New method for component-labeling in binary image
一种新的连通区域标记算法*

When solving anycast routing problems with traditional evolutionary algorithms, feasible routes to parts of servers in anycast group were only included in initial population and load of servers was not considered, in addition, the penalty functions of QoS parameter were set too simple; these methods easily resulted in local optimal solutions. In order to overcome these shortcomings, proposed an anycast routing algorithm based on the load of anycast group to initialize population, and firstly introduced penalty functions of QoS parameter with higher distinction degree to compose the fitness function. Simulation experiments are carried out in networks that randomly generated by the model of Waxman, the results show that the algorithm can achieve the optimal route with better delay and bandwidth in less generations compared with traditional algorithms.