Search

It is still a challenging task to perform the semantic segmentation with high accuracy due to the complexity of real picture scenes. Many semantic segmentation methods based on traditional deep learning insufficiently captured the semantic and appearance information of images, which put limit on their generality and robustness for various application scenes. Thus, in this paper, we proposed a novel strategy that reformulated the popularly used convolution operation to multi-layer convolutional sparse coding block in semantic segmentation method to ease the aforementioned deficiency. To prove the effectiveness of our idea, we chose the widely used U-Net model for the demonstration purpose, and we designed CSC-Unet model series based on U-Net. Through extensive analysis and experiments, we provided credible evidence showing that the multi-layer convolutional sparse coding block enables semantic segmentation model to converge faster, extract finer semantic and appearance information of images, and improve the ability to recover spatial detail information. The best CSC-Unet model significantly outperforms the results of the original U-Net on three public datasets with different scenarios, i.e., 87.14% vs. 84.71% on DeepCrack dataset, 68.91% vs. 67.09% on Nuclei dataset, and 53.68% vs. 48.82% on CamVid dataset, respectively. In addition, the proposed strategy could be possibly used to significantly improve segmentation performance of any semantic segmentation model that involves convolution operations and the corresponding code is available at https://github.com/NZWANG/CSC-Unet.

For semantic segmentation of remote sensing images (RSI), trade-off between representation power and location accuracy is quite important. How to get the trade-off effectively is an open question, where current approaches of utilizing very deep models result in complex models with large memory consumption. In contrast to previous work that utilizes dilated convolutions or deep models, we propose a novel two-stream deep neural network for semantic segmentation of RSI (RSI-Net) to obtain improved performance through modeling and propagating spatial contextual structure effectively and a decoding scheme with image-level and graph-level combination. The first component explicitly models correlations between adjacent land covers and conduct flexible convolution on arbitrarily irregular image regions by using graph convolutional network, while densely connected atrous convolution network (DenseAtrousCNet) with multi-scale atrous convolution can expand the receptive fields and obtain image global information. Extensive experiments are implemented on the Vaihingen, Potsdam and Gaofen RSI datasets, where the comparison results demonstrate the superior performance of RSI-Net in terms of overall accuracy (91.83%, 93.31% and 93.67% on three datasets, respectively), F1 score (90.3%, 91.49% and 89.35% on three datasets, respectively) and kappa coefficient (89.46%, 90.46% and 90.37% on three datasets, respectively) when compared with six state-of-the-art RSI semantic segmentation methods.

Faults, such as broken rotor bars, in induction motors may be detected by estimating the spectral signature of the stator currents, particularly the sidebands around the supply line frequency. However, the amplitude of the fundamental frequency (50 Hz) is considerably greater than the sideband amplitude. How to demodulate the signature frequency components under the heavy background of fundamental frequency, or how to remove the fundamental frequency, is becoming a key problem in motor current signature analysis. This paper puts forward a novel transform demodulation algorithm to solve the problem. The three-phase currents are transformed to a magnetic-torque (M-T) coordinate using this algorithm. It is found that the signature frequency components are demodulated in the magnetizing and torque-producing currents obtained by the transformation. Thus, the two demodulated M-T currents can be used to extract the enhanced signature frequency components of faults, and the incipient fault detection of induction motors is easy to realize. With both simulated and experimental data of broken rotor bars, it shows that the proposed algorithm can extract more detailed fault signature frequency components and realize the incipient fault detection of induction motors. [ABSTRACT FROM AUTHOR]

In this paper, we use a circular distribution to quantify certain domain properties and model the hysteresis behavior of piezoceramic materials. The model is constructed by bridging the characteristics of microscopic domain distribution into the macroscopic (or bulk) behavior. Contributions, other than those associated with the polarization of domains, to bulk quantities are also counted. A domain orientation distribution function is first selected and the corresponding distribution function parameters are chosen as the internal state variables. For the two-dimensional model, a von Mises-Fisher circular distribution is used. Instead of micromechanical analysis of domain motions that would involve large computation efforts, the delineation of domain evolution is simplified by considering the evolution of the domain orientation distribution, which is determined by the dynamic variations of the internal state variables. We also develop a procedure to identify the material constants introduced in the constitutive equations. The models are used to quantitatively characterize various hysteresis loops observed in piezoceramic materials. [ABSTRACT FROM AUTHOR]

Self-heating or dissipation of piezoelectric ce- ramic elements is observed to be severe under dynamic operations even in the linear range. In this paper, a nonequilibrium thermodynamic model is developed to delineate the coupled irreversible mechanical, electric, and thermal processes, which jointly contribute to dissipation. Specifically, additional nonequilibrium state variables, also known as thermodynamic fluxes, are brought in to describe each of these processes. The characteristic relaxation of these processes is modeled. The nonnegative rate of entropy pro- duction is found to be in quadratic form of thermodynamics fluxes. The energy balance equation, which governs the transformation between different energy forms, is obtained in the framework of extended irreversible thermodynamics. Using this model, the dissipation of a piezoceramic stack actuator under harmonic electric or mechanical loadings in linear operation range is studied. The harmonic-balance methods are utilized as solution techniques. In contrast to the existing piezoelectric dissipation models, the dissipation by the developed model is verified to nonlinearly depend on operating frequency, with a peak dissipation occurring at some operating frequency that is related to characteristic relaxation of irreversible processes. The measurements of newly introduced parameters are also discussed. [ABSTRACT FROM AUTHOR]