| Full text | |
| Author(s): |
Total Authors: 3
|
| Affiliation: | [1] Aeronaut Inst Technol, Dept Civil Engn, Sao Jose Dos Campos, SP - Brazil
[2] Univ Estadual Paulista, UNESP, Inst Sci & Technol, Dept Environm Engn, Sao Jose Dos Campos, SP - Brazil
Total Affiliations: 2
|
| Document type: | Journal article |
| Source: | GEOSYNTHETICS INTERNATIONAL; v. 26, n. 3, p. 261-272, JUN 2019. |
| Web of Science Citations: | 0 |
| Abstract | |
The use of digital images to evaluate geotextile characteristics has appeared in the scientific literature, particularly for estimating the pore size distribution (PSD) of nonwoven geotextiles used for filtration purposes. While numerous image analysis techniques for PSD curve estimation have been proposed, automation of the analysis methods has received almost no attention. The goal of the present study is to quantify PSD curves from images of representative geotextile specimens using a fully automatic algorithm and considering the concept of largest inscribing opening size. These features are incorporated in a software program (GeotexInspector) developed by the second author. GeotexInspector is based on the concept of classification using the Support Vector Machine technique and digital image processing. These techniques are combined to develop a tool with a robust learning algorithm to identify geotextile fibers and pores. PSD curves from continuous filament and staple fiber nonwoven geotextiles are estimated using GeotexInspector and then compared with theoretical models that have appeared in the literature. The results indicate that the curves obtained from image analysis are in good agreement with model outcomes for models focused on the structure of geotextiles that are similar to the geotextiles used in this study. (AU) | |
| FAPESP's process: | 14/14830-8 - Study and development of new Kernel functions with applications on remote sensing image classification |
| Grantee: | Rogério Galante Negri |
| Support Opportunities: | Regular Research Grants |