| Autor: |
Gang Yao, Wentong Sun, Yang Yang, Yujia Sun, Liangjin Xu, Jian Zhou |
| Jazyk: |
angličtina |
| Rok vydání: |
2022 |
| Předmět: |
|
| Zdroj: |
Frontiers in Materials, Vol 9 (2022) |
| Druh dokumentu: |
article |
| ISSN: |
2296-8016 |
| DOI: |
10.3389/fmats.2022.851555 |
| Popis: |
Chromatic aberration is one of the quality defects in the appearance of fair-faced concrete (FFC). The mainly surface chromatic aberration identification (CAI) method being applied is manual observation, which is subjective and time-consuming. A multi-scale lightweight structured data algorithm (MSLSDA) for CAI in FFC is proposed in this manuscript. An unmanned aerial vehicle (UAV) is used for image acquisition. 2368 FFC sample images are collected to build the datasets. The FFC chromatic aberration features are identified by the improved Residual Network Convolutional Neural Network (CNN) framework to achieve chromatic aberration samples quantitative analysis. The method proposed in this manuscript can verify the generalization prediction ability of the MSLSDA for different building samples by generalization prediction set. The results show that the accuracy in CAI samples and chromatic aberration generalization prediction samples can achieve 92.1 and 99.6%, respectively. The FFC chromatic aberration detection platform (FFC-CADP) built by color space conversion, histogram equalization, image color recognition, image noise reduction and image mask algorithm is able to calculate boundary features, geometric parameter features (length and width), chromatic aberration ratio features, total chromatic aberration ratio and number of chromatic aberration. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
