Web based progressive transmission for browsing remotely sensed imagery

This paper describes an image representation technique that entails progressive refinement of user specified regions of interest (ROI) of large images. Progressive refinement to original quality can be accomplished in theory. However, due to heavy burden on storage resources for the authors´ applications, they restrict the refinement to about 25% of the original data resolution. A wavelet decomposition of the data combined with scalar and vector quantization (VQ) of the high frequency components and JPEG/DCT compression of low frequency component is used as representation framework. Their software will reconstruct the region selected by the user from its wavelet decomposition such that it fills up the preview window with the appropriate subimages at the desired resolution, including full resolution stored for preview. Further refinement from the first preview can be obtained progressively by transmitting high frequency coefficients from low resolution to high resolution which are compressed by variant of vector quantization called model-based VQ (MVQ). The user will have an option for progressive build up of the ROIs until full resolution stored or terminate the transmission at any time during the progressive refinement