In order to solve the problem of bridge crack extraction by traditional clustering algorithm, the recognition accuracy is low and the feature information is easy to be lost due to the uneven shadow and light, a method of bridge crack extraction based on hybrid bald eagle-aquila optimizer (HBAO) cross iteration was proposed to improve Fuzzy C-Means (FCM) clustering. Firstly, coupled chaotic mapping initialization was introduced, and refraction learning was integrated to increase population diversity. Secondly, in order to enhance the performance of the global search phase of the bald eagle search (BES) algorithm, this phase was replaced with the expanded and narrowed search strategy of the BES optimization, which significantly improved the convergence trend and global search ability of BES, and improved the success rate of finding the optimal clustering center of FCM. Then, HBAO and weighted entropy method were used to optimize FCM clustering algorithm, which improved the robustness and enhanced the search accuracy, and obtained better clustering results. Finally, the clustering performance evaluation experiment is carried out on UCI standard data set with 6 comparison algorithms, and the comprehensive performance of the proposed algorithm is verified. Furthermore, the proposed algorithm is tested on 4 different fracture patterns. Experimental results show that compared with other algorithms, the proposed algorithm has the best extraction effect.