Wavelet Transform Method

Wavelet transform analysis is a powerful tool to understand phenomena with mutiresolutional nature. Its advantage has been deeply exploited in many applications. My research in WT focus on the following two aspects:

(1) Find appropriate wavelet basis for computer graphics and scientific visualization applications.

(2) Analyze how WT affects various visualization and graphics rendering methods. Further more, we can exploit the advantage of the wavelet representation to improve the performance of these algorithms.

Biorthogonal WT ( 5k )

Biorthogonal wavelet is a family of symmetric compactly supported wavelet basis. Because of its interesting properties, it is suitable for data reduction in computer graphics and visualization applications. For more detailed information, see paper on IEEE visualization'94 (see publications).

Scalar field ( 35k )

Apply the BWT to the thickness data of Hurburt's ocean model. The compression ratio is 50:1 ( upper one is the original data ). For a frame data with 468x337 floating point samples, the speed of reconstruction is 10 frame/sec on SGI Indigo2.

Vector field & stream lines ( 101k )

Apply the BWT to the velocity field of Hurburt's ocean model. The compression ratio is 50:1 ( upper one is the original data ).

Vector field & topology ( 134k )

Topology of 2D velocity data has been extracted both from the original data and the data reconstructed from 1/4 (middle) and 1/16 (right) of BWT coefficients. The global structures remain unchanged, even at a high compression ratio.

Go back