US 12,340,552 B2
Iterative recognition-guided thresholding and data extraction
Christopher W. Thrasher, Rochester, NY (US); Alexander Shustorovich, Pittsford, NY (US); Stephen Michael Thompson, Oceanside, CA (US); Jan W. Amtrup, Silver Spring, MD (US); and Anthony Macciola, Irvine, CA (US)
Assigned to Tungsten Automation Corporation, Irvine, CA (US)
Filed by Tungsten Automation Corporation, Irvine, CA (US)
Filed on Jun. 15, 2021, as Appl. No. 17/348,584.
Application 17/348,584 is a continuation of application No. 16/267,205, filed on Feb. 4, 2019, granted, now 11,062,163.
Application 16/267,205 is a continuation of application No. 15/214,351, filed on Jul. 19, 2016, granted, now 10,242,285, issued on Mar. 26, 2019.
Claims priority of provisional application 62/194,783, filed on Jul. 20, 2015.
Prior Publication US 2021/0383150 A1, Dec. 9, 2021
Int. Cl. G06V 10/28 (2022.01); G06T 7/11 (2017.01); G06T 7/136 (2017.01); G06T 7/187 (2017.01); G06V 10/25 (2022.01); G06V 10/44 (2022.01)
CPC G06V 10/28 (2022.01) [G06T 7/11 (2017.01); G06T 7/136 (2017.01); G06T 7/187 (2017.01); G06V 10/25 (2022.01); G06V 10/457 (2022.01); G06T 2207/20104 (2013.01)] 13 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
identifying a region of interest of a document depicted within a color digital image;
generating a plurality of binarized images based on the region of interest, wherein the plurality of binarized images are each independently generated using a different one of a plurality of binarization thresholds; and
extracting data from some or all of the plurality of binarized images;
wherein the extracted data comprises one or more connected components represented in the plurality of binarized images;
wherein one or more of the connected components at least partially overlap or are at least partially obscured by a plurality of unique background elements;
wherein the plurality of unique background elements are each independently characterized by different color profiles within the color digital image; and
independently normalizing color within individual regions the color digital image prior to generating the plurality of binarized images, wherein the normalizing comprises stretching a range of intensity values observed in each of a plurality of color channels corresponding to a plurality of pixels of the color digital image such that:
a minimum intensity value observed in each of the plurality of color channels corresponds to a minimum possible intensity value in the color digital image, and
a maximum intensity value observed in each of the plurality of color channels corresponds to a maximum possible intensity value in the color digital image.