TY - JOUR
T1 - Wavelets for self-calibrated bundle block adjustment
AU - Ye, Jun Fu
AU - Tsay, Jaan Rong
N1 - Funding Information:
We sincerely appreciate the Ministry of Science and Technology (MOST), Taiwan, for the financial supports to our research projects for four years so far with the research grants MOST 105-2119-M-006-032, MOST 106-2119-M-006-023, MOST 107-2119-M-006-019 and MOST 108-2621-M-006-002. Also, we truly appreciate the GeoForce Technologies Co., Ltd, Taipei, Taiwan, for providing the aerial images taken with UltraCam Xp Wide Angle digital aerial camera over the calibration field for testing our wavelet model for camera self-calibration. Moreover, best thanks must be expressed to the National Land Surveying and Mapping Center (NLSC), Ministry of the Interior (MOI), Taiwan, for providing the ground target point data in the calibration field. Finally, we would like to express our sincere thanks to the blind reviewers of ISPRS 2020 for their valuable comments to this paper.
Publisher Copyright:
© 2020 International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives.
PY - 2020/8/6
Y1 - 2020/8/6
N2 - This paper entails a methodological novelty and builds upon prior research on a wavelets-based model for digital camera self-calibration. We introduce a new kernel function based on the compactly supported orthogonal third-order asymmetric Daubechies wavelet to correct systematic image distortion errors. Tests are done by using aerial images taken with a high-resolution metric digital aerial mapping camera. The quality of experimental results is evaluated by using reliable and high precision ground check points in the calibration field. For example, a four-fold block with this wavelet self-calibration model has the external accuracy of about 0.28 GSD (Combining double low lineground sampling distance) in the horizontal direction, and about 0.43 GSD in the vertical direction, respectively, where 1GSD & 4.6cm. The posterior standard deviations σ& 0 of unit weight are reduced from 0.37 pixel to 0.27 pixel. The residual vector lengths are also significantly reduced after our wavelet additional parameters are used. Experimental results support the proposal and demonstrate the applicability of this new model.
AB - This paper entails a methodological novelty and builds upon prior research on a wavelets-based model for digital camera self-calibration. We introduce a new kernel function based on the compactly supported orthogonal third-order asymmetric Daubechies wavelet to correct systematic image distortion errors. Tests are done by using aerial images taken with a high-resolution metric digital aerial mapping camera. The quality of experimental results is evaluated by using reliable and high precision ground check points in the calibration field. For example, a four-fold block with this wavelet self-calibration model has the external accuracy of about 0.28 GSD (Combining double low lineground sampling distance) in the horizontal direction, and about 0.43 GSD in the vertical direction, respectively, where 1GSD & 4.6cm. The posterior standard deviations σ& 0 of unit weight are reduced from 0.37 pixel to 0.27 pixel. The residual vector lengths are also significantly reduced after our wavelet additional parameters are used. Experimental results support the proposal and demonstrate the applicability of this new model.
UR - http://www.scopus.com/inward/record.url?scp=85091182636&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85091182636&partnerID=8YFLogxK
U2 - 10.5194/isprs-archives-XLIII-B1-2020-407-2020
DO - 10.5194/isprs-archives-XLIII-B1-2020-407-2020
M3 - Conference article
AN - SCOPUS:85091182636
SN - 1682-1750
VL - 43
SP - 407
EP - 414
JO - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
JF - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
IS - B1
T2 - 2020 24th ISPRS Congress - Technical Commission I
Y2 - 31 August 2020 through 2 September 2020
ER -