Title:
AI CRACK-DETECTING METHOD USING BRIDGE PIER DRIVING DEVICE
Document Type and Number:
WIPO Patent Application WO/2024/080436
Kind Code:
A1
Abstract:
The present invention proposes an AI crack-detecting method using a bridge pier driving device, wherein reinforcement data is generated by expanding bridge crack learning data which is insufficient as learning data, machine learning is conducted on the basis thereof, thereby securing more excellent crack-detecting ability, and processing resources necessary for image processing for crack detection are minimized, thereby detecting cracks accurately and quickly. To this end, the present invention uses a bridge pier driving device which captures images of a bridge pier by using multiple cameras, and may comprise: a) a preprocessing step of removing image areas obtained by capturing images of subjects other than the bridge pier, and correcting images from respective cameras in conformity with the form of the bridge pier; b) a crack-detecting step of stitching the preprocessed images with regard to respective cameras, thereby forming a bridge pier image, delimiting the bridge pier image into multiple tiles, and detecting a cracked tile with reference to crack learning data with regard to respective tiles, thereby reducing image processing load; and c) a numericalization step of applying skeletonization to the cracked tile such that the center line of the crack included in the cracked tile is left, and same is measured and numericalized.
Inventors:
CHOI JONG DAI (KR)
Application Number:
PCT/KR2022/017202
Publication Date:
April 18, 2024
Filing Date:
November 04, 2022
Export Citation:
Assignee:
FD CO LTD (KR)
International Classes:
G01N21/88; G06N3/08; G06T5/00; G06T5/50; G06T7/00
Foreign References:
| KR20200013148A | 2020-02-06 | |||
| CN110378879A | 2019-10-25 | |||
| KR102239377B1 | 2021-04-13 | |||
| KR20210077322A | 2021-06-25 | |||
| CN114689600A | 2022-07-01 |
Attorney, Agent or Firm:
POLARIS INTERNATIONAL PATENT & LAW FIRM (KR)
Download PDF: