100 likes | 573 Views
Embedded Image coding using zero-trees of Wavelet Transform. Authors: Harish Rajagopal Brett Buehl. Project Overview. Purpose: create a software implementation of EZW coder for image compression Overall process flow-chart:. Step 1: Wavelet Transform.
E N D
Embedded Image coding using zero-trees of Wavelet Transform Authors: Harish Rajagopal Brett Buehl
Project Overview • Purpose: create a software implementation of EZW coder for image compression • Overall process flow-chart:
Step 1: Wavelet Transform • Step 1a: Replace each image row with its 1D DWT • Step 1b: Replace each image column with its 1D DWT • Step 1c: Repeat steps (1) and (2) on the lowest subband to create the next scale • Step 1d: Repeat step (3) until the desired number of scales has been created Original Image Discrete Wavelet Transform
Step 2: EZW Coding • Flow-chart for EZW encoding: (See next slide for more details)
EZW Coding Cont’d • The EZW consists of two algorithms: 1. Dominant Pass Flow-Chart 2. Subordinate Pass Flow-Chart
Step 3: Arithmetic Encoding • Uses symbol probabilities for lossless compression • Added to EZW implementation to increase compression • Encodes output from EZW encoder • P,N,T,Z,E or 0,1,E Symbol Probabilities & subintervals Process of encoding a1a2a3a4
Results: • Compression ratios similar to JPEG image compression • Demonstration image results: • cameraman.bmp – 16.8:1 compression ratio • lena.bmp – 7.5:1 compression ratio • barbara.bmp – 7.5:1 compression ratio
Image Demonstration: Original Image Reconstructed image Difference Image Initial DWT Reconstructed DWT