Abstract: This paper proposes an algorithm to obtain the statistic data of quantum dots from atomic force microscopy photos.Starting from identifying the dynamic values of each regional maximum,the peak of each qualified quantum dot is located.Their positions are used as the markers for the next step, which is to apply the marker watershed transform to obtain a rough segmentation of the quantum dots.According to the boundary of the coarse partition,each quantum dot is cut from the original photo.A process is then carried out to filter the possible attached substrates based on the area-height distribution of the current quantum dot.After all the above stages,all the quantum dots can be accurately and robustly extracted and thus their properties, such as height,lateral size,and volume,can easily be measured.

Key words: mathematical morphology, watershed transform, quantum dots characterization, automatic statistics