A random object-oriented population study of household paints measured by infrared spectroscopy

Forensic paint analysis heavily relies on population studies and creation of databases for estimating frequencies of apparition and perform discrimination studies. An important part of the paint market is currently led by household paints, that can be recovered in cases of burglaries of forced entries. Here, we present a database of 166 household paint samples collected at random in a waste recycling plant, and measured by infrared spectroscopy. We provide general demographic information on the paint such as their type, brand, model, and color. Based on their infrared spectra, we present the differentiation among samples and the calculation of the discriminating power for various attributes. We compare our visual discrimination results with a principal component analysis. This study provides valuable information for paint evaluation and association evidence, but will also be used to create a searchable database to tentatively determine the chemical composition, brand, or model of a questionned specimen in absence of reference material.