Search

Cable incipient fault (CIF) is a potential fault which may recur over time and evolve into permanent fault eventually. The identification of CIF can help to reduce the likelihood of a permanent fault and enhance power supply reliability. This paper fully considers the randomness and uncertainty of CIF waveforms, and proposes a CIF identification method using power disturbance waveform feature learning. Firstly, the shallow features are extracted to characterize the transient components of different disturbance current waveforms, by conducting a stationary wavelet transform. Then, a dropout deep belief network (DDBN) is constructed using the extracted shallow features by pre-training and fine-tuning. Finally, the well- constructed DDBN model is used to identify CIF from other similar disturbance events. The performance of the proposed method is verified by simulation data and experimental data, for different disturbance events, such as sub-cycle cable incipient fault and multi-cycle cable incipient fault, as well as other over-current disturbance events such as capacitor switching and inrush current. Both the accuracy and generalization ability are higher than other methods. The proposed method provides new insights on possible applications on monitoring of cable status and timely warning of cable faults.

This paper considers the problem of estimating the motion parameters of a low-observable high-speed maneuvering target with high-order motions, where the target’s complex motions will result in the range migration (RM) and Doppler frequency migration (DFM) effects within the observation time interval, which will significantly affect the radar performance. Based on the Hough transform (HT) and the polynomial Chirplet transform (PCT), a novel method for the motion parameter estimation, which we coin as HPCT, is proposed in this paper. It first employs the trajectory feature extracted from the image of range profiles by the HT to correct the first-order RM and resolve the Doppler ambiguity. Then, it compensates the residual range curvature and the DFM with the motion parameters estimated from the time-frequency ridge, which is extracted from the highly concentrated time-frequency representation (TFR) generated by the PCT. After that, the energy of radar returns can be coherently integrated along the target’s moving trajectory. In addition, we evaluate the parameter estimation performance, integration performance and computational cost of the proposed method via theoretical analysis or numerical experiments. The results show that in terms of the anti-noise ability, cross-term interference avoidance, measurement accuracy and computational complexity, the HPCT is superior to some common methods for the motion parameter estimation. Finally, we extend it to the scenario of multiple targets with different motion orders.

The incidence of skin cancer around the world is increasing year by year. However, early diagnosis and treatment can greatly improve the survival rate of patients. Skin lesion boundary segmentation is essential to accurately locate lesion areas in dermatoscopic images. It is true that accurate segmentation of skin lesions is still challenging dues to problems such as blurred borders, which requires an accurate and automatic skin lesion segmentation method. In this paper, we propose an end-to-end framework which can perform skin lesion segmentation automatically and efficiently, called the CSARM-CNN (Channel & Spatial Attention Residual Module) model. Each CSARM block of the model combines channel attention and spatial attention to form a new attention module to enhance segmentation results. The multi-scale input images are obtained by the spatial pyramid pooling. Finally, a weighted cross-entropy loss function is used at each side of the output layer to sum the total loss of the model. We evaluated in two published standard datasets, ISIC 2017 and PH2, and achieved competitive results in terms of specificity and accuracy, with 99.03% and 99.45% specificity, 94.96% and 95.23% accuracy, respectively.

Image registration is an important technique in many computer vision applications such as image fusion, image retrieval, object tracking, face recognition, change detection and so on. Local feature descriptors, i.e., how to detect features and how to describe them, play a fundamental and important role in image registration process, which directly influence the accuracy and robustness of image registration. This paper mainly focuses on the variety of local feature descriptors including some theoretical research, mathematical models, and methods or algorithms along with their applications in the context of image registration. The existing local feature descriptors are roughly classified into six categories to demonstrate and analyze comprehensively their own advantages. The current and future challenges of local feature descriptors are discussed. The major goal of the paper is to present a unique survey of the state-of-the-art image matching methods based on feature descriptor, from which future research may benefit.

In this paper, we study the sparse signal reconstruction with nonconvex regularization, mainly focusing on two popular nonconvex regularization methods, minimax concave penalty (MCP) and smoothly clipped absolute deviation (SCAD). An approximate message passing (AMP) algorithm is an effective method for signal reconstruction. Based on the AMP algorithm, we propose an improved MCP iterative thresholding algorithm and an improved SCAD iterative thresholding algorithm. Furthermore, we analyze the convergence of the new algorithms and provide a series of experiments to assess the performance of the new algorithms. The experiments show that the new algorithms based on AMP have stronger reconstruction capabilities, higher phase transition for sparse signal reconstruction, and better variable selection ability than the original MCP iterative thresholding algorithm and the original SCAD iterative thresholding algorithm.

Accurate segmentation of retinal vessels is a basic step in diabetic retinopathy (DR) detection. Most methods based on deep convolutional neural network (DCNN) have small receptive fields, and hence they are unable to capture global context information of larger regions, with difficult to identify pathological. The final segmented retina vessels contain more noise with low classification accuracy. Therefore, in this paper, we propose a DCNN structure named as D-Net. In the encoding phase, we reduced the loss of feature information by reducing the downsampling factor, which reduced the difficulty of tiny thin vessels segmentation. We use the combined dilated convolution to effectively enlarge the receptive field of the network and alleviate the “grid problem” that exists in the standard dilated convolution. In the proposed multi-scale information fusion module (MSIF), parallel convolution layers with different dilation rates are used, so that the model can obtain more dense feature information and better capture retinal vessel information of different sizes. In the decoding module, the skip layer connection is used to propagate context information to higher resolution layers, so as to prevent low-level information from passing the entire network structure. Finally, our method was verified on DRIVE, STARE, and CHASE dataset. The experimental results show that our network structure outperforms some state-of-art method, such as N4-fields, U-Net, and DRIU in terms of accuracy, sensitivity, specificity, and ${AUC_{ROC}}$ . Particularly, D-Net outperforms U-Net by 1.04 %, 1.23 %, and 2.79% in DRIVE, STARE, and CHASE dataset, respectively.

In this paper, a fully electronically reconfigurable reflectarray antenna that has 14×14 reflecting elements and features a higher aperture efficiency was designed, fabricated, and tested. A new configuration of the reflecting element was first developed so that all the bias circuit and the PIN diode are arranged on the opposite side of the reflecting surface, avoiding the potential unexpected reflection and aiming at the improvement of the antenna efficiency. The new reflecting element, which has a smaller size (0.365λ), was then applied to the design of a 14 × 14 reconfigurable reflectarray antenna. A field programmable gate array (FPGA)-based beam control system was also developed for the realization of an electronically beamscanning performance. The measurements of the prototype of the integrated reconfigurable reflectarray antenna show good beam-scanning radiation performance, a peak gain of 19.2 dBi, and an aperture efficiency of about 25%.

Exposure fusion (EF) directly generates a low dynamic range (LDR) image with similar image quality to the high dynamic range (HDR) image by combining images of different exposures to overcome the limited dynamic range of common digital cameras. EF for dynamic scenes is challenging not only due to the difficulty in removing various ghost artifacts but also because of the requirement of preserving details, especially in the saturated regions. Recently, low-rank (LR) matrix completion (LRMC) has been shown to be efficient in separating the latent background from the sparse motion in the irradiance domain. However, the performance of LRMC models strongly relies on the estimated irradiance images. To address this problem, this paper proposes a novel EF method combining Retinex theory and the LRMC model. The proposed method consists of four steps. First, according to Retinex theory, an image is decomposed into illumination and reflection components. Second, the LRMC model is applied to the reflection component to generate the background reflection component and the sparse error. Third, the motion map is modeled as a Markov random field (MRF), integrating the sparse error with the ordering constraint across all the illumination components. Finally, all the illumination components are fused via a pyramid-based method, where the weight maps are defined based on the obtained motion map and illumination. The experimental results show that the proposed method outperforms the state-of-the-art methods particularly in preserving details in saturated regions.

The cross-polarized scattering (HV) of polarimetric synthetic aperture radar (PolSAR) data is caused not only by forest but also by urban buildings with azimuth orientation angles. Since the general double scattering in model-based decomposition does not support their dominant scattering mechanism, it's still a challenge for modeling the scattering mechanism of oriented buildings. In this paper, the HV induced by oriented buildings is modeled by a rotated dihedral corner reflector. The cross scattering matrix of oriented building is obtained by averaging a cosine squared distribution with its peak at the dominant polarimetric orientation angle (DPOA) of an area and an adaptive width. The relation between DPOA and distribution width is acquired. Then an optimization strategy which eliminates the negative power and balances time and efficiency is proposed to estimate the scattering contributions. The proposed algorithm is tested on AIRSAR data of San Francisco and UAVSAR of San Diego and the results confirm the effectiveness.