Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
FEATURE VALUE GENERATION DEVICE, FEATURE VALUE GENERATION METHOD, INFORMATION PROCESSING DEVICE, AND INFORMATION PROCESSING METHOD
Document Type and Number:
WIPO Patent Application WO/2019/039059
Kind Code:
A1
Abstract:
In the present invention, a feature value generation unit 26a of an information processing unit 20-1 uses spectral data in a multispectral image acquired from a multispectral camera 10 to generate a feature value indicating the relationship between the spectral reflectances at two points within an irradiation range of the same light source. A recognition processing unit 27 uses the feature value generated by the feature value generation unit 26a to carry out subject recognition processing, taking at least one of the two points as the position of a subject to be recognized within an image range in the multispectral image. As a result, the feature value generation unit 26a is capable of simply generating a feature value corresponding to the subject that is robust to light source fluctuation. Further, the recognition processing unit 27 is capable of accurately carrying out recognition processing even if the light source fluctuates during the registration and recognition of the subject.

Inventors:
UEMORI TAKESHI (JP)
ITO ATSUSHI (JP)
GOMI SHINICHIRO (JP)
Application Number:
PCT/JP2018/023112
Publication Date:
February 28, 2019
Filing Date:
June 18, 2018
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
SONY CORP (JP)
International Classes:
G01J3/46; G06T7/00; G06V10/60
Foreign References:
US20100208951A12010-08-19
JP2006130308A2006-05-25
US20130076913A12013-03-28
JP2017133966A2017-08-03
JP2016075624A2016-05-12
Other References:
AKAZAWA, FUMITSUGU ET AL.: "A Biometric Authentication Method by Hyperspectral Datacube with 0.93nm Resolution", THE JOURNAL OF THE INSTITUTE OF ELECTRONICS, INFORMATION AND COMMUNICATION ENGINEERS THE 1ST BIOX, 27 August 2012 (2012-08-27), pages 29 - 38, XP055578292
Attorney, Agent or Firm:
MIYATA, Masaaki et al. (JP)
Download PDF: