| Autor: |
Iryna Ivasenko, Roman Vorobel, Oleksandr Maksymenko, Olena Berehulyak |
| Rok vydání: |
2022 |
| Zdroj: |
Information extraction and processing. 2022:62-68 |
| ISSN: |
0474-8662 |
| DOI: |
10.15407/vidbir2022.50.062 |
| Popis: |
The application of optimization method of image segmentation to estimate the percentage content of concrete components is considered. Non-destructive testing methods (ultrasonic, magnetic, radiographic, image processing) are actively used to assess the condition of concrete structures and structures of long-term use. Recently, the share of studies of the mechanical properties of concrete based on the processing of images of sections of its samples has increased significantly. The relationship between the parameters obtained by digital image processing methods and the compressive strength of concrete is established on the basis of regression analysis. A method of segmentation of color images of test sample sections of concrete based on the Gaussian mixture method and level-sets model is developed. Based on the analysis of the differences in the color characteristics of the background and the object, it is concluded that they can be divided into two classes in the RGB color space. For this purpose appropriate training samples are created, which contain image pixel samples with typical features of the respective classes. The training sample consists of a set of feature vectors of image pixels. The parameters of the segmentation model have been adjusted. The experimental results of the segmentation of color images of sections of test concrete samples by the proposed method are presented. An analysis of the obtained results is carried out. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
