New Dissimilarity Measures for Ultraviolet Spectra Identification Andr´ es Eduardo Guti´ errez-Rodr´ ıguez 1 , Miguel Angel Medina-P´ erez 1 , Jos´ e Fco. Mart´ ınez-Trinidad 2 , Jes´ us Ariel Carrasco-Ochoa 2 , and Milton Garc´ ıa-Borroto 1,2 1 Centro de Bioplantas. Carretera a Mor´ on km 9, Ciego de ´ Avila, Cuba 2 Instituto Nacional de Astrof´ ısica, ´ Optica y Electr´onica, Luis Enrique Erro No. 1, Sta. Mar´ ıa Tonanzintla, Puebla, M´ exico, C.P. 2840 {andres,migue,mil}@bioplantas.cu, {fmartine,ariel}@ccc.inaoep.mx Abstract. Ultraviolet Spectra (UVS) analysis is a frequent tool in tasks like diseases diagnosis, drugs detection and hyperspectral remote sens- ing. A key point in these applications is the UVS comparison function. Although there are several UVS comparisons functions, creating good dissimilarity functions is still a challenge because there are different sub- stances with very similar spectra and the same substance may produce different spectra. In this paper, we introduce a new spectral dissimilarity measure for substances identification, based on the way experts visu- ally match the spectra shapes. We also combine the new measure with the Spectral Correlation Measure. A set of experiments conducted with a database of real substances reveals superior results of the combined dissimilarity, with respect to state-of-the-art measures. We use Receiver Operating Characteristic curve analysis to show that our proposal get the best tradeoff between false positive rates and true positive rates. Keywords: Ultraviolet Spectra, Ultraviolet Spectra Comparisons Func- tions, Substance Identification, Dissimilarity Measures. 1 Introduction Ultraviolet Spectra (UVS) represent, for a given substance, the relation between ultraviolet light absorbance and light wavelength. Due to UVS are unique for each pure substance, they are frequently used for substance identification in dif- ferent areas such as medicine, geology, criminalistics, and industrial applications [1–4]. Identifying substances by UVS is a challenge because there are different sub- stances with very similar spectra shape (Fig. 1). Additionally, different concen- trations of the same substance produce different spectra, dilated or contracted, according to Lambert-Beer Law [5] (Fig. 2). In this paper, we focus on substance identification (mainly drugs, medicines, poisons, pesticides and other organic substances) by ranking its spectrum accord- ing to its dissimilarity values against the spectra in a database. These substances J.A. Carrasco-Ochoa et al. (Eds.): MCPR 2010, LNCS 6256, pp. 220–229, 2010. c  Springer-Verlag Berlin Heidelberg 2010