Citation Context

...gion. Moreover, spatial information provides important hints of object boundaries. Methods that combine an initial intensity segmentation with motion information have been proposed recently [12] [15] =-=[20]-=-. Given an oversegmentation of the current frame, objects are formed by merging together segments with spatio-temporal similarity. The region merging approaches have two disadvantages. Firstly, the in...

Citation Context

.... On the other hand, spatial segmentation can provide accurate boundaries of regions, but it is difficult to detect moving objects. The well-known watershed approach is an important region-based tool =-=[9, 10]-=-. The conventional watershed algorithm has the problem of over-segmentation. In fusion of spatial and temporal segmentation results, the threshold which determines a region obtained in spatial segment...

Citation Context

...ntation. I. INTRODUCTION VIDEO segmentation is a key technique for semantic object extraction and plays an important role in content-based video coding in MPEG4 [4], [5], [11]–[13], [15], [16], [18], =-=[22]-=-. It is often necessary to separate moving objects from their shadows in video segmentation for many applications such as traffic monitoring, surveillance and video conferencing [17]. There are some d...

Citation Context

...eillance system, the object location and shape information are important for object behavior analysis, object recognition, and video indexing. Several video segmentation algorithms have been proposed =-=[4, 5, 7, 12, 13, 19, 20, 27, 29, 30, 32, 35]-=-. Most of these algorithms can be covered by the video segmentation framework proposed in MPEG-4 standard [21], which is shown in Fig. 1. Camera Motion Compensation and Scene Cut Detection first compe...

Citation Context

... provides a content-based representation for the pixel-based or block-based source video, and it greatly benefits many multimedia applications including region of interest (ROI) coding, intelligent video surveillance, interactive video editing and manipulation, and specific object query and retrieval. There have been many approaches proposed for video object segmentation, both automatic and semi-automatic dependent on the target applications. Although researches on video object segmentation have progressed a lot in the recent years, automatic segmentation is still applicable to moving objects [1, 2] or some specific objects with a prior knowledge [3]. Semi-automatic segmentation is a more practical way to segment generic objects in a variety of video sequences [4, 5, 6, 7, 8, 9]. A tracking based paradigm is commonly adopted in both automatic and semi-automatic video segmentation approaches, in which the video objects in the first frame can be automatically or interactively segmented by the end user, and the video objects of subsequent frames are generated by tracking of previous video objects. Therefore, the initial video object may be obtained by different ways, but the following video...

Citation Context

... [2], see Figure 1, results in over segmented regions, and efficient merging algorithms are required to merge adjacent and similar regions, which can be time consuming. A region adjacency graph (RAG) =-=[3]-=- provides a simple-connectivity view of the image as Department of Electronics Systems Engineering University of Essex, Wivenhoe Park Colchester, UK {elanet, ekhan, woodjt, ghan@essex.ac.uk} illustrat...