摘要 |
PROBLEM TO BE SOLVED: To improve the convergence of a parameter value to a global solution and prevent processing time from extremely increasing in optimization using an evaluation function. SOLUTION: From optimum value candidates of alternative parameters, which are found by optimization convergence computation using a plurality of auxiliary evaluation functions, a partial set of the optimum value candidates which are small in variation is selected, and the initial values of the alternative parameters are each determined from the optimum value candidates included in the partial set to carry out optimization convergence computation. By this, processing time is prevented from increasing, and the convergence of a common parameter to a global solution is improved. Camera calibration was actually performed, and the result is that a square error between an optimum value of a camera parameter found by the optimization convergence computation and a known ideal value of a camera parameter value is small in almost all trials, and processing time is suppressed. COPYRIGHT: (C)2007,JPO&INPIT
|