 |
Technical Program
Paper Detail
| Paper: | TA-P7.9 |
| Session: | Image and Video Filtering |
| Time: | Tuesday, October 10, 09:40 - 12:20 |
| Presentation: |
Poster
|
| Title: |
VARIATIONAL TEXTURED IMAGE DECOMPOSITION WITH IMPROVED EDGE SEGREGATION |
| Authors: |
Reza Shahidi; Memorial University of Newfoundland | | | | | Cecilia Moloney; Memorial University of Newfoundland | | |
| Abstract: |
Image decomposition consists of splitting an image into two or more components. One component is piecewise smooth and models the objects in the image. Another consists of the texture in the image, possibly including some noise. Decomposition is useful for many image processing tasks, e.g. texture segmentation and image inpainting. In this paper, we consider ways of improving both the speed and quality of decomposition using Meyer's well-known variational approach of Meyer by adding extra regularization terms. A measure of improved quality of image decomposition found in the literature is the lack of cartoon edges in the texture component. In this paper, a method called Improved Edge Segregation is introduced, which ensures this quality measure is high. When combined with active contour texture discrimination, improved results are obtained over conventional methods. |
|
