Optimization, genetic algorithm
Most of the digital signal processing procedures can be specified in terms of optimisation procedure. In the case of multidimensional signal processing (image and video), the optimisation space is usually huge and the optimisation procedure highly time consuming. When a real time processing is not required, evolutionary methods offer a viable alternative to overcome large optimisation space problem.
Evolution of species first described by Darwin solve the most difficult problems in the most elegant way. Can much simpler but still complex problems be solved using the same principles? The theory of genetic algorithms and evolutionary programming proved that difficult optimisation problems can be efficiently solved by what is known as a genetic algorithm – a simulation of natural evolution. Surprisingly, this is not the case for simple problems when simple straightforward methods outperform evolutionary algorithms.