US 12,437,532 B2
Intelligent multi-visual camera system and method
Isabelle Butterfield, Bellevue, WA (US); Paul Barsic, Seattle, WA (US); and Alex Terekhov, Bellevue, WA (US)
Assigned to TerraClear Inc., Bellevue, WA (US)
Filed by TerraClear Inc., Bellevue, WA (US)
Filed on Jul. 9, 2021, as Appl. No. 17/372,285.
Claims priority of provisional application 63/050,672, filed on Jul. 10, 2020.
Prior Publication US 2022/0012494 A1, Jan. 13, 2022
Int. Cl. G06V 20/10 (2022.01); G06T 7/246 (2017.01); G06T 7/292 (2017.01); G06T 7/73 (2017.01); G06V 20/58 (2022.01); H04N 13/388 (2018.01); H04N 23/90 (2023.01)
CPC G06V 20/10 (2022.01) [G06T 7/246 (2017.01); G06T 7/292 (2017.01); G06T 7/73 (2017.01); G06V 20/58 (2022.01); H04N 13/388 (2018.05); H04N 23/90 (2023.01); G06T 2207/10016 (2013.01); G06T 2207/20084 (2013.01)] 48 Claims
OG exemplary drawing
 
32. An intelligent multi-camera system, the system comprising:
a multiple sensor array including multiple cameras spaced apart from each other, wherein each of the multiple cameras is mounted on a support frame, and wherein each camera acquires its own independent image; and
a control system that receives input from the multiple cameras, the control system including a processor and a memory storing computer instructions that, when executed by the processor, cause the processor to:
initiate a registration system that projects images from the multiple cameras into a single three-dimensional volume;
detect one or more features of interest in the images from the multiple cameras, wherein detection of features of interest includes classification, pixel location in the image, and boundary representation in the image;
track one or more features of interest in one frame in an image from the multiple cameras, and identify the same one or more features of interest in a subsequent frame in an image from the multiple cameras;
project the features of interest from the multiple cameras into the three-dimensional volume to identify instances where the same objects have been identified by multiple cameras and ensure unique features of interest;
limiting the three-dimensional volume around a ground plane, wherein a search volume is a truncated curved cone shaped volume that covers the three-dimensional volume represented by a rectangular box;
reducing a size of the search volume by rotating the search volume, transforming a position of the search volume from oblique to orthographic, and further truncating the curved cone shaped volume; and
predicting, through the search volume, a presence of an object in the absence of object detection by logging positions of objects over time;
communicate information regarding the features of interest that have been detected.