VOE: A new sparsity-based camera network placement framework

In this paper, we propose a stepwise sparsity-based framework for camera network placement. Unlike most previous methods which are developed for specific tasks, our approach is universal and can generalize well for different application scenarios. There are three steps in our approach: visibility analysis, optimization and evaluation (VOE), which are employed sequentially and iteratively. First, we use a cascaded visibility filter model to construct a visibility matrix, where each column describes the appearance representation of the surveillance area. Then, we formulate camera network layout as a sparse representation problem, and employ an l1-optimization algorithm to obtain a feasible solution. Our framework is general enough and applicable to various objectives in practical applications. Experiment results are presented to show the effectiveness and efficiency of the proposed framework.