Title:
METHOD FOR TRAINING DEVICE TO REMOVE NOISE FROM X-RAY IMAGE AND COMPUTING DEVICE THEREFOR
Document Type and Number:
WIPO Patent Application WO/2023/229385
Kind Code:
A1
Abstract:
Disclosed is a method for preprocessing an X-ray image that inputs a first image converted from an X-ray image into a first network to obtain a second image from which noise has been removed. The first network is trained by means of supervised learning comprising the steps of: converting a training image selected from a set of training X-ray images into a first training image; generating a first sub-image and a second sub-image by downsampling the first training image; and training the first network through supervised learning so as to reduce an error between an output image converted from the first sub-image and the second sub-image. The size of the first training image is N2*N4, and the sizes of the first sub-image (805), the output image (803), and the second sub-image are all N1*N3 (where N2=2*N1 and N4=2*N3). In addition, the size of the X-ray image is R0*C0, and the sizes of both the first image and the second image are N1*N3 (where R0 >= N1 and C0 >= N3).
Inventors:
PARK JUHYUNG (KR)
SHIN DONGMYUNG (KR)
SHIN DONGMYUNG (KR)
Application Number:
PCT/KR2023/007109
Publication Date:
November 30, 2023
Filing Date:
May 24, 2023
Export Citation:
Assignee:
RADISEN CO LTD (KR)
International Classes:
G06T5/00; G06T3/40; G06T11/00
Foreign References:
| KR20210152048A | 2021-12-14 | |||
| KR20090056515A | 2009-06-03 | |||
| KR20210115010A | 2021-09-24 | |||
| KR20210116923A | 2021-09-28 | |||
| KR20220056067A | 2022-05-04 |
Attorney, Agent or Firm:
YANG, Ki-Hyuk (KR)
Download PDF: