Weakly supervised target detection in remote sensing images based on transferred deep features and negative bootstrapping

Target detection in remote sensing images (RSIs) is a fundamental yet challenging problem faced for remote sensing images analysis. More recently, weakly supervised learning, in which training sets require only binary labels indicating whether an image contains the object or not, has attracted considerable attention owing to its obvious advantages such as alleviating the tedious and time consuming work of human annotation. Inspired by its impressive success in computer vision field, in this paper, we propose a novel and effective framework for weakly supervised target detection in RSIs based on transferred deep features and negative bootstrapping. On one hand, to effectively mine information from RSIs and improve the performance of target detection, we develop a transferred deep model to extract high-level features from RSIs, which can be achieved by pre-training a convolutional neural network model on a large-scale annotated dataset (e.g. ImageNet) and then transferring it to our task by domain-specifically fine-tuning it on RSI datasets. On the other hand, we integrate negative bootstrapping scheme into detector training process to make the detector converge more stably and faster by exploiting the most discriminative training samples. Comprehensive evaluations on three RSI datasets and comparisons with state-of-the-art weakly supervised target detection approaches demonstrate the effectiveness and superiority of the proposed method.     

Multi-view and multi-plane data fusion for effective pedestrian detection in intelligent visual surveillance

For the robust detection of pedestrians in intelligent video surveillance, an approach to multi-view and multi-plane data fusion is proposed. Through the estimated homography, foreground regions are projected from multiple camera views to a reference view. To identify false-positive detections caused by foreground intersections of non-corresponding objects, the homographic transformations for a set of parallel planes, which are from the head plane to the ground, are applied. Multiple features including occupancy information and colour cues are extracted from such planes for joint decision-making. Experimental results on real world sequences have demonstrated the good performance of the proposed approach in pedestrian detection for intelligent visual surveillance.     

Survivable deployment of cloud-integrated fiber-wireless networks against multi-fiber failure

Cloud-integrated fiber-wireless (FiWi) networks inheriting advantages of optical and wireless access networks have a broad prospect in the future. As various component failures may occur in cloud-integrated FiWi networks, survivability is becoming one of the key important issues. It is necessary to provide survivability strategies for cloud-integrated FiWi networks. Hence, this paper mainly focuses on the survivability of cloud-integrated FiWi networks against multiple fibers failure. Firstly, in this paper, a novel integer linear programming (ILP) solution is proposed to tolerate the failure of multiple distribution fibers with capacity and coverage constraints in the context of urban area. Then, considering the complexity of ILP models, an efficient heuristic scheme is proposed, in order to get the approximate solutions of ILP. Simulation results and analysis give the configurations of optical network units (ONUs) and wireless routers with different constraints and show the network coverage of clients for different number of ONUs and wireless routers with ILP solution and heuristic approach, respectively.     

Wavelength channel minimization-based energy-aware mechanism in a multichannel EPON

We propose an energy-aware mechanism (EAM) applicable to the multichannel Ethernet Passive Optical Network that can minimize the number of wavelength channels used and save energy. Wavelength channel minimization is processed by collecting the information such as request message, allocated grant, and start time of each optical network unit (ONU) transmission in the previously elapsed scheduling cycles and comparing it with the buffer occupancy and packet delay conditions of the ONUs required by the user’s quality-of-service requirement. They are exchanged between the optical line terminal (OLT) and the ONUs via the multipoint control protocol. In this way, at the beginning of each scheduling cycle, the ONU’s buffer occupancy and packet delay conditions can be evaluated, and then, the OLT decides the smallest number of wavelength channels to be used in the current scheduling cycle. By turning off the OLT receivers corresponding to the unused wavelength channels, the OLT can save energy. The performance of the proposed EAM was evaluated through simulations using nonjoint off-line dynamic bandwidth allocation and dynamic wavelength assignment algorithms. The results showed that the OLT receivers’ power consumption could be reduced by 48 % on average.     

Performance analysis of MIMO–OFDM free space optical communication system with low-density parity-check code

In this paper, we have evaluated the performance of a low-density parity-check (LDPC)-coded multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) free space optical (FSO) communication system. Closed form expressions for the average bit error rate and throughput with diversity using equal gain combining have been obtained for the system under consideration. The Monte Carlo simulation has been carried out for the verification of the results. The performance of the QPSK and 16 QAM modulations is evaluated for different weather and atmospheric turbulence conditions. The results are also compared for both, QPSK and 16 QAM for SISO–OFDM, \(2\times 2\) and \(4 \times 4\) MIMO–OFDM FSO communication system. The results show that the performance of the system under consideration improves, as we move from SISO–OFDM to \(4 \times 4\) MIMO–OFDM. The results also show that the effect of weather is very much pronounced on the system and the performance in terms of average bit error rate of QPSK is better than 16 QAM in the presence of every weather condition. However, the later provides better throughput. Regular LDPC codes with code rate 1/2 have been applied to the simulated results, yielding high coding gains. Coding gain of 29.5 and 22 dB is achieved for QPSK and 16 QAM, respectively, for \(4 \times 4\) MIMO–OFDM.     

Throughput analysis of a CSMA based WLAN with successive interference cancellation under Rayleigh fading and shadowing

In this paper, we consider a carrier sense multiple access based wireless local area network (WLAN) with a successive interference cancellation (SIC) technique. We develop an analytical model to compute the average throughput of a user in a WLAN with the SIC technique in presence of path loss, Rayleigh fading and log-normal shadowing. We then validate the model via simulation. By means of the developed analytical model, we compute the average throughput of a user in WLAN systems without and with the SIC technique and evaluate the throughput gain provided by the SIC technique. We find that the throughput gain provided by the SIC technique is significant. However, the throughput gain varies significantly depending on the parameters of network and wireless channel. We find that the throughput gain provided by the SIC technique increases with increasing the number of users in WLAN, medium access rate of the users and the variance in shadowing and it decreases with increasing the data transmission rate. We also investigate the effect of the decoding capability of the SIC technique on the throughput performance. We find that throughputs obtained with decoding capability of 2 and 3 packets are very close.     

FIR system identification based on a nonparametric Bayesian model using the Indian buffet process

In this paper, we propose a nonparametric Bayesian model combined with the Indian buffet process (IBP) for a finite impulse response (FIR) system. We develop an FIR system identification method that can simultaneously estimate the number of FIR taps and coefficients. In the proposed model, each FIR tap consists of a coefficient and a gain, and the gain is a binary value. An infinite-dimensional binary vector is composed of binary values, and we assume that this binary vector is generated by the IBP. To identify the FIR system, we specify the likelihood function and prior distributions of the parameters and derive their posterior distributions. We can simultaneously estimate the number of FIR taps and coefficients by sampling from posterior distributions using the Gibbs sampler. Our simulations demonstrate that although the number of FIR taps is unknown, the identification performance of the proposed method in a high signal-to-noise ratio environment is similar to or better than that of the conventional least square solution.     

A dynamic search pattern motion estimation algorithm using prioritized motion vectors

Motion estimation is one of the critical parts in video compression standards with a high computational load. Many motion estimation algorithms have been developed to reduce the number of search points compared to a full-search algorithm without losing the quality considerably. Most of them use fixed search patterns in their first step which may suffer from trapping into local minima or searching unnecessary blocks due to inappropriate size and type of search patterns. In this paper, a new dynamic search pattern using motion vectors of spatial and temporal neighboring blocks is proposed. The motion vectors of neighboring blocks are prioritized, in order to efficiently use of halfway stop technique. The simulation results indicate that proposed algorithm is very close to the full-search algorithm in quality, compared to other rivals. Moreover, the average number of searches is often less than other algorithms.     

Salient object detection via images frequency domain analyzing

Salient object detection has become an important direction in image processing and computer vision. The traditional center-priori theory believes that salient target should be closer to the central area of the image. However, false detection will often occur when the salient object is closer to the image boundary. So, this paper obtains center coordinates of the salient object by using Harris corner detection algorithm and convex hull. Accordingly, an improved center-priori saliency detection model is obtained by applying the frequency-tuned method. And then, the local saliency is set up by wavelet transforming which has the local characteristic information representation ability in the time domain and frequency domain. In addition, we obtain the global saliency by spectral residual analyzing. Finally, an advanced center-priori saliency model is established. The experimental results show that the model in this paper has better detection effects and higher target detection rates.     

Optimization of the Security-Performance Tradeoff in RC4 Encryption Algorithm

Wireless Personal Communications - In this paper, we have investigated different vulnerabilities in RC4 and its enhanced variants to overcome the security attacks. It is established that in spite...