A framework for optimization of a multiresolution remote image retrieval system

Studies the tradeoffs involved in choosing the bit allocation in a multiresolution remote image retrieval system. Such a system uses a multiresolution image coding scheme so that a user accessing the database will first see a coarse version of the images and will be able to accept or discard a given image faster, without needing to receive all the image data. The authors formalize the problem of choosing the bit allocation (e.g., in the two resolution case, how many bits should be given to the coarse image and the additional information, respectively?) so that the overall delay in the query is minimized. They provide analytical methods to find the optimal solution under different configurations and show how a good choice of the bit allocation results in a significant reduction of the overall delay in the query (by up to a factor of two in some cases)