A Segmentation Algorithm for Overlapping Nuclear Track Image

Solid state nuclear track detection technology is a significantly important nuclear technology based on measuring track density and geometric parameters, but particle track overlap hinders automatic identification, which seriously affects the measurement accuracy of track parameters. According to the characteristics of nuclear track images, a segmentation method for overlapping nuclear tracks is proposed, which includes three steps: preprocessing, background segmentation and overlap separation. In the preprocessing, bilateral filtering is used to denoise the image, and homomorphic filtering is used to improve the impact of uneven illumination. Subsequently, the background segmentation uses adaptive iterative threshold segmentation to obtain the initial contour of the track. Finally, mathematical morphology and an improved watershed algorithm are used to separate overlapping tracks. The result suggests that this method can correctly segment the tracks with different degrees of overlap and effectively improve the accuracy of automatic measurement.