Image Recognition In Complex Environments Based On Improved FCM Algorithm

In recent years, image recognition has been commonly utilized in the field of computer vision, while the challenge of image recognition in complex environments has become a topical study. The fuzzy C-mean (FCM) is a clustering algorithm based on fuzzy set theory and achieves clustering by assigning affiliation to each data point. However, in complex environments, the FCM algorithm may fall into local optimal solutions and lead to a decrease in recognition accuracy. To solve this problem, this paper proposes an improved approach based on a genetic algorithm and simulated annealing algorithm applied in the FCM algorithm. The genetic algorithm searches and optimizes the clustering centers by simulating the biological evolution process to find more optimal parameters for the experiment. The simulated annealing algorithm performs global optimization. It generates new solutions by adding a small perturbation to avoid the clustering centers from falling into local optima. Finally, a series of image recognition experiments are conducted in complex environments. The experiment results show that the proposed approach has higher recognition accuracy and robustness in complex environments compared with the traditional FCM algorithm. The algorithm can improve the accuracy of image recognition and has good adaptability in complex environments.