Expert system for bus styling evaluation based on neural network and feature extraction

This paper firstly examines the traditional vehicle styling evaluation methods and issues, and then presents a new approach which uses ANN (artificial neural network) to build an expert system for bus styling evaluation. It describes the key technical issues of quasi-three-dimensional bus styling evaluation expert system from data collection, graphical pre-processing, graphics feature extraction, knowledge base construction to neural network model design and training. On this basis, authors have developed a bus styling evaluation expert system software, and satisfactory results are obtained in actual bus styling retrofit project.