Title:
情報処理装置、情報処理プログラム及び情報処理方法
Document Type and Number:
Japanese Patent JP6798619
Kind Code:
B2
Abstract:
To automatically generate an image classification program capable of classifying images with high accuracy, when automatically generating the image classification program on the basis of genetic programming, an image classification device accepts input of an NG region of a learning image, and creates NG teaching region data indicating a degree of overlap between the NG region and each of divided regions obtained by dividing the learning image (ROI). Furthermore, the image classification device performs image conversion on the learning image by a predetermined image classification program (child individual), calculates an image feature value of each of the divided regions from an image obtained, and creates image feature distribution data corresponding to each of the divided regions. Then, the image classification device calculates a similarity between the NG teaching region data and the image feature distribution data, and uses a result (fitness) of evaluating the image classification program (child individual) on the basis of the similarity, for execution of the genetic programming.
More Like This:
Inventors:
Hiroaki Okamoto
Takeshi Nagato
Tetsuo Koizuka
Takeshi Nagato
Tetsuo Koizuka
Application Number:
JP2019533730A
Publication Date:
December 09, 2020
Filing Date:
July 31, 2017
Export Citation:
Assignee:
富士通株式会社
International Classes:
G06T7/00; G06V10/764
Domestic Patent References:
| JP2013125524A | ||||
| JP2004239870A | ||||
| JP2006260410A | ||||
| JP2009104275A |
Other References:
Yuichiro Yoshimura et al.,"Automatic Construction of Image Inspection Algorithm by using Image Processing Network Programming",Proceedings of SPIE,米国,SPIE,2017年 5月14日,Vol.10338,pp.1-8
Daniel Yapi et al.,"A Learning-Based Approach for Automatic Defect Detection in Textile Images",IFAC-PaperOnLine,NL,Elsevier B.V.,2015年 8月31日,Vol.48, No.3,pp.2423-2428
Daniel Yapi et al.,"A Learning-Based Approach for Automatic Defect Detection in Textile Images",IFAC-PaperOnLine,NL,Elsevier B.V.,2015年 8月31日,Vol.48, No.3,pp.2423-2428
Attorney, Agent or Firm:
Shuhei Katayama