Recent progress and prospects in multimedia, humancomputer interaction, visual communications, semantic web, and cognitive vision call for and can benefit from applications of advanced image and video analysis technologies. Adaptive robust systems are required for analysis, indexing, and summarization of large amounts of audio-visual data. Advanced image analysis technologies are needed for next-generation description and browsing services characterized by structured, object-and content-based

more »

... presentations. Automatic extraction of semantic information from still or moving images and the analysis of their content are necessary for automatic annotation, indexing, and categorization. The aim of this special issue is to bring together contributions from the latest developments in the field of objectoriented and semantic image and video analysis applications. Ten papers have been selected following the reviewing process and appear in this issue, which are briefly described below. In the first paper, Cavallaro and Ebrahimi tackle semantic video object extraction by interacting between color change detection and region-based processing, achieving high spa-tial accuracy and temporal coherence. In the second paper, H.-Y. Wang and Ma propose a video object segmentation approach, involving image segmentation and motion estimation; the approach is based on spatial-constrained motion mask generation and motion-constrained spatial region merging. Video object segmentation is also the topic of the third paper by Porikli and Y. Wang. The authors perform a spatiotemporal decomposition of the data, defining simple homogeneous, in terms of low-level visual descriptors, components; the latter, called volumes, are then expanded and grouped into objects, using hierarchical clustering. In the next paper, Li et al. use a Markov random field model to obtain object-based semantic image segmentation, focusing on remote sensing applications; their approach includes a Wold model decomposition of the original image generating both stochastic and structural texture image components. The next two papers deal with technologies used in semantic image and video object analysis. In the first paper, Tsechpenakis et al. propose a model-based snake approach for object tracking, using a priori shape knowledge;