As a nondestructive testing method

magnetic particle testing is widely used in power industry. The comprehensive sensitivity of the magnetic particle detection can be tested by A1 standard shims. However

at present

the evaluation of the clarity of magnetic trace on standard shims still depends on subjective judgment of the tester. In order to eliminate the subjective factors in the evaluation

this paper proposes a standard magnetic trace evaluation method based on machine vision. Based on Python programming language and OpenCV function library

the initial obtained magnetic trace is processed by image correction

magnetic trace extraction

quantitative analysis and evaluation using computer program. On this basis

the influence of the thickness of non-magnetic layer on the comprehensive sensitivity of magnetic particle detection is investigated using this method. It is shown that the magnetic trace evaluation method based on machine vision is more objective and accurate than the conventional manual evaluation.