Scaled total-least-squares-based registration for optical remote sensing imagery
- 255 Downloads
In optical image registration, the reference control points (RCPs) used as explanatory variables in the polynomial regression model are generally assumed to be error free. However, this most frequently used assumption is often invalid in practice because RCPs always contain errors. In this situation, the extensively applied estimator, the ordinary least squares (LS) estimator, is biased and incapable of handling the errors in RCPs. Therefore, it is necessary to develop new feasible methods to address such a problem. This paper discusses the scaled total least squares (STLS) estimator, which is a generalization of the LS estimator in optical remote sensing image registration. The basic principle and the computational method of the STLS estimator and the relationship among the LS, total least squares (TLS) and STLS estimators are presented. Simulation experiments and real remotely sensed image experiments are carried out to compare LS and STLS approaches and systematically analyze the effect of the number and accuracy of RCPs on the performances in registration. The results show that the STLS estimator is more effective in estimating the model parameters than the LS estimator. Using this estimator based on the error-in-variables model, more accurate registration results can be obtained. Furthermore, the STLS estimator has superior overall performance in the estimation and correction of measurement errors in RCPs, which is beneficial to the study of error propagation in remote sensing data. The larger the RCP number and error, the more obvious are these advantages of the presented estimator.
KeywordImage registration Polynomial regression model Error-in-variables model Ordinary least squares Scaled total least squares Singular value decomposition
This research was supported in part by the Knowledge Innovation Program of the Chinese Academy of Sciences (Grant No. KZCX2-EW-QN303) and the National Natural Science Foundation of China (Grant No. 40971222). The authors are grateful to two anonymous referees for their constructive comments, which helped to improve the quality of the paper.
- Evans FH (1998) Statistical methods in remote sensing. In: Proceedings of the 3rd National Earth Resource Assessment workshop, Brisbane, Australia, 16-18, Nov., pp 1-26Google Scholar
- Gillard JW (2006) An historical overview of linear regression with errors in both variables. Technical Report, Cardiff School of Mathematics, UK, Available at http://www.cardiff.ac.uk/maths/resources/Gillard_Tech_Report.pdf
- Jensen JR (1996) Introductory Digital Image Processing: a remote sensing perspective. Prentice Hall, New JerseyGoogle Scholar
- Lunetta RS, Congalton RG, Fenstermaker LK, Jensen JR, McGwire KC, Tinney LR (1991) Remote sensing and geographic information system data integration: error sources and research issues. Photogramm Eng Rem S 57(6):677–687Google Scholar
- Markovsky I (2010) Bibliography on total least squares and related methods. Statistics and Its Interface 3(3):329–334Google Scholar
- Paige CC, Strakoš Z (2002b) Unifying least squares, total least squares and data least squares. In: Van Huffel S, Lemmerling P (eds) Total least squares and errors-in-variables modeling, Kluwer Academic Publishers, pp 25–34Google Scholar
- Rao BD (1997) Unified treatment of LS, TLS and truncated SVD methods using a weighted TLS framework. In: Van Huffel S (ed) Recent advances in total least squares techniques and errors-in-variables modelling, SIAM, pp 11–20Google Scholar
- Richards JA, Jia XP (1999) Remote Sensing Digital Image Analysis: an introduction, 3rd edn. Springer, Berlin, New YorkGoogle Scholar
- Van Huffel S (ed) (1997) Recent advances in total least squares techniques and errors-in-variables modeling. SIAM, PhiladelphiaGoogle Scholar
- Van Huffel S, Lemmerling P (eds) (2002) Total least squares and errors-in-variables modeling: analysis, algorithms and applications. Kluwer Academic Publishers, DordrechtGoogle Scholar
- Wansbeek T, Meijer E (2000) Measurement error and latent variables in econometrics. Elsevier, New YorkGoogle Scholar