Global Gray-level Thresholding Based on Object Size
2016 (English)In: Cytometry Part A, ISSN 1552-4922, E-ISSN 1552-4930, Vol. 89A, no 4, 385-390 p.Article in journal (Refereed) Published
In this article, we propose a fast and robust global gray-level thresholding method based on object size, where the selection of threshold level is based on recall and maximum precision with regard to objects within a given size interval. The method relies on the component tree representation, which can be computed in quasi-linear time. Feature-based segmentation is especially suitable for biomedical microscopy applications where objects often vary in number, but have limited variation in size. We show that for real images of cell nuclei and synthetic data sets mimicking fluorescent spots the proposed method is more robust than all standard global thresholding methods available for microscopy applications in ImageJ and CellProfiler. The proposed method, provided as ImageJ and CellProfiler plugins, is simple to use and the only required input is an interval of the expected object sizes.
Place, publisher, year, edition, pages
John Wiley & Sons, 2016. Vol. 89A, no 4, 385-390 p.
pattern recognition, automated, algorithms, microscopy
Medical Image Processing
Research subject Computerized Image Processing
IdentifiersURN: urn:nbn:se:uu:diva-275180DOI: 10.1002/cyto.a.22806ISI: 000374730400008PubMedID: 26800009OAI: oai:DiVA.org:uu-275180DiVA: diva2:899146
FunderScience for Life Laboratory - a national resource center for high-throughput molecular bioscienceeSSENCE - An eScience CollaborationSwedish Research Council, 2012-4968