Design and Simulation of Photogrammetric Networks Using Genetic Algorithms

This work describes the use of genetic algorithms in the design of photogrammetric networks. A system made for this purpose EPOCA (Evolving POsitions of CAmeras) is presented. With the system, a photogrammetric network can be designed using three-dimensional CAD-models of the object. The system provides the attitude of each camera in the network, taking into account the imaging geometry of the system (contribution to intersection angles), as well as, several major constraints like the incidence angle constraint, workspace constraint and the problem of visibility. When planning a photogrammetric network, the cameras should be placed so that each point can be seen at least from two cameras. When the object is three-dimensional a combinatorial problem is presented. Genetic algorithms are stochastic optimization techniques, which have proved useful at solving computationally difficult problems. In order to solve the problem we pose it in terms of a global optimization process. The system is capable of minimizing the error while taking into account the computational burden (the mathematical error plays a role as it leads to the error to be optimized). Our system reproduces configurations reported in the literature, of well-known international experts, like Fraser and Mason. Moreover, the system can design networks for several adjoining planes and complex objects opening new interesting research ways. The results obtained confirm the effectiveness and efficiency of the solution.