| Summary: | 碩士 === 國立中正大學 === 電機工程所 === 96 === When we want to design a digital filter circuit, the rounding error between floating point and fixed point calculation is an important issue for hardware designers to consider before they decide the specification. The topic of this thesis will focus on the analysis of the precision, which is the discrete wavelet transform calculation for JPEG2000. Besides, we will propose a methodology algorithm that has a flexible fixed point calculation feature, and propose the flexible hardware design concepts for the algorithm. We implement the algorithm with Matlab tool, and implement the hardware concepts with Hspice simulation.
The goal of this thesis is that we try to propose a methodology to estimate the rounding error between the different fixed point calculation efficiently, and assist the designer to adjust a flexible discrete wavelet transform circuit in different conditions.
|
|---|
