This award will enable Dr. Huseyin Abut, San Diego State University, to collaborate with Dr. Oguz Manas, Aegean University, Izmir, Turkey. They will study applications of vector and matrix quantization to image coding, image classification, image enhancement, and recognition problems in digital signal processing. They will also establish a signal processing graduate course and a laboratory in Izmir. The researchers' goals are to extend linear prediction theory to include low-rate image coding and image classification problems. They will reformulate autoregressive analysis and windowing techniques to cover multidimensional signals. They propose to use information from the multidimensional signal studies to do image coding below 0.5 bits per pixel, and to apply vector and matrix linear prediction concepts to image classification and segmentation problems on which there has been relatively little research. The tools of digital signal processing developed in this research have applications in studying speech, seismic activity, imaging and machine vision, and in designing aids for the handicapped. This award will help the researchers continue their collaboration on image classification and recognition using vector linear prediction coefficients.
