Wireless visual sensor Network (WVSN) has become a new important research area where camera-equipped sensor nodes can capture, process, and transmit visual (image/video) information. While the traditional wireless sensor networks (WSN) can only transmit the scalar information (e.g., temperature), WVSN enables a much wider range of applications, such as visual security surveillance and visual wildlife monitoring. However, visual data is much bigger in size and more complicated to process than scalar data. In addition, visual sensor coverage in WVSN is significantly different from traditional sensor coverage in WSN, because a camera sensor might fail to detect objects within the sensing range due to camera's viewing angle or occlusion. This makes the capturing, processing, and transmission of visual information in WVSN even more challenging than in WSN. Meeting these challenges in WVSN requires interdisciplinary and cross-layered approaches, including visual quality aware image/video processing, signal quality aware communications, and power aware networking. This raises new interesting and challenging research issues and opportunities in WVSN.
This special issue includes eight papers with various aspects in WVSN research. The paper “Adaptive monitoring relevance in camera networks for critical surveillance applications” by D. G. Costa et al. proposes a two-tier hierarchy where the scalar sensors in the first tier are to detect the event and camera sensor in the second tier is assigned based on the relevance measures to capture the visual data of the event. Based on the three monitoring relevance measures, the visual data coding and routing strategy could be chosen so that the critical event has better visual resolution and short delay time.
The paper “An enhanced approach for reliable bulk data transmission based on erasure-resilient codes in wireless sensor networks” by T. Park et al. proposes an error control technique, called M-FEC, to provide reliable bulk data transmission in WVSN. M-FEC is an enhanced version of forward error correction (FEC) to fragment the bulk data into multiple small blocks and transmit the blocks. If there is an error in the frame, only the block that contains the error will be dropped. This frame will be buffered until the frame is in full size and then transmits again. With this idea of reassembling the error-free blocks into a full-sized frame at the relay node before the relay node transmits the frame, the number of transmissions could be reduced.
The paper “Approximate techniques in solving optimal camera placement problems” by J. Zhao et al. discusses strengths and weaknesses of various approaches for finding optimal placement of visual sensors. In-depth analysis of accuracy, efficiency, and scalability of each approach is also covered in this paper.
The paper “Distributed Bayesian inference for consistent labeling of tracked objects in nonoverlapping camera networks” by J. Wan and L. Lill presents a VSN application for targets tracking. The proposed distributed Bayesian inference framework does not require prior knowledge of the number of objects presenting in the monitored region, and no assumptions are made regarding the appearance distribution of a single object. The proposed technique is evaluated using real dataset covering both indoor and outdoor scenarios.
When wireless VSN is applied to monitor flying objects in the sky, challenges arise for efficient and effective data acquisition with flying objects observed at different altitudes. The paper “Modeling and verification of a heterogeneous sky surveillance visual sensor network” by N. Ahmad et al. presents the design of a heterogeneous VSN system for monitoring a given area with multiple altitude limits. In comparison to a homogeneous VSN setup, full volume coverage can be achieved with a significant cost reduction by adopting the proposed system.
The paper “Efficient image transmission over WVSNs using two-measurement matrix based CS with enhanced OMP” by H. Rajendra et al. proposes a compressive sampling (CS) based image transmission system as well as an improved version of the orthogonal matching pursuit (OMP) algorithm to reduce the amount of dataflow across the network, for the purpose of increasing the life time of a WVSN. By utilizing two-measurement matrix (TMM) based sampling strategy with CS only and nonuniform sampling (NUS) based CS techniques with DCT and binDCT as sparsity basis, the required number of measurements is reduced considerably. With these reduced measurements, the image quality is maintained in the acceptable level, which in turn enables the low bit rate transmission of the images in WVSNs.
The paper “Complexity analysis of vision functions for comparison of wireless smart cameras” by M. Imran et al. presents an abstract model for the comparison and generalization of vision solutions for wireless smart cameras, with the assistance of system taxonomy, to predict the arithmetic complexity and memory requirements. The use of the proposed model is demonstrated by analyzing a number of published systems as a case study and classifying them with the system taxonomy. The proposed approach will assist in proposing efficient generic solutions for the same class of problems with reduced design and development costs.
The paper “Hardware architecture for real-time computation of image component feature descriptors on a FPGA” by A. W. Malik et al. presents a hardware architecture for real-time image component labeling and the computation of image component feature descriptors. With component labeling and feature calculation running in parallel, the presented hardware architecture is suitable for the efficient implementation of machine vision systems and can support robust, high speed, low latency, and low power smart camera applications.
Hong-Hsu Yen
Hongkai Xiong
Ivan Lee
