Introduction from Session Chair

Abstract

There is a vast amount of existing digitised video data and each day new digitised content is created. But simply encoding video data in one of the existing formats (e.g., MPEG-X or H.263) is not sufficient to fully support applications that make use of video content. Video processing is a very useful mechanism to support application like content management, content creation, or even video surveillance. The latter is a good example to demonstrate the importance of video processing. A video surveillance application can create a large amount of video data. Analysing this large amount of data manually is very time consuming. Thus, applications that perform content recognition automatically are necessary to allow an efficient data analysis. Another example is an automated annotation of video data supported by video processing. These annotations can, e.g., be used for data retrieval. Unfortunately, existing mechanisms are still not mature enough to make use of them in commercial applications. In this session two new mechanisms for video processing are presented. The first paper presents an approach on the recognition of textural regions for colour video analysis. Results show that the presented mechanism has an accuracy from 90% to 97% and, thus, can be applied in multimedia applications as a colour texture recognition system. The second paper presents a new method for feature extraction of 3D object based on Wavelet Transform. An experiment shows that the feature extraction method generates feature vectors which uniquely describe 3D objects.