US 10,891,329 B2
Image recognition method and image recognition apparatus
Hiroto Yoshii, Tokyo (JP); and Masakazu Matsugu, Yokohama (JP)
Assigned to Canon Kabushiki Kaisha, Tokyo (JP)
Filed by CANON KABUSHIKI KAISHA, Tokyo (JP)
Filed on Nov. 29, 2017, as Appl. No. 15/826,396.
Application 15/826,396 is a continuation of application No. 13/375,448, granted, now 9,852,159, previously published as PCT/JP2010/060413, filed on Jun. 15, 2010.
Claims priority of application No. 2009-145457 (JP), filed on Jun. 18, 2009; and application No. 2010-064316 (JP), filed on Mar. 19, 2010.
Prior Publication US 2018/0089187 A1, Mar. 29, 2018
Int. Cl. G06F 16/51 (2019.01); G06T 7/33 (2017.01); G06K 9/62 (2006.01); G06K 9/46 (2006.01)
CPC G06F 16/51 (2019.01) [G06K 9/4642 (2013.01); G06K 9/6202 (2013.01); G06K 9/6255 (2013.01); G06K 9/6282 (2013.01); G06K 9/6292 (2013.01); G06T 7/337 (2017.01); G06T 2207/20021 (2013.01); G06T 2207/20081 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An apparatus for processing an image including a target object, comprising:
one or more processors; and
at least one memory coupled to the one or more processors, the at least one memory having stored thereon instructions which, when executed by the one or more processors, cause the apparatus to:
obtain a plurality of partial target images from a target image including the target object;
discriminate a feature of each of the plurality of partial target images;
estimate, for each of the plurality of partial target images, a class of the target object and a position of the target object in the target image, by referring to a dictionary based on the discriminated feature and a position of each of the plurality of partial target images, wherein in the dictionary, for a plurality of partial learning images obtained by dividing respective one of a plurality of learning images each including the target object, a feature of each partial learning image is registered in correspondence with a class of the target object included in a learning image which contains the partial learning image;
vote, with respect to each of the plurality of partial target images, for a combination of the estimated class of the target object and the estimated position of the target object in the target image; and
recognize the class of the target object and the position of the target object in the target image by aggregating results of the voting for the plurality of partial target images.