Uma metodologia para validação fotométrica em sistemas interativos veiculares baseada em inteligência computacional

Abstract

This work proposes a methodology for automatically validating the internal lighting system of an automobile, by assessing the visual quality of an instrument cluster (IC) based on the user's perceptions. Although the visual quality assessment of an instrument is a subjective matter, it is inuenced by some photometric features of the instrument, such as the light intensity distribution. In this work a methodology aiming to identify and quantify non homogeneous regions in the lighting distribution of these instruments, starting from a digital image, is presented. In order to accomplish this task, a set of 107 digital images of some gauges (speedometer, tachometer, temperature and fuel) was acquired and preprocessed. The same instruments were evaluated by users to identify their non-homogenous regions. Then, for each instrument region, we extracted a set of homogeneity descriptors. It is also proposed in this work, a relational descriptor to study the homogeneity inuence of a region in relation with the others regions in the gauges. These descriptors were associated with the results of the user labeling, and given to two machine learning algorithms (Articial Neural Network - ANN and Support VectorMachine - SVM). These algorithms were trained to identify a region as being homogeneous or not. The work also accomplished a meticulous analysis of the user and the specialist evaluation. After the analysis of the results, the proposed methodology obtained a precision above 94%, for both regions and nal classications.