Video Segmentation Using Fast Marching and Region Growing Algorithms
- PDF / 3,469,386 Bytes
- 10 Pages / 600 x 792 pts Page_size
- 45 Downloads / 209 Views
Video Segmentation Using Fast Marching and Region Growing Algorithms Eftychis Sifakis Department of Computer Science, University of Crete, P.O. Box 2208, Heraklion, Greece Email: [email protected]
Ilias Grinias Department of Computer Science, University of Crete, P.O. Box 2208, Heraklion, Greece Email: [email protected]
Georgios Tziritas Department of Computer Science, University of Crete, P.O. Box 2208, Heraklion, Greece Email: [email protected] Received 31 July 2001 The algorithm presented in this paper is comprised of three main stages: (1) classification of the image sequence and, in the case of a moving camera, parametric motion estimation, (2) change detection having as reference a fixed frame, an appropriately selected frame or a displaced frame, and (3) object localization using local colour features. The image sequence classification is based on statistical tests on the frame difference. The change detection module uses a two-label fast marching algorithm. Finally, the object localization uses a region growing algorithm based on the colour similarity. Video object segmentation results are shown using the COST 211 data set. Keywords and phrases: video object segmentation, change detection, colour-based region growing.
1.
INTRODUCTION
Video segmentation is a key step in image sequence analysis and its results are extensively used for determining motion features of scene objects, as well as for coding purposes to reduce storage requirements. The development and widespread use of the international coding standard MPEG-4 [1], which relies on the concept of image/video objects as transmission elements, has raised the importance of these methods. Moving objects could also be used for content description in MPEG-7 applications. Various approaches have been proposed for video or spatio-temporal segmentation. An overview of segmentation tools, as well as of region-based representations of image and video, are presented in [2]. The video object extraction could be based on change detection and moving object localization, or on motion field segmentation, particularly when the camera is moving. Our approach is based exclusively on change detection. The costly and potentially inaccurate motion estimation process is not needed. We present here some relevant work from the related literature for better situating our contribution.
Spatial Markov Random Fields (MRFs) through the Gibbs distribution have been widely used for modelling the change detection problem [3, 4, 5, 6, 7, 8]. These approaches are based on the construction of a global cost function, where interactions (possibly nonlinear) are specified among different image features (e.g., luminance, region labels). Multiscale approaches have also been investigated in order to reduce the computational overhead of the deterministic cost minimization algorithms [7] and to improve the quality of the field estimates. In [9], a motion detection method based on an MRF model was proposed, where two zero-mean generalized Gaussian distributions were used to model the interf
Data Loading...
