An alternative approach to spectrum base line estimation

Abstract

We present a new form of merit function which measures agreement between a large number of data and the model function with a particular choice of parameters. We demonstrate the efficiency of the proposed merit function on the common problem of finding the base line of a spectrum. When the base line is expected to be a horizontal straight line, the use of minimization algorithms is not necessary, i.e. the solution is achieved in a small number of steps. We discuss the advantages of the proposed merit function in general, when explicit use of a minimization algorithm is necessary. The hardcopy text is accompanied by an electronic archive, stored on the SAE homepage at http://www1.elsevier. com/homepage/saa/sab/content/lower.htm. The archive contains fully functional demo program with tutorial, examples and Visual Basic source code of the key subroutine. © 2005 Elsevier B.V. All rights reserved.