera poses accurately in ofine environment, the
problem of correcting the camera pose change in online environment is still a
research gap. In this paper, we propose a camera pose optimization method for
surround view system in online environment. Our method consists of two models:
Ground Model and Ground-Camera Model, both of which correct the camera poses by
minimizing photometric errors between ground projections of adjacent cameras.
Experiments show that our method can effectively correct the geometric
misalignment of the surround view image caused by camera pose changes. Since our
method is highly automated with low requirement of calibration site and manual
operation, it has wide application scenario and is convenient for the end users.
To make our results fully reproducible, the dataset and the relevant source code
have been made publicly available on this website.
Source Codes
1. code.zip
This is the code of Ground Model and Ground-Camera Model. The code can be compiled with cmake. The prerequisite of compiling can be found in CMakeLists.txt.
These are some surround-view images for testing. Images in ALL/ are captured by the Front, Left, Back, Right camera respectively. Same index means the images are captured at the same time. Images in surround view/ are the surround-view images corresponded to the images in ALL/.
Results of the test cases before and after being processed by our algorithm.
Last update: Aug. 14, 2019