Image Transforms For Data Compression
The goal of this computer assignment is to study and compare two different image transforms, namely 2-D DCT and DWT, for the purpose of image data compression. The “Pepper” image (see course web page under Images) is used for this assignment.
1. DCT Approach: Partition the image into non-overlapping blocks and apply DCT to each block. Use at least two different blocks sizes, e.g., 8x8 and 16x16. Devise a scheme to reduce the blocked data in the DCT domain using an energy-based or another ‘meaningful’ criterion (e.g., histogram of the DCT coefficients). For each block-size and data reduction criterion, reconstruct the image from the retrained coefficients and display the results. Compare the results of your reconstructions using performance measures such as distortion or signal-to-
2
σo ) where σo2 is the variance of the original image noise ratio (use SNRdB =10log10(2 σe
and σe2 is the variance of the error image) as well as visual evaluation of the results.
2. DWT Approach: Apply 2-D DWT to the same image and display the results for two and three-level decompositions and two different choices of wavelets e.g., Daubechies (db), Symlet (sym),), etc. Present the histograms of the sub-images and compare the data reduction ability of the two transforms in Parts 1 and 2. Compare the best results of the DWT with that of the DCT for the cases studied.
3. Provide a detailed discussion on the effectiveness of this transform for image data reduction and representation .
