Quantitative method of evaluating for recognize people, based on thermograms
G. Owczarek, Robert Strąkowski, M. Więcek · Proceedings of the 2010 International Conference on Quantitative InfraRed Thermography · 2010
This article present quantitative method of evaluating for recognize people, based on thermograms.The main purpose of the research was to write a computer program for quantitative method of evaluating for recognize people, based on thermograms.The application makes a selection of discrimination properties from thermal image -the analyze is provided for I level statistical parameters obtained from luminance histogram and II level statistical parameters obtained from: cooccurrence matrix parameterised by offset (4 directions and 3 neighborhoods), luminance matrix (for mask 3x3 5x5), Run Length Matrix, wavelet transform.The best discrimination parameters are selected by Fisher-Tippett.The artificial neural network (ANN) is used to determine the number of people from thermograms.