<p>Train wheels are critical components for the safe operation of railway trains. Weak magnetic detection technology has been proven to have advantages such as fast speed and high efficiency in detecting wheel tread cracks, but sensor jitter and external electromagnetic interference during the detection process remain issues to be addressed. This study innovatively proposes a comparative detection method based on weak magnetic detection technology. First, a magnetic dipole model was established in the polar coordinate system, revealing the weak magnetic signal characteristics of cracks on train wheel treads and the influence of lift-off height on signal amplitude. Second, a comparative detection method was proposed, and a contrast difference recognition algorithm was designed. Artificial defects were prefabricated on two wheels, and detection experiments were conducted to verify the feasibility of the comparative detection method and the contrast difference recognition algorithm. The corrected magnetic anomaly amplitude from the algorithm was used to perform quantitative analysis of crack depth, and a quantitative equation was established. Experimental results show that: The comparative detection method based on weak magnetic detection technology is feasible for detecting wheel tread crack defects; the designed contrast difference recognition algorithm can effectively identify wheel tread cracks; there is a logarithmic correlation between crack depth and magnetic anomaly amplitude; and the depth quantization equation fitted with a logarithmic basis function provides accurate quantification of crack depth, with an average error rate of 5.705%. The comparative detection method and contrast difference recognition algorithm based on weak magnetic detection technology have engineering applicability in the detection and evaluation of train wheel tread cracks.</p>