发明名称 |
Method and apparatus for solving position and orientation from correlated point features in images |
摘要 |
A method and apparatus for determining a position and attitude of at least one camera by calculating and extracting estimated rotation and translation values from an estimated fundamental matrix based on information from at least a first and second 2D image. Variable substitution is utilized to strengthen derivatives and provide a more rapid convergence. A solution is provided for solving position and orientation from correlated point features in images using a method that solves for both rotation and translation simultaneously. |
申请公布号 |
US8953847(B2) |
申请公布日期 |
2015.02.10 |
申请号 |
US201013877252 |
申请日期 |
2010.10.01 |
申请人 |
SAAB AB |
发明人 |
Modén Anders |
分类号 |
G06K9/00;G06T7/00 |
主分类号 |
G06K9/00 |
代理机构 |
Venable LLP |
代理人 |
Venable LLP ;Franklin Eric J. |
主权项 |
1. A method for determining a position and attitude of at least one camera by calculating and extracting an estimated rotation and translation values from an estimated fundamental matrix based on information from at least a first and second 2D image, said method comprising:
generating four equations for the estimated rotation and translation values from the estimated fundamental matrix using single value decomposition; testing said four equations for the estimated rotation and translation values with a number of known points deriving one initial-equation which describes an initial calculated first estimate of the rotation and translation values; defining an equation system of the equations and substituting the heading, pitch and roll variables by changing heading, pitch and roll variables with a 3×3 matrix and substituting position variables by changing z the position variables with natural logarithms of the position variables; using a parameterization for each 3D point containing a distance in a first camera to each matched point; calculating the equation system with k+12 unknown variables, where k equals a number of 3D points and where the 12 unknown variables come from the 3×3 matrix yielding 9 unknown variables and 3 unknown position variables from the translation by solving the equation system with the input values from the initial estimated rotation and translation from the generating and testing using LevenBerg-Marquardt or any similar process, thereby obtaining a final estimate of the rotation and translation values. |
地址 |
Linköping SE |