Abstract Solid state nuclear track detection plays an extremely important role in the field of nuclear technology, but artificial cognition is extremely time-consuming and labor-consuming. In this paper, an automatic measurement system of solid state nuclear track is proposed, in which the microscopic image of the nuclear track is obtained by the camera and processed digitally, and then use software to identify the image content. Based on the image restoration theory, this paper preprocesses such as noise elimination and luminance correction, and the mathematical morphology algorithm is used for track counting and parameter estimation. The comparison results between automatic recognition and manual recognition show that the algorithm can effectively identify track points from images under different conditions and improve work efficiency.