{"title":"High-accuracy homography estimation for robust pan-tilt-zoom camera calibration","authors":"Jiaoru Yang, Rui Wang","doi":"10.1109/CIACT.2017.7977293","DOIUrl":null,"url":null,"abstract":"A robust and full automatic pan-tilt-zoom (PTZ) camera calibration of intrinsic parameters is proposed in this paper. This method elegantly combined the estimation of lens distortion coefficient and the homography given an image pair taken by the camera undergoing an arbitrary pan-tilt rotation in a fixed zoom. Addressing the problem that the feature-based homography estimation approaches loose the optimality in presence of feature location noise which was assumed as isotropic and identical distributed, we propose a better homography and lens distortion estimation by using covariance matrix weighted Ransac under lens distortion called as CWRLD in which the feature location noise is considered as anisotropic non identical distributed noise. The robustness and effectiveness of our method are demonstrated on both synthetic and real data.","PeriodicalId":218079,"journal":{"name":"2017 3rd International Conference on Computational Intelligence & Communication Technology (CICT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2017-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2017 3rd International Conference on Computational Intelligence & Communication Technology (CICT)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CIACT.2017.7977293","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 3
Abstract
A robust and full automatic pan-tilt-zoom (PTZ) camera calibration of intrinsic parameters is proposed in this paper. This method elegantly combined the estimation of lens distortion coefficient and the homography given an image pair taken by the camera undergoing an arbitrary pan-tilt rotation in a fixed zoom. Addressing the problem that the feature-based homography estimation approaches loose the optimality in presence of feature location noise which was assumed as isotropic and identical distributed, we propose a better homography and lens distortion estimation by using covariance matrix weighted Ransac under lens distortion called as CWRLD in which the feature location noise is considered as anisotropic non identical distributed noise. The robustness and effectiveness of our method are demonstrated on both synthetic and real data.