Browsing and Placement of Multiresolution Images on SecondaryStorage

With the rapid advances in computer and communication technologies, there is anincreasing demand to build and maintain large image repositories. In order toreduce the demands on I/O and network resources, multiresolutionrepresentations of images are being proposed for the storage organization ofimages. Image decomposition techniques such as {\\em wavelets} can be used toprovide multiple resolution images. The image is represented by severalcoefficients, one of them with visual similarity to the original image but at alower resolution. Thus, these visually similar coefficients can be thought ofas the {\\em thumbnails} or {\\em icons} of the original image. This paperaddresses the problem of storing these multiresolution coefficients on disk(s)so that thumbnail browsing as well as image reconstruction can be performedefficiently. Several strategies are evaluated to store the image coefficientson parallel disks. These strategies can be classified into two broad classesdepending on whether the browsing pattern of the images is used in theplacement of image coefficients. Disk simulation is used to evaluate theperformance of these strategies. The data used in the simulation are of twotypes: the texture samples from the Brodatz collection as well as 50,000 reallandsat images. The results indicate that significant performance improvementscan be achieved with as few as four disks by placing image coefficients basedupon the browsing access patterns.