Algoritmos de extracción de características a partir de imágenes de resonancia magnética para evaluar parámetros de calidad en productos cárnicos mediante minería de datos

Abstract

This PhD thesis proposes a new methodology to determine the quality characteristics of meat products (Iberian loin and ham) in a non-destructive way, by using of Magnetic Resonance Imaging (MRI) and computer vision algorithms. Firstly, MRI are obtained from meat products, evaluating three acquisition sequence (Spin Echo (SE), Gradient Echo (GE) and Turbo 3D (T3D)). Later, MRI is analyzed by applying different texture (GLCM, GLRLM and NGLDM) and fractals algorithms (CFA, FTA and OPFTA); the last two have been developed in this PhD thesis. These algorithms extract texture features from the MRI. At the same time, the meat products are also analyzed by means of physico-chemical and sensory techniques. Finally, different data mining techniques are applied on all obtained data: deductive (Multiple linear regression, MLR), classification (Decision trees, DT and Rules-based systems, SBR) and prediction techniques (MLR and Isotonic regression, IR). The accuracy of the analysis of quality parameters is affected by the MRI acquisition sequence, the algorithm used to analyze them and the data mining technique applied. In general, It could be indicated the use of SE as MRI acquisition sequence, and GLCM or OPFTA as image analysis algorithm. Considering the data mining techniques, MLR and DT are appropriate, respectively, to deduce physico-chemical parameters and to classify as a function of salt content. Regarding to the predictive technique, MLR could be indicated. It offers reliable equations to determine the quality parameters, and, allows analysing the quality of meat products in a non-destructive, efficient, effective and accurate way.