Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignment
In surveying, mapping and geographic information systems, building extraction from remote sensing imagery is a common task. However, there are still some challenges in automatic building extraction. First, using only single-scale depth features cannot take into account the uncertainty of features su...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2022-09-01
|
| Series: | International Journal of Applied Earth Observations and Geoinformation |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1569843222001637 |
| _version_ | 1811272002912124928 |
|---|---|
| author | Wangbin Li Kaimin Sun Hepeng Zhao Wenzhuo Li Jinjiang Wei Song Gao |
| author_facet | Wangbin Li Kaimin Sun Hepeng Zhao Wenzhuo Li Jinjiang Wei Song Gao |
| author_sort | Wangbin Li |
| collection | DOAJ |
| description | In surveying, mapping and geographic information systems, building extraction from remote sensing imagery is a common task. However, there are still some challenges in automatic building extraction. First, using only single-scale depth features cannot take into account the uncertainty of features such as the hue and texture of buildings in images, and the results are prone to missed detection. Moreover, extracted high-level features often lose structural information and have scale differences with low-level features, which results in less accurate extraction of boundaries. To simultaneously address these problems, we propose pyramid feature extraction (PFE) to construct multi-scale representations of buildings, which is inspired by the feature extraction of scale-invariant feature transform. We also apply attention modules in channel dimension and spatial dimension to PFE and low-level feature maps. Furthermore, we use the structural-cue-guided feature alignment module to learn the correlation between feature maps at different levels, obtaining high-resolution features with strong semantic representation and ensuring the integrity of high-level features in both structural and semantic dimensions. An edge loss is applied to get a highly accurate building boundary. For the WHU Building Dataset, our method achieves an F1 score of 95.3% and an Intersection over Union (IoU) score of 90.9%; for the Massachusetts Buildings Dataset, our method achieves an F1 score of 85.0% and an IoU score of 74.1%. |
| first_indexed | 2024-04-12T22:32:25Z |
| format | Article |
| id | doaj.art-11643341a04640a4902b50ac80303255 |
| institution | Directory Open Access Journal |
| issn | 1569-8432 |
| language | English |
| last_indexed | 2024-04-12T22:32:25Z |
| publishDate | 2022-09-01 |
| publisher | Elsevier |
| record_format | Article |
| series | International Journal of Applied Earth Observations and Geoinformation |
| spelling | doaj.art-11643341a04640a4902b50ac803032552022-12-22T03:13:56ZengElsevierInternational Journal of Applied Earth Observations and Geoinformation1569-84322022-09-01113102970Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignmentWangbin Li0Kaimin Sun1Hepeng Zhao2Wenzhuo Li3Jinjiang Wei4Song Gao5The State Key Lab, LIESMARS, Wuhan University, Wuhan, 430072, Hubei Province, ChinaThe State Key Lab, LIESMARS, Wuhan University, Wuhan, 430072, Hubei Province, China; Corresponding author.Project Management Center, Marine Equipment Department, 100072, Beijing, ChinaThe State Key Lab, LIESMARS, Wuhan University, Wuhan, 430072, Hubei Province, ChinaThe State Key Lab, LIESMARS, Wuhan University, Wuhan, 430072, Hubei Province, ChinaThe State Key Lab, LIESMARS, Wuhan University, Wuhan, 430072, Hubei Province, ChinaIn surveying, mapping and geographic information systems, building extraction from remote sensing imagery is a common task. However, there are still some challenges in automatic building extraction. First, using only single-scale depth features cannot take into account the uncertainty of features such as the hue and texture of buildings in images, and the results are prone to missed detection. Moreover, extracted high-level features often lose structural information and have scale differences with low-level features, which results in less accurate extraction of boundaries. To simultaneously address these problems, we propose pyramid feature extraction (PFE) to construct multi-scale representations of buildings, which is inspired by the feature extraction of scale-invariant feature transform. We also apply attention modules in channel dimension and spatial dimension to PFE and low-level feature maps. Furthermore, we use the structural-cue-guided feature alignment module to learn the correlation between feature maps at different levels, obtaining high-resolution features with strong semantic representation and ensuring the integrity of high-level features in both structural and semantic dimensions. An edge loss is applied to get a highly accurate building boundary. For the WHU Building Dataset, our method achieves an F1 score of 95.3% and an Intersection over Union (IoU) score of 90.9%; for the Massachusetts Buildings Dataset, our method achieves an F1 score of 85.0% and an IoU score of 74.1%.http://www.sciencedirect.com/science/article/pii/S1569843222001637Building extractionStructural-cue-guided feature alignmentConvolutional neural network (convNet) |
| spellingShingle | Wangbin Li Kaimin Sun Hepeng Zhao Wenzhuo Li Jinjiang Wei Song Gao Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignment International Journal of Applied Earth Observations and Geoinformation Building extraction Structural-cue-guided feature alignment Convolutional neural network (convNet) |
| title | Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignment |
| title_full | Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignment |
| title_fullStr | Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignment |
| title_full_unstemmed | Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignment |
| title_short | Extracting buildings from high-resolution remote sensing images by deep ConvNets equipped with structural-cue-guided feature alignment |
| title_sort | extracting buildings from high resolution remote sensing images by deep convnets equipped with structural cue guided feature alignment |
| topic | Building extraction Structural-cue-guided feature alignment Convolutional neural network (convNet) |
| url | http://www.sciencedirect.com/science/article/pii/S1569843222001637 |
| work_keys_str_mv | AT wangbinli extractingbuildingsfromhighresolutionremotesensingimagesbydeepconvnetsequippedwithstructuralcueguidedfeaturealignment AT kaiminsun extractingbuildingsfromhighresolutionremotesensingimagesbydeepconvnetsequippedwithstructuralcueguidedfeaturealignment AT hepengzhao extractingbuildingsfromhighresolutionremotesensingimagesbydeepconvnetsequippedwithstructuralcueguidedfeaturealignment AT wenzhuoli extractingbuildingsfromhighresolutionremotesensingimagesbydeepconvnetsequippedwithstructuralcueguidedfeaturealignment AT jinjiangwei extractingbuildingsfromhighresolutionremotesensingimagesbydeepconvnetsequippedwithstructuralcueguidedfeaturealignment AT songgao extractingbuildingsfromhighresolutionremotesensingimagesbydeepconvnetsequippedwithstructuralcueguidedfeaturealignment |