Techniques from sparse signal representation are beginning to see significant impact in computer vision, often on non-traditional applications where the goal is not just to obtain a compact high-fidelity representation of the observed signal, but also to extract semantic information. The choice of dictionary plays a key role in bridging this gap: unconventional dictionaries consisting of, or learned from, the training samples themselves provide the key to obtaining state-of-theart results and to attaching semantic meaning to sparse signal representations. Understanding the good performance of such unconventional dictionaries in turn demands new algorithmic and analytical techniques. This review paper highlights a few representative examples of how the interaction between sparse signal representation and computer vision can enrich both fields, and raises a number of open questions for further study.

We present an open-source platform for wireless body sensor networks called DexterNet. The system supports real-time, persistent human monitoring in both indoor and outdoor environments. The platform utilizes a three-layer architecture to control heterogeneous body sensors. The first layer called the body sensor layer (BSL) deals with design of heterogeneous body sensors and their instrumentation on the body. At the second layer called the personal network layer (PNL), the body sensors on a single subject communicate with a mobile base station, which supports Linux OS and the IEEE 802.15.4 protocol. The BSL and PNL functions are abstracted and implemented as an open-source software library, called Signal Processing In Node Environment (SPINE). A DexterNet network is scalable, and can be reconfigured on-the-fly via SPINE. At the third layer called the global network layer (GNL), multiple PNLs communicate with a remote Internet server to permanently log the sensor data and support higher-level applications. We demonstrate the versatility of the DexterNet platform via several real-world applications. 1

Sensor network scenarios are considered where the underlying signals of interest exhibit a degree of sparsity, which means that in an appropriate basis, they can be expressed in terms of a small number of nonzero coefficients. Following the emerging theory of compressive sensing, an overall architecture is considered where the sensors acquire potentially noisy projections of the data, and the underlying sparsity is exploited to recover useful information about the signals of interest, which will be referred to as distributed sensor perception. First, we discuss the question of which projections of the data should be acquired, and how many of them. Then, we discuss how to take advantage of possible joint sparsity of the signals acquired by multiple sensors, and show how this can further improve the inference of the events from the sensor network. Two practical sensor applications are demonstrated, namely, distributed wearable action recognition using low-power motion sensors and distributed object recognition using high-power camera sensors. Experimental data support the utility of the compressive sensing framework in distributed sensor perception.

Copyright is held by the author/owner(s).

Abstract. Movement recognition constitutes a central task in homebased assisted living environments and in many application domains where activity recognition is crucial. Solutions in these application areas often rely on an heterogeneous collection of body-sensors whose diversity and lack of precision has to be compensated by advanced techniques for feature extraction and analysis. Although there are well established quantitative methods in machine learning for robotics and neighboring fields for addressing these problems, they lack advanced knowledge representation and reasoning capacities that may help understanding through contextualization. Such capabilities are not only useful in dealing with lacking and imprecise information, but moreover they allow for a better inclusion of semantic information and more general domain-related knowledge. We address this problem and investigate how a lexical approach to multisensor analysis can be combined with answer set programming to support movement recognition. A semantic notion of contextual coherence is formalized and qualitative optimization criteria are introduced in the reasoning process. We report upon a first experimental evaluation of the lexical approach to multi-sensor analysis and discuss the potentials of knowledge-based contextualization of movements in reducing the error rate. 1