Artificial neural network models were developed to classify road pavement types into the transverse-tined, the longitudinal-tined, NGCS(Next Generation Concrete Surface), Diamond Grinding, and Stone Mastic Asphalt by utilizing tire-pavement noise and road surface images. Tire-pavement noise data were collected by OBSI(On-Board Sound Intensity) method, and analyzed to obtain sound intensity level, sound pressure level, and sound quality indices. Road surface image data was analyzed through image feature extraction algorithms of Hough transformation and HOG(Histogram of gradient). The important features among the acoustic and image characteristics were selected by a random forest model. The acoustic features selected by the random forest algorithm are the overall sound intensity level of 400~5kHz 1/3-octave bands, the sound intensities (W/m2) of 800~2kHz 1/3-octave bands, loudness, fluctuation strength and tonality. The image features selected are the number of longitudinal lines extracted from Hough transform algorithm and HOG of the central cell. The two groups of the selected features were applied separately or together to an artificial neural network model to find classification performance. The classification accuracy rates of the models using acoustic features only, image features only and both acoustic and image features combined were 90.8%, 88.8%, and 97.3%, respectively.