Browsing and Transmission Optimization of Multiresolutuion Multiview Image Retr ieval System In Distr ibuted Network Environments

* This work is supported in part by KOSEF, Korea, under the grant KOSEF971-0919-119-2 Ab st r a ct Problems of transmission delay caused by shared network or low speed network, become serious in case of making a fast decision for target image. In distributed network environments, trade-offs are needed between user and system when retrieving large volume of multiview image DBs in the aspect of computer processing power and transmission delay for service quality control. In this paper browsing and transmission optimization techniques are proposed for fast retrieval of multiview image DB by the multiresolution(wavelet) image modeling and coding based on progressive transmission. Interactive user-system using multiresolution representation has shown much better performance in transmission delay minimization over the single resolution image retrieval system. The proposed network model was analyzed and evaluated for system’s performance improvement. The paper also describes multiresolution epipolar plane image modeling for fast representation and interpolation of multiview image sets To this end structure (trace line) and texture information are extracted from coarse version of a multi-view image set using the Viterbi optimization method and then this information is exploited in order to progressively synthesize, compress and display a multi-view image set. Experimental results that verify the proposed algorithm have shown that about 38 dB PSNR at a center-view image was obtained when compressed a multi-view image set consisting of 5-view images by about 30:1.