 |
.. (לתיקייה המכילה) | |
Matlab Part | |
| Please note that you should apply the K-term approximation on the zero-mean image, for all the considered transforms (i.e., DFT, DCT and KLT). |
Matlab Part: K-term approximation using DFT | |
|
Since you examine the DFT of a real signal, the coefficients should be conjugate symmetric. Therefore, keeping this property in the thresholding (i.e., selecting the K coefficients) will result in a real-valued reconstruction. You can obtain this by considering only even K values for the DFT-based approximation. Note 1: Reconstructed image-signal with complex values is not a valid result. Note 2: For the DCT and KLT you should evaluate also the odd K values. |
Question #1 | |
| Please note that a typo was fixed, as you need to prove that E[y^T e]=0. |
Matlab Part: The 'Peppers' Image | |
|
Some of you noticed that their Matlab's buit-in 'Peppers' image is invalid (e.g., white margins, etc.). You may, however not required, to use the 'Peppers' image that is provided in the Homework #4 section under the assignment tab (you just need to transform it to grayscale). |