US 12,254,694 B2
Image-based tube top circle detection with multiple candidates
Yao-Jen Chang, Princeton, NJ (US); Stefan Kluckner, Berlin (DE); Benjamin S. Pollack, Jersey City, NJ (US); and Terrence Chen, Princeton, NJ (US)
Assigned to Siemens Healthcare Diagnostics Inc., Tarrytown, NY (US)
Appl. No. 16/630,286
Filed by Siemens Healthcare Diagnostics Inc., Tarrytown, NY (US)
PCT Filed Jun. 25, 2018, PCT No. PCT/US2018/039283
§ 371(c)(1), (2) Date Jan. 10, 2020,
PCT Pub. No. WO2019/013961, PCT Pub. Date Jan. 17, 2019.
Claims priority of provisional application 62/531,108, filed on Jul. 11, 2017.
Prior Publication US 2020/0232908 A1, Jul. 23, 2020
Int. Cl. G01N 21/25 (2006.01); G01N 35/00 (2006.01); G01N 35/04 (2006.01); G06V 20/52 (2022.01)
CPC G06V 20/52 (2022.01) [G01N 21/253 (2013.01); G01N 35/00732 (2013.01); G01N 35/04 (2013.01); G01N 2035/0418 (2013.01); G01N 2035/0498 (2013.01)] 16 Claims
OG exemplary drawing
 
1. A method of image-based sample tube top circle detection across multiple images, the method comprising:
receiving, from at least one camera, a series of images of a tray, the tray comprising a plurality of tube slots, the tray holding a given sample tube in one of the plurality of tube slots;
extracting, using a processor, for the given sample tube a plurality of candidates in each image of the series of images;
computing, using the processor, a plurality of consistency scores, each of the plurality of consistency scores between candidates across images in the series of images from possible combinations of the plurality of candidates;
accumulating, using the processor, the plurality of consistency scores;
selecting, using the processor, a true tube top circle for the given sample tube for each image in the series of images based on a highest consistency score between candidates across images in the series of images;
analyzing the true tube top circle for the given sample tube to determine tube characteristics; and
determining a processing step for the given sample tube based on the determined tube characteristics.