Download presentation
Presentation is loading. Please wait.
Published byTara Hersey Modified over 10 years ago
1
Bilateral Filter Wojciech Korycki wojciech.korycki@gmail.com 23.06.2010 Wrocław Implementations compare
2
Bilateral Filter- introduction Bilateral filtering will preserve edges but… To high parameters, that is R = MAX, might change bilateral filter to normal Gaussian filter. (MAX- usually 255 or 1)
3
Bilateral Filter- implementations Application implemetations Bilateral Filter Normal Bilateral Filter CPU CPU (indexing) Fast approximation CUDA Cross Bilateral Filter CPU CPU (indexing) Fast approximation CUDA
4
Bilateral Filter- testing Test images
5
Bilateral Filter- testing Processing time tabble for 8-bit gray scale image Image Lena.bmp (512x512, 8-bit Gray Scale) Bilateral Filter Processing type Parameters SR SR S R 3107 11 10 t [s] CPU2,0236,82114,394 CPU (indexing)0,6941,1822,091 Fast Approximation0,1410,1370,201 CUDA0,0190,0510,108
6
Bilateral Filter- testing Processing time diagram for 8-bit gray scale image
7
Bilateral Filter- testing Visual results
8
Bilateral Filter- testing Processing time tabble for 24-bit RGB image image Iwonka.bmp (24-bit RGB) Bilateral Filter Processing type Parameters SR SR S R 3107 11 10 t [s] CPU2,3568,1217,728 CPU (indexing)0,9752,0854,11 Fast Approximation0,5380,3390,444 CUDA0,0260,080,174
9
Bilateral Filter- testing Processing time diagram for 24-bit RGB image
10
Bilateral Filter- testing Visual results
11
Bilateral Filter- testing Processing result
12
Bilateral Filter- summary General profit in CUDA processing (time in [s])
13
Bilateral Filter- addons Result of Cross Bilateral Filter processing + =
14
Bilateral Filter- addons Iterations= 10 iterations= 20
15
Bilateral Filter- side effects Fog effect
16
Bilateral Filter- side effects Edges blur
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.