1: \begin{abstract}

2: In the big data era, one has often  to conduct the problem of parcimonious data representation. 

3: In this paper, the data under study are curves and the sparse representation stands on a semi-parametric model. 

4: Indeed, we propose an original registration model for noisy curves. The model is built transforming an unknown function by plane similarities.  We develop a statistical method that allows to estimate the parameters characterizing  the  plane similarities. The properties of the statistical procedure is studied. We show the convergence and the asymptotic normality of the estimators. Numerical simulations and a real life aeronautic example  illustrate and demonstrate the strength of our methodology.\\

5: 

6: 

7: \noindent

8: \begin{it}Keywords:\end{it} Semiparametric model, Registration of curves, Statistical learning of a physical system \\

9: 

10: 

11: % \PACS{PACS code1 \and PACS code2 \and more}

12: \noindent

13: \begin{it} MSC Classification:\end{it} 62F12, 62F30,  62P30

14: \end{abstract}

15: