Saliency-directed prioritization of visual data in wireless surveillance networks

Abstract

Item Type Article Authors Mehmood, Irfan; Sajjad, M.; Ejaz, W.; Baik, S.W. Citation Mehmood I, Sajjad M, Ejaz W et al (2015) Saliency-directed prioritization of visual data in wireless surveillance networks. Information Fusion. 24: 16-30. RightsAbstract: In wireless visual sensor networks (WVSNs), streaming all imaging data is impractical due to resource constraints. Moreover, the sheer volume of surveillance videos causes intermittency on the analysts' ability to extract actionable intelligence from it. In this work, an energy-efficient image prioritization framework is presented to cope with the fragility of the traditional WVSN. The proposed framework selects semantically relevant information before transmission to the sink node. This is based on salient motion detection, which works on the principle of human cognitive processes. Each camera node estimates background bootstrapping, which helps to increase the efficiency of salient motion detection. Based on the salient motion, each sensor node is classified into a high or a low-priority node. This classification is dynamic such that camera nodes toggle between high-priority and low-priority status depending on the coverage of region of interest. The high-priority camera nodes are allowed to access the reliable radio channels for the sake of timely and reliable transmission of data. We compared the performance of this framework with other state-of-the-art methods in the cases of both single and multi-camera monitoring. The results indicate the usefulness of the proposed method in terms of salient event coverage and the reduction of computational and transmission costs as well as in helping analysts find semantically relevant visual information.Keywords: image prioritization, wireless visual sensor networks, monitoring applications, salient activity detection. Corresponding Author networks. In contrast to scalar sensors, WVSNs capture visual data, which offers rich information. Moreover, owing to their high versatility, small size, and dense spatial coverage, WVSNs can be deployed flexibly in various applications, such as in remote patient monitoring, distributed multimedia-based surveillance, and security systems [1][2][3][4]. In such setups, visual data about the event area has tremendous potential to influence decision making because the WVSNs address real-time observation of the events occurring within a busy environment. Despite the existence of numerous ideas of attractive surveillance applications, their actual implementation using the WVSNs remains a challenge. This is a challenge particularly because surveillance systems consist of numerous sensors that consume a large amount of bandwidth to transmit their raw video streams. Further, these video streams require extensive processing to detect events and anomalies. Recently, there have been attempts to advance the current WVSN technologies by implementation of intelligent methods for automatically extracting the relevant data at the source node to reduce the consumption of network bandwidth.WVSN based mo...

Audio-Visual and EEG-Based Attention Modeling for Extraction of Affective Video Content

“…Fig. 2 illustrates that the proposed method achieves high values for Informativeness, Enjoyability and Rank as compared to other techniques [1,2]. This indicates that the generated summaries are more informative, precise, and are closer to human perception.…”

“…It facilitates users to quickly browse semantically relevant videos. In this context, researchers have presented various summarization schemes [1,2]. Most of the state-of-the-art summarization schemes employ the concept of the visual attention model to extract the summaries closer to human perception.…”

Example of IoT platform usage for wireless video surveillance with support of NoSQL and cloud systems

“…Related to wireless video surveillance systems, there are various and interesting examples of application of these systems in different areas such as precision agriculture [14], using 6LoWPAN [15] and WiMAX [17] technology, or in using efficient prioritization techniques [16].…”

Power efficient dataflow design for a heterogeneous smart camera architecture

“…Visual Saliency (VS) or attention modelling characterises the scene to segment regions of visual interest. Hence it is a suitable concept for assessing the relevance of a region in a frame for further processing in a vision system, e.g., saliency based person re-identification [6], resource allocation for driver assistance [7] or prioritization of visual data in wireless surveillance networks [8]. This significantly reduces data dimensionality resulting in accelerated processing, reduced power consumption and bandwidth demand.…”