β Scribed by Gerald Schaefer
No coin nor oath required. For personal study only.
Image retrieval and image compression are both fields of intensive research. As lossy image compression degrades the visual quality of images and hence changes the actual pixel values of an image, low level image retrieval descriptors which are based on statistical properties of pixel values will change, too. In this article we investigate how image compression affects the performance of lowβlevel colour descriptors. Several image retrieval algorithms are evaluated on a speciated image database compressed at different image quality levels. Extensive experiments reveal that while distributionβbased colour descriptors are fairly stable with respect to image compression a drop in retrieval performance can nevertheless be observed for JPEG compressed images. On the other hand, after application of JPEG2000 compression only a negligible performance drop is observed even at high compression ratios. Β© 2008 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 18, 101β112, 2008
π SIMILAR VOLUMES
In this contribution we investigate the potential influence between assessors' perceived completion of their work task at hand and their actual assessment of usefulness of the retrieved information.The results indicate that the number of useful documents found by assessors does not influence their p
The results of three experiments suggest that a memory trace for an event is not altered by witnessing similar events\ but that postevents can interfere with its retrieval[ On an immediate recall test\ details from an original story "e[g[ wrench# were recalled less often if a subsequent story mentio