We propose a new framework for estimating depth information from a single image. Our framework is relatively small and straightforward by employing a two-stage architecture: a residual network and a simple decoder network. Our residual network in this paper is a remodeled of the original ResNet-50 architecture, which consists of only thirty-eight convolution layers in the residual block following by pair of two up-sampling and layers. While the simple decoder network, stack of five convolution layers, accepts the initial depth to be refined as the final output depth. During training, we monitor the loss behavior and adjust the learning rate hyperparameter in order to improve the performance. Furthermore, instead of using a single common pixel-wise loss, we also compute loss based on gradient-direction, and their structure similarity. This setting in our network can significantly reduce the number of network parameters, and simultaneously get a more accurate image depth map. The performance of our approach has been evaluated by conducting both quantitative and qualitative comparisons with several prior related methods on the publicly NYU and KITTI datasets.

We combine the siamese network and the recurrent regression network, proposing a two-stage tracking framework termed as SiamReg. Our method solves the problem that the classic siamese network can not judge the target size precisely and simplifies the procedures of regression in the training and testing process. We perform experiments on three challenging tracking datasets: VOT2016, OTB100, and VOT2018. The results indicate that, after offline trained, SiamReg can obtain a higher expected average overlap measure.

Crack detection is a vital task to maintain a bridge's health and safety condition. Traditional computer-vision based methods easily suffer from disturbance of noise and clutters for a real bridge inspection. To address this limitation, we propose a two-stage crack detection approach based on Convolutional Neural Networks (CNN) in this letter. A predictor of small receptive field is exploited in the first detection stage, while another predictor of large receptive field is used to refine the detection results in the second stage. Benefiting from data fusion of confidence maps produced by both predictors, our method can predict the probability belongs to cracked areas of each pixel accurately. Experimental results show that the proposed method is superior to an up-to-date method on real concrete surface images.

This paper presents accurate DC and high frequency power-supply rejection (PSR) models for low drop-out (LDO) regulators using different types of active loads and pass transistors. Based on the proposed PSR model, we suggest design guidelines to achieve a high DC PSR or flat bandwidth (BW) by choosing appropriate active loads and pass transistors. Our PSR model captures the intricate interaction between the error amplifiers (EAs) and the pass devices by redefining the transfer function of the LDO topologies. The accuracy of our model has been verified through SPICE simulation and measurements. Moreover, the measurement results of the LDOs fabricated using the 0.18 µm CMOS process are consistent with the design guidelines suggested in this work.

In this paper, a discrete particle swarm optimization method is proposed to solve the multi-objective task assignment problem in distributed environment. The objectives of optimization include the makespan for task execution and the budget caused by resource occupation. A two-stage approach is designed as follows. In the first stage, several artificial particles are added into the initialized swarm to guide the search direction. In the second stage, we redefine the operators of the discrete PSO to implement addition, subtraction and multiplication. Besides, a fuzzy-cost-based elite selection is used to improve the computational efficiency. Evaluation shows that the proposed algorithm achieves Pareto improvement in comparison to the state-of-the-art algorithms.

Phishing attacks target financial returns by luring Internet users to exposure their sensitive information. Phishing originates from e-mail fraud, and recently it is also spread by social networks and short message service (SMS), which makes phishing become more widespread. Phishing attacks have drawn great attention due to their high volume and causing heavy losses, and many methods have been developed to fight against them. However, most of researches suffered low detection accuracy or high false positive (FP) rate, and phishing attacks are facing the Internet users continuously. In this paper, we are concerned about feature engineering for improving the classification performance on phishing web pages detection. We propose a novel anti-phishing framework that employs feature engineering including feature selection and feature extraction. First, we perform feature selection based on genetic algorithm (GA) to divide features into critical features and non-critical features. Then, the non-critical features are projected to a new feature by implementing feature extraction based on a two-stage projection pursuit (PP) algorithm. Finally, we take the critical features and the new feature as input data to construct the detection model. Our anti-phishing framework does not simply eliminate the non-critical features, but considers utilizing their projection in the process of classification, which is different from literatures. Experimental results show that the proposed framework is effective in detecting phishing web pages.

In this paper, resource-efficient multiple description coding (MDC) multicast is investigated in cognitive radio networks with the consideration of imperfect spectrum sensing and imperfect channel feedback. Our objective is to maximize the system goodput, which is defined as the total successfully received data rate of all multicast users, while guaranteeing the maximum transmit power budget and the maximum average received interference constraint. Owing to the uncertainty of the spectrum state and the non-closed-form expression of the objective function, it is difficult to solve the problem directly. To circumvent this problem, a pretreatment is performed, in which we first estimate the real spectrum state of primary users and then propose a Gaussian approximation for the probability density functions of transmission channel gains to simplify the computation of the objective function. Thereafter, a two-stage resource allocation algorithm is presented to accomplish the subcarrier assignment, the optimal transmit channel gain to interference plus noise ratio (T-CINR) setting, and the transmit power allocation separately. Simulation results show that the proposed scheme is able to offset more than 80% of the performance loss caused by imperfect channel feedback when the feedback error is not high, while keeping the average interference on primary users below the prescribed threshold.

To achieve high quality output speech synthesis systems, data-driven grapheme-to-phoneme (G2P) conversion is usually used to generate the phonetic transcription of out-of-vocabulary (OOV) words. To improve the performance of G2P conversion, this paper deals with the problem of conflicting phonemes, where an input grapheme can, in the same context, produce many possible output phonemes at the same time. To this end, we propose a two-stage neural network-based approach that converts the input text to phoneme sequences in the first stage and then predicts each output phoneme in the second stage using the phonemic information obtained. The first-stage neural network is fundamentally implemented as a many-to-many mapping model for automatic conversion of word to phoneme sequences, while the second stage uses a combination of the obtained phoneme sequences to predict the output phoneme corresponding to each input grapheme in a given word. We evaluate the performance of this approach using the American English words-based pronunciation dictionary known as the auto-aligned CMUDict corpus[1]. In terms of phoneme and word accuracy of the OOV words, on comparison with several proposed baseline approaches, the evaluation results show that our proposed approach improves on the previous one-stage neural network-based approach for G2P conversion. The results of comparison with another existing approach indicate that it provides higher phoneme accuracy but lower word accuracy on a general dataset, and slightly higher phoneme and word accuracy on a selection of words consisting of more than one phoneme conflicts.

In this paper, an efficient method to reduce computational complexity for pedestrian detection is presented. Since trilinear interpolation is not used, the amount of required operations for histogram of oriented gradient (HOG) feature calculation is significantly reduced. By calculating multi-scale HOG features with integral HOG in a two-stage approach, both high detection rate and speed are achieved in the proposed method.

A sensing efficiency optimization scheme based on two-stage spectrum sensing that maximizes the achievable throughput of the secondary network and minimizes the average sensing time is proposed in this paper. A selection method for the threshold is proposed and proved to ensure optimal sensing performance. An effective iterative algorithm is presented to solve the constructed efficiency optimization problem.

In this letter, a two-stage TOA estimation scheme is proposed for positioning in OFDM-based WLAN systems under indoor environments. The estimation scheme consists of coarse estimation and fine estimation. The presented scheme effectively exploits the preamble of the OFDM-based WLAN for accurate estimation. The simulation results exhibit that the performance of the proposed approach is comparable to that of super-resolution estimation even with lower computational complexity.

A spectrum sensing scheme for cognitive radio that includes coarse and fine sensing stages based on cyclostationarity is proposed in this paper. The cyclostationary feature detection (CFD) based on a single cyclic frequency (SCF) is used in the coarse sensing stage and that based on multiple cyclic frequencies (MCF) is employed in the fine sensing stage. Whether the fine sensing stage is performed or not is decided by comparing the statistic constructed in the coarse sensing stage with two thresholds. Theoretical analyses and simulation results show that the proposed sensing scheme has superior sensing performance and needs shorter sensing time.

Due to the advancement of software radio and RF technology, cognitive radio(CR) has become an enabling technology to realize dynamic spectrum access through its spectrum sensing and reconfiguration capability. Robust and reliable spectrum sensing is a key factor to discover spectrum opportunity. Single cognitive radios often fail to provide such reliable information because of their inherent sensitivity limitation. Primary signals that are subject to detection by cognitive radios may become weak due to several factors such as fading and shadowing. One approach to overcome this problem is to perform spectrum sensing by using multiple CRs or multiple spectrum sensors. This approach is known as distributed sensing because sensing is carried out through cooperation of spatially distributed sensors. In distributed sensing, sensors should perform spectrum sensing and forward the result to a destination where data fusion is carried out. Depending on the channel conditions between sensors (sensor-to-sensor channel) and between the sensor and the radio (user-channel), we explore different spectrum sensing algorithms where sensors provide the sensing information either cooperatively or independently. Moreover we investigate sensing schemes based on soft information combining (SC), hard information combining (HC). Finally we propose a two-stage detection scheme that uses both SC and HC. The newly proposed detection scheme is shown to provide improved performance compared to sensing based on either HC or SC alone. Computer simulation results are provided to illustrate the performances of the different sensing algorithms.

This paper proposes a low power and high speed track and hold circuit (T/H circuit) based on the two-stage structure. The proposed circuit consists of two internal T/H circuits connected in cascade. The first T/H circuit converts an input signal into a step voltage and it is applied to the following second T/H circuit which drives large load capacitors and consumes large power. Applying the step voltage to the second T/H circuit prevents the second T/H circuit from charging and discharging its load capacitor during an identical track phase and enables low power operation. Thanks to the two-stage structure the proposed T/H circuit can save 29% of the power consumption compared with the conventional one. An optimum design procedure of the proposed two stage T/H circuit is explained and its validity is confirmed by HSPICE simulations.

A stability of the cascade two-stage Power-Factor-Correction converter is investigated. The first stage is boost PFC converter to achieve a near unity power factor and the second stage is forward converter to regulate the output voltage. Previous researches studied the system using linear analysis. However, PFC boost converter is a nonlinear circuit due to the existence of the multiplier and the large variation of the duty cycle. Moreover, the effect of the second stage DC/DC converter on the first stage PFC converter adds more complexity to the nonlinear circuit. In this issue, low-frequency instability has been detected in the two-stage PFC converter assuring the limitation of the prior linear models. Therefore, nonlinear model is proposed to detected and explain these instabilities. The borderlines between stable and unstable operation has been made clear. It is cleared that feedback gains of the first stage PFC and the second stage DC/DC converters are the main affected parts to the total system stability. Then, a simplified nonlinear model is provided. Experiment confirm the two models with a good agreement. These nonlinear models have introduced new PFC design scheme by choosing the minimum required output capacitor and the feedback loop design.

This paper presents the improvement of the transient response and stability for a two-stage DC-DC converter by removing the output inductor. The conventional two-stage converter consists of a buck converter used as the first stage and a half-bridge converter used as the second stage. The proposed circuit topology removing the output inductor and the conventional topology are compared. Removing the output inductor results in the system-order reduction of the transfer function. As a result, the stability is improved, and the crossover frequency of the open-loop transfer function becomes higher. The effectiveness of the proposed circuit topology was experimentally confirmed.

In this paper, we proposed a novel two-stage channel estimation (2S-CE) method. In contrast to conventional channel estimation methods, this method makes the maximum use of the information contributed by the known data in every transmission burst. In the first stage, the least-squares (LS) algorithm was used to estimate the channel impulse response (CIR) based on the normal training sequence. Then the maximum channel memory was estimated and used to locate the uncorrupted data in the guard interval. In the second stage, the uncorrupted data together with the normal training sequence were sent to the LS algorithm again to obtain the fine-tuned CIR. To verify the efficiency of the proposed 2S-CE method, both a theoretical analysis and computer simulations have been done. Computer simulation results confirm the analysis results and demonstrate that the proposed 2S-CE method outperforms a conventional single-stage channel estimation method.

In a content-based image retrieval (CBIR) system, both the retrieval relevance and the response time are very important. This letter presents the condensed two-stage search method as a new fast image retrieval approach by making use of the property of Cauchy-Schwarz inequality. The method successfully reduces the overall processing time for similarity computation, while maintaining the same retrieval relevance as the conventional exhaustive search method. By the extensive computer simulations, we observe that the condensed two-stage search method is more effective as the number of images and dimensions of the feature space increase.

MOSFETs can be used as capacitors, but its capacitance can vary by 5 to 7 times as its terminal voltage varies. To reduce the voltage dependence of the capacitance, this paper proposed two types of devices: one is called accumulation MOSFET (AMOS) and the other is formed by two conventional PMOS connected in anti-parallel. These two devices are readily available in the standard digital CMOS processes. The proposed capacitors were implemented in three different CMOS processes. The measured results show that the capacitances of both devices have less voltage dependence than a single PMOS. The voltage dependence of the AMOS capacitance can be as small as 17%. The minimum capacitance per unit area of the AMOS is 1.8 times that of the double-poly capacitor in an analog/mixed-mode CMOS process. To verify the usefulness of these two types of capacitors, they are used as compensation capacitors in a conventional two-stage amplifier. The measured results show that the amplifier compensated by the AMOS capacitor has little variation (6%) of the unity-gain frequency over the input common-mode range. Due to its smaller die area and cheaper digital process, AMOS can be used as compensation capacitor without resorting to more expensive analog process.

In this paper we propose a two-stage address coding scheme to transmit two data symbols at once within a frame in a MFSK/FH-CDMA system. We compare it with the conventional system using single-stage address coding. Assumed that the address codes of all users are known in the receiver. A multiuser detection scheme is applied and the performance is evaluated by computer simulations to show the improvement in bit error rate (BER) compairing to the conventional system. We also investigate the performance of error-correcting coding and decoding in the two-stage address coded MFSK/FH-CDMA system. An erasure decoding scheme is modified for the two-stage address coded system and is utilized to improve spectral efficiency or to increase user capacity in the MFSK/FH-CDMA system. Finally, we investigate a hybrid scheme of combining the multi-user detection scheme and the error-correcting decoding scheme for the two-stage address coded MFSK/FH-CDMA system. The performance is evaluated by computer simulations.