{"_id":"TjchMCH6sKKzNE8zd","bibbaseid":"besbes-benazzabenyahia-roadnetworkextractionbyahigherordercrfmodelbuiltoncenterlinecliques-2015","authorIDs":[],"author_short":["Besbes, O.","Benazza-Benyahia, A."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"firstnames":["O."],"propositions":[],"lastnames":["Besbes"],"suffixes":[]},{"firstnames":["A."],"propositions":[],"lastnames":["Benazza-Benyahia"],"suffixes":[]}],"booktitle":"2015 23rd European Signal Processing Conference (EUSIPCO)","title":"Road network extraction by a higher-order CRF model built on centerline cliques","year":"2015","pages":"1631-1635","abstract":"The goal of this work is to recover road networks from aerial images. This problem is extremely challenging because roads not only exhibit a highly varying appearance but also are usually occluded by nearby objects. Most importantly, roads are complex structures as they form connected networks of segments with slowly changing width and curvature. As an effective tool for their extraction, we propose to resort to a Conditional Random Field (CRF) model. Our contribution consists in representing the prior on the complex structure of the roads by higher-order potentials defined over centerline cliques. Robust PN-Potts potentials are defined over such relevant cliques as well as over background cliques to integrate long-range constraints within the objective model energy. The optimal solution is derived thanks to graph-cuts tools. We demonstrate promising results and make qualitative and quantitative comparisons to the state of the art methods on the Vaihingen database.","keywords":"geophysical image processing;image representation;random processes;road network extraction;higher-order CRF model;centerline clique;aerial imaging;conditional random field model;robust PN-Potts potential;graph-cuts tool;Vaihingen database;Roads;Image segmentation;Shape;Feature extraction;Color;Robustness;Detectors;Road network;higher-order CRF;centerline cliques;graph-cuts;aerial images","doi":"10.1109/EUSIPCO.2015.7362660","issn":"2076-1465","month":"Aug","url":"https://www.eurasip.org/proceedings/eusipco/eusipco2015/papers/1570103659.pdf","bibtex":"@InProceedings{7362660,\n author = {O. Besbes and A. Benazza-Benyahia},\n booktitle = {2015 23rd European Signal Processing Conference (EUSIPCO)},\n title = {Road network extraction by a higher-order CRF model built on centerline cliques},\n year = {2015},\n pages = {1631-1635},\n abstract = {The goal of this work is to recover road networks from aerial images. This problem is extremely challenging because roads not only exhibit a highly varying appearance but also are usually occluded by nearby objects. Most importantly, roads are complex structures as they form connected networks of segments with slowly changing width and curvature. As an effective tool for their extraction, we propose to resort to a Conditional Random Field (CRF) model. Our contribution consists in representing the prior on the complex structure of the roads by higher-order potentials defined over centerline cliques. Robust PN-Potts potentials are defined over such relevant cliques as well as over background cliques to integrate long-range constraints within the objective model energy. The optimal solution is derived thanks to graph-cuts tools. We demonstrate promising results and make qualitative and quantitative comparisons to the state of the art methods on the Vaihingen database.},\n keywords = {geophysical image processing;image representation;random processes;road network extraction;higher-order CRF model;centerline clique;aerial imaging;conditional random field model;robust PN-Potts potential;graph-cuts tool;Vaihingen database;Roads;Image segmentation;Shape;Feature extraction;Color;Robustness;Detectors;Road network;higher-order CRF;centerline cliques;graph-cuts;aerial images},\n doi = {10.1109/EUSIPCO.2015.7362660},\n issn = {2076-1465},\n month = {Aug},\n url = {https://www.eurasip.org/proceedings/eusipco/eusipco2015/papers/1570103659.pdf},\n}\n\n","author_short":["Besbes, O.","Benazza-Benyahia, A."],"key":"7362660","id":"7362660","bibbaseid":"besbes-benazzabenyahia-roadnetworkextractionbyahigherordercrfmodelbuiltoncenterlinecliques-2015","role":"author","urls":{"Paper":"https://www.eurasip.org/proceedings/eusipco/eusipco2015/papers/1570103659.pdf"},"keyword":["geophysical image processing;image representation;random processes;road network extraction;higher-order CRF model;centerline clique;aerial imaging;conditional random field model;robust PN-Potts potential;graph-cuts tool;Vaihingen database;Roads;Image segmentation;Shape;Feature extraction;Color;Robustness;Detectors;Road network;higher-order CRF;centerline cliques;graph-cuts;aerial images"],"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"inproceedings","biburl":"https://raw.githubusercontent.com/Roznn/EUSIPCO/main/eusipco2015url.bib","creationDate":"2021-02-13T17:31:52.462Z","downloads":0,"keywords":["geophysical image processing;image representation;random processes;road network extraction;higher-order crf model;centerline clique;aerial imaging;conditional random field model;robust pn-potts potential;graph-cuts tool;vaihingen database;roads;image segmentation;shape;feature extraction;color;robustness;detectors;road network;higher-order crf;centerline cliques;graph-cuts;aerial images"],"search_terms":["road","network","extraction","higher","order","crf","model","built","centerline","cliques","besbes","benazza-benyahia"],"title":"Road network extraction by a higher-order CRF model built on centerline cliques","year":2015,"dataSources":["eov4vbT6mnAiTpKji","knrZsDjSNHWtA9WNT"]}