Compression Techniques
INTRODUCTION
This research describes an image representation technique that
entails progressive refinement of user specified regions of
interest (ROI) of large images. Progressive refinement of
original quality can be accomplished in theory. However, due to
heavy burden on storage resources for our applications, we
restrict the refinement to about 25% of the original data
resolution. Wavelet decomposition with Vector Quantization (VQ)
of the high frequency components and JPEG/DCT compression of low
frequency component is used as representation framework. Our
software will reconstruct the region selected by the user from
its wavelet decomposition at desired resolution. 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 Vector Quantization. The user
will have an option of progressive build up of the ROI?s until
full resolution stored or terminate
the transmission at any time during the progressive refinement.
IMAGE COMPRESSION
The entire architecture of the program is based on object
oriented programming using C++. A multiresolution decomposition
into wavelet coefficients provides the most useful image
representation or image browsing. Image decomposition is done
recursively into wavelet coefficients using the technique [2]
shown in Figure 1. The given image is decomposed into low and
high frequency bands along the rows and columns.
VECTOR QUANTIZATION
The high frequency subbands
image, high, frequency, coefficients, wavelet, vector, using, resolution, quality, progressive, hvs, hl, hh, user, technique, subbands, refinement, reconstruction, quantization, program, mvq, low, lh, first, decomposition, compressed, components, based, vq, vectors, vector, quantization, variant, subband, shown, scalar, roi, representation, quantized, model, ll, jpeg/dct, figure, distribution, component, codebook, called, bands, about, zeros, due, zero
