Motion Differential SPIHT for Image Sequence and Video Coding

Efficient image sequence coding exploits both intraand inter-frame correlations. SPIHT is efficient in intra-frame decorrelation for still images. Based on SPIHT, differential-SPIHT removes inter-frame redundancy by re-using the significance map of a SPIHT coded frame. The motion differential SPIHT (MD-SPIHT) automatically decides the coding methods for each frame, according to the inter-frame correlations in terms of PSNR. It can be seen as coding correlated sources (frames) into common and private messages, where the common message is the re-used significance map. From the simulation results of hyperspectral images, 3D/4D medical images, and video, MD-SPIHT is more efficient, compared with conventional 2D SPIHT and motion JPEG2000, while retaining the desirable features of random frame access, flexible frame rates, scalable bit rates, parallel coding, and computational simplicity.