Assignment 3: Tone mapping using bilateral filter

 

1. The following three pictures are: original Einstein photo, contaminated by AWGN with a variance of 0.0025 (26dB), recovered by implemented bilateral filter.

                           

             1.a. Original Einstein                                      1.b. Contaminated Einstein                                        1.c. Filtered Einstein

 

2. The Gaussian filter is adopted to generate the coarse part in tone mapping procedure. The final result is shown as follows:

                             2.a. Tone mapping result using Gaussian filter 

The halo effect, which is caused by the fact that Gaussian Filter cannot tell the object boundaries is obvious from this image.

3. If we replace the Gaussian filter with bilateral filter, the halo will be largely canceled since bilateral filter take object boundary into account.

                             3.a. Tone mapping result using bilateral filter 

4. My HDR picture consists of ten photo of different exposure time. The optimal LDR image is shown as follows, which is compared with a less desirable compressed counterpart.

                                               4.a. Optimized image                                                                        4.b. Compressed image with significant halo effect

The tone mapping and bilateral filter MATLAB code can be found: here.