A Stereoscopic Image Hole Filling Based on Hole-Classification and Fill Route
2013
Hochschulschrift
Zugriff:
101
The so-called 2D to 3D conversion technique is to use the original 2D image together with the distance information of depth map to move the pixel of the front object, and then the disparity could be created on human eye. Through such way, it can be converted into left and right eye image, and 3D stereoscopic image can be viewed. However, after left and right viewing angles are generated, the shift of pixel of the original image will change along with different depth, and the exposed part after pixel shift is then called disoccluded area. The border of the object of interest in the image that belongs to disoccluded area might not have pixel information, which is then called hole. The main objective of this research is to use fast and effective way to fill the hole generated by these disoccluded areas. First, the original depth map and the original image are made with the same shift action, in the meantime, the shift depth map is taken with simple depth map filling process so as to know correct location and range of each object. Based on this, the computation of filling object in the original image can be reduced, and the filling speed can then be speeded up. In addition, during the hole filling process, since the border of the disoccluded area does not contain correct pixel information needed for the filling process, it usually causes a great difference between the filled result and the real situation. Consequently, the filled result figure usually looks unreasonable. In order to solve such problem, filling method adopting a different route is used to reduce the generation of mistake. Next, these disoccluded areas are divided into three types, and three different filling routes and filling methods are applied according to these three types of disoccluded areas. Through the method provided in this paper, a more real-time way can be used to generate a left and right eye viewing image with fewer mistakes. From the experimental results, it can be seen that holes that are mistakenly filled have been done with better corrections, in the meantime, it becomes more natural and closer to the reality when it is viewed.
Titel: |
A Stereoscopic Image Hole Filling Based on Hole-Classification and Fill Route
|
---|---|
Autor/in / Beteiligte Person: | Loo, Mu-Di ; 盧慕迪 |
Link: | |
Veröffentlichung: | 2013 |
Medientyp: | Hochschulschrift |
Sonstiges: |
|