Initial value search method for multi-viewpoint stereo measurement using genetic algorithms

Presents a new method utilizing a genetic algorithms to search for initial values for a nonlinear optimization process in multi-viewpoint stereo measurement. First, the authors take a one dimensional matrix called “individual” which consists of parameters to be determined. Next, the authors form a population which consists of several individuals. The authors perform three kinds of genetic operation on the population, that is, crossover, mutation and selection, which evolve the population to fit the environment. The authors reiterate this evolution process and use the evolved result as the initial values for the iteration calculation. The efficiency, of this method is confirmed by simulation and an outdoor experiment