Rate distortion optimal bit allocation methods for volumetric data using JPEG 2000

Area of Science:

• Computer Science
• Data Compression
• Scientific Visualization

Background:

• Three-dimensional (3-D) data from computer modeling and sensors are large and require efficient compression.
• The JPEG 2000 standard offers a method for compressing volumetric data by processing it slice by slice.
• JPEG 2000 lacks a defined method for allocating bits across these data slices.

Purpose of the Study:

• To propose and evaluate new bit allocation algorithms for JPEG 2000 compression of 3-D volumetric data.
• To address the challenge of optimal bit allocation among separate slices in 3-D data compression.
• To compare the performance and computational complexity of proposed algorithms against optimal methods.

Main Methods:

• Developed a rate distortion optimal (RDO) bit allocation algorithm for mean squared error, analogous to JPEG 2000's internal optimization.
• Introduced the mixed model (MM) approach, which uses a two-region mathematical model for slice rate distortion curves to improve low bit-rate accuracy.
• Applied both algorithms to a 3-D meteorological dataset for evaluation.

Main Results:

• The proposed algorithms effectively compress 3-D volumetric data using the JPEG 2000 standard.
• The mixed model (MM) approach demonstrated distortion results nearly identical to the computationally intensive RDO method.
• The MM approach significantly reduced the computational complexity compared to the optimal method.

Conclusions:

• The mixed model (MM) provides an efficient and effective solution for bit allocation in JPEG 2000 compression of 3-D data.
• This approach offers a practical alternative for compressing large 3-D datasets, particularly in fields like meteorology.
• The study highlights the potential of tailored bit allocation strategies to enhance volumetric data compression.