Multi-Threshold Image Segmentation Simulation Based on Bayesian Algorithm Optimization
In the process of multi-threshold image segmentation,low segmentation accuracy may directly affect the image segmentation effect.In order to improve the accuracy of image segmentation,a method for multi-threshold image segmentation was proposed based on Bayesian algorithm optimization.After determining the influencing factors during the optimization of the Bayesian algorithm,we constructed the fitness function correspondingly and thus updated the fitness value of individuals in the algorithm in real-time.Then,we used the minimum hill-climbing method to test the independence of the variables.Meanwhile,we constructed the optimization framework of the algo-rithm.Moreover,we used the ant colony algorithm to learn the structure of the Bayesian network and find the high-or-der excellent pattern of the population of the algorithm,thus optimizing the Bayesian algorithm.Based on the Bayesian optimization algorithm,we calculated the mean value of the occurrence probability of the image gray level,thus obtai-ning the variance between image classes.Finally,we determined the optimal segmentation threshold and thus comple-ted the segmentation of the multi-threshold image.Experimental results show that the Probability Rand Index(PRI)of the proposed method is close to 1.Meanwhile,the Global Consistency Error(GCE)and Value of Information(VOI)are small,indicating that the method has high performance and good image segmentation effect.
Multi-threshold imageBayesian algorithmHill-climbing methodPattern ant colony algorithmFitness function
NETL
NSTL
万方数据
