Tournament selection based fruit fly optimization and its application in template matching

In this paper, an improved fruit fly optimization algorithm based on tournament selection mechanism (TS-FFO) is put forward. In TS-FFO, considering the fact that the aggregation way in the vision optimization phase will easily cause the loss of biodiversity and make the population jump into the local extreme, tournament selection mechanism is embed into FFO to randomly generate a new conductive individual to replace the current best fruit fly. In addition, in view of the blind search surrounding the best individual in the osphresis optimization phase, the evolutionary formula is also redefined by incorporating the current individual's own information to effectively control the evolution direction and step size. Six high dimensional benchmark functions are used to test and evaluate the TS-FFO. The experimental results demonstrate that TS-FFO has quicker optimizing efficiency and better accuracy compared with the standard FFO and several advanced algorithms. TS-FFO is also used to solve the image template matching problems and the statistical results show that our proposed approach is more effective and efficient than the particle swarm optimization (PSO) and differential search algorithms (DSA).