AUTOMATIC IMAGE COMPRESSION FEATURE EXTRACTION AND RECOGNITION

WE PROPOSE A STUDY TO EXPLORE THE POSSIBILITY OF DEVELOPING AN IMPROVED TECHNIQUE TO AUTOMATICALLY COMPRESS AN IMAGE AND EXTRACT FEATURES THAT REPRESENT THE IMAGE, OR PARTS OF IT, IN TERMS OF A FEW RELATIONSHIPS CHARACTERIZED BY FRACTALS. WE PLAN TO USE EXISTING TOOLS OF FRACTAL GEOMETRY LIKE HAUSDORFF DIMENSION, ATTRACTORS OBTAINED BY ITERATED FUNCTION SYSTEMS, FRACTAL INTERPOLATION FUNCTIONS TO MODEL AN IMAGE. SPECIFICALLY EDGE DETECTION AND/OR TEXTURE CLASSIFICATION WILL BE CONSIDERED. SIMULATION STUDIES WILL BE CARRIED OUT USING SYNTHETIC IMAGES. FURTHER, WE ASPIRE TO DEVELOP A PATTERN RECOGNITION SYSTEM FOR CLASSIFYING TEXTURES. THIS WILL BE USEFUL FOR THE IDENTIFICATION OF GEOMETRICAL STRUCTURES IN A NATURAL BACKGROUD, WHICH IS OUR PHASE II PROPOSAL.