To address the issues of low accuracy and poor practical performance in existing bridge crack recognition algorithms, this paper took the high-precision image of a bridge surface as the research object and proposed a crack recognition algorithm based on the characteristics of image connected domain and the calculation principle of the maximum inscribed circle (CIACM). Firstly, traditional image processing algorithms such as graying, uniform light filtering, and edge detection were used to show the edge features of cracks. Then, based on the characteristics of the connected domain and the calculation principle of the maximum inscribed circle, the crack was identified, and the maximum width of the crack was screened. The experimental results show that the absolute error of the algorithm is not more than 0.02 mm; the average relative error is 2.47%, and the standard deviation is 1.52%. For fine cracks with widths <0.2 mm, the average relative error and standard deviation are 4.71% and 1.54%, respectively, meeting bridge inspection precision requirements. The study demonstrates that the CIACM algorithm significantly improves crack recognition precision, particularly for fine cracks, and it provides reliable technical support for the automated assessment of bridge surface damage.