Name
Abstract | ||
|---|---|---|
In the robotic waste sorting lines, most existing segmentation methods will fail due to irregular shapes and color degradation of solid waste. Especially in the cases of adhesion and occlusion, errors may frequently occur while labeling the ambiguous regions between solid waste objects. In this paper, we propose an efficient RGB-D based segmentation method for construction waste segmentation in harsh environment. First, an efficient background modeling strategy is designed to separate the solid waste regions from the cluttered background. Second, we propose an ambiguous regions extracting method to deal with the problem of adhesion and occlusion. Finally, a relabeling method is developed for ambiguous regions and a high precision segmentation will be obtained. A dataset of construction materials consists of RGB-D images is built to evaluate the proposed method. Results show that our approach outperforms other state-of-the-art methods in harsh environment. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1007/978-981-10-7305-2_40 | Communications in Computer and Information Science |
Keywords | DocType | Volume |
Solid waste recycling,Object segmentation,Ambiguous region extraction | Conference | 773 |
ISSN | Citations | PageRank |
1865-0929 | 0 | 0.34 |
References | Authors | |
0 | 5 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Chao Wang | 1 | 895 | 190.04 |
| Sheng Liu | 2 | 5 | 8.58 |
| Jianhua Zhang | 3 | 25 | 5.97 |
| Yuan Feng | 4 | 218 | 38.29 |
| Sheng-Yong Chen | 5 | 1077 | 114.06 |