This paper is concerned with the problem of distributed H ∞ -consensus filtering for target tracking over a wireless filter network. The wireless filter network consists of a large number of filter nodes. The network topology is supposed to be switching or changeable over time. Data communication between filter nodes is subject to some network-induced constraints including packet dropouts and channel fading. Different from some existing results, packet dropouts and channel fading are assumed to occur not only in the measurement process for sensors but also in the transmission process among filter nodes. Within such settings, a distributed H ∞ -consensus filtering method is developed to ensure both the estimation accuracy of filters and the robustness of the filtering error system against network-induced constraints. Criteria on designing desired distributed H ∞ filters are derived in terms of a set of linear matrix inequalities. Numerical examples are given to verify the effectiveness of the proposed target tracking filtering algorithm.

No abstract

A fast and effective fade detection algorithm is proposed in this paper, which directly operates in compressed domain and suitable for real-time implementation. By analyzing the prediction directions of B frames, which are revealed in the macroblock types, the candidate fades can be found. Then, uncommon intracoded macroblocks of the P frame can be applied as an indicator of fade. As a result, locating fades are operated by a sliding window. Extensive experiments illustrate that the proposed algorithm achieves superiority measured by recall and precision rates, providing a useful technique for fast and on-line video content processing.

No abstract

No abstract