Abstract
This paper proposes a dynamical task decomposition approach for tracking a large number of people in a wide area with multiple cameras. By exploiting geometric relations between sensing geometry and people's positions, the method is able to dynamically decompose the overall tracking task (i.e. tracking all people using all available cameras) into a number of nearly independent subtasks. Each subtask tracks a subset of people with a subset of cameras. The method hereby reduces task complexity dramatically and helps to boost parallelization, balance computational load and maximize the system's real time throughput and resource utility. The optimal task decomposition is computed by minimum cost flow. We demonstrate the efficiency of our approach by conducting experiments with a challenging sequence. Copyright 2014 ACM.