US 12,243,262 B2
Apparatus and method for estimating distance and non-transitory computer-readable medium containing computer program for estimating distance
Wadim Kehl, Tokyo-to (JP); and Hiroharu Kato, Tokyo-to (JP)
Assigned to TOYOTA JIDOSHA KABUSHIKI KAISHA, Toyota (JP)
Filed by TOYOTA JIDOSHA KABUSHIKI KAISHA, Toyota (JP)
Filed on Sep. 9, 2022, as Appl. No. 17/930,826.
Claims priority of application No. 2021-156991 (JP), filed on Sep. 27, 2021.
Prior Publication US 2023/0102186 A1, Mar. 30, 2023
Int. Cl. G06T 7/50 (2017.01); G06T 3/4007 (2024.01); G06T 7/73 (2017.01); G06V 10/77 (2022.01); G06V 20/56 (2022.01)
CPC G06T 7/73 (2017.01) [G06T 3/4007 (2013.01); G06V 10/7715 (2022.01)] 8 Claims
OG exemplary drawing
 
1. An apparatus for estimating distance, comprising a processor configured to:
extract a reference feature map from a reference image generated by a reference camera for taking a picture of an object from a predetermined reference position, the reference feature map representing features corresponding to respective pixels included in the reference image;
extract a source feature map from each of one or more source images respectively generated by one or more source cameras for taking a picture of the object from a position different from the reference position, the source feature map representing features of respective pixels included in the source image;
project the source feature map onto hypothetical planes to generate a cost volume in which coordinates on the hypothetical planes are associated with features, the hypothetical planes being hypothetically disposed by transforming the features in the reference feature map corresponding to respective pixels included in the reference image so that the features in the reference feature map correspond to respective pixels of an image corresponding to an image surface of the reference camera for the case that the image surface is moved in the direction of an optical axis of the reference camera;
set sampling points on a ray in the cost volume extending from the reference position in a direction corresponding to one of the pixels included in the reference image;
interpolate, for each of the sampling points, a feature corresponding to the sampling point, using features associated with coordinates near the sampling point and on some of the hypothetical planes disposed near the sampling point in the cost volume;
calculate occupancy probabilities corresponding to the respective sampling points by inputting features corresponding to the interpolated sampling points into a classifier, the classifier being trained to output the occupancy probabilities each depending on a feature corresponding to coordinates on one of the hypothetical planes and indicating how likely the coordinates are inside the object; and
add up products of the occupancy probabilities corresponding to the respective sampling points and the distances from the reference position to the corresponding sampling points to estimate the distance from the reference position to a surface of the object.