Multiresolution analysis and chemometrics for pattern enhancement and resolution in spectral signals and images

The chapter illustrates the benefits and improvements of the integration of the wavelet transform with multivariate data analysis upon multiresolution analysis. This approach can be used for feature extraction both in signals and images in a broad sense, focusing on the capability to simultaneously accomplish de-noising and feature enhancement / selection. Different contexts are presented, ranging from feature selection applied to spectroscopic signals in classification and regression tasks, to multiresolution multivariate image analysis with special attention to quality monitoring, fault detection and classification. The proposed cases of study cover applications in the food and materials sciences.