Due to the bad service environment of sea-crossing bridge and port engineering, the crack images collected in most cases have complex background and high noise. A method of extracting image cracks feature based on fractal theory and secondary segmentation is proposed in this paper to overcome the shortcomings of the existing technology. In this method, fractal parameters are used as the characteristic parameters of the crack image, which can preferentially suppress the problem of excessive interference in the crack image, and effectively overcome the interference factors such as uneven gray level, large noise and complex background. At the same time, two different algorithms are combined based on the quadratic segmentation theory. The first segmentation is used to eliminate the interference area, and the second segmentation is used to precisely segment the cracks in the target area, so as to accurately and effectively extract the cracks of concrete structure and achieve better segmentation effect.