Publications / 2000 Proceedings of the 17th ISARC, Taipei, Taiwan
Multi-resolution pattern classification (MPC) is an image segmentation technique that extracts features from an image and segments the image into several featured areas. Two resolution levels are contained in the MPC approach, the fine resolution level and the coarse resolution level. Pre-selected features, such as means and variances, will be computed, and then clustered using a linear or nonlinear classifier. This paper introduces the MPC approach to process steel bridge painting images with the hope of obtaining better segmentation results on rust identification.