Spatial object detection in JPEG bitstreams

Because of the large amount of raw data involved, images are generally compressed before storage or transmission. If one were interested in detecting and localizing spatial objects of interest, it would be considerably more efficient to do so in the compressed domain because less data would need to be processed and the computations required to decode the image would be avoided. In this paper, we study detection and localization in the bitstream of JPEG-compressed imagery and compare these new results with our earlier work in which detection was performed in the DCT domain.