In this paper, a novel design of theWideband Bandstop Filter, based on Split Ring Resonators, (SRR) is proposed. The Defect Ground Structure (DGS) is achieved by opening the SRR-shaped slot in the ground plane, and th...
详细信息
How to fast, accurately and robustly recognize wheat diseases, particularly for those diseases with mild-to-moderate severity, is a challenge for prevention and control of crop disease timely. In this study, image pro...
详细信息
How to fast, accurately and robustly recognize wheat diseases, particularly for those diseases with mild-to-moderate severity, is a challenge for prevention and control of crop disease timely. In this study, image processing technique was applied to segment the infected regions of disease leaves. Twenty disease features were extracted, and eighteen larger weight features were selected by Relief-F algorithm to generate the models of Support Vector Machine (SVM), Relevance Vector Machine (RVM) and Back Propagation Neural Network (BPNN). Subsequently, these models were used to identify two kinds of wheat diseases, namely, wheat stripe rust and powdery mildew. Total 136 samples, including 68 training samples and 68 test samples with different infection severities were used to study the recognition capabilities of the three models. Results showed that high predictive accuracies in identification of two wheat diseases with varying severity for all three models. Overall accuracy of RVM was 89.71%, which was superior to 83.82% of SVM and inferior to 92.64% of BPNN. Meanwhile, the recognition accuracies of SVM, RVM and BPNN models for mild-to-moderate disease were 83.33%, 88.33% and 91.67%, respectively. The prediction time of RVM was less than those of SVM and BPNN, with differences as large as 7.96 and 31.68 times, respectively. Therefore, RVM appeared to be the most suitable for real-time identifying wheat leaf diseases among the three models, which can provide important technical support for wheat diseases management.
Regarding the performance of traditional endpoint detection algorithms degrades as the environment noise level increases, a recursive calculating algorithm for higher-order cu- mulants over a sliding window is propose...
详细信息
Regarding the performance of traditional endpoint detection algorithms degrades as the environment noise level increases, a recursive calculating algorithm for higher-order cu- mulants over a sliding window is proposed. Then it is applied to the speech endpoint detection. Furthermore, endpoint detection is carried out with the feature of energy. Experimental results show that both the computational efficiency and the robustness against noise of the proposed algorithm are improved remarkably compared with traditional algorithm. The average prob- ability of correct point detection (Pc-point) of the proposed voice activity detection (VAD) is 6.07% higher than that of G.729b VAD in different noisy at different signal-noise ratios (SNRs) environments.
Now, the highway toll system still uses a single license plate recognition, this method has a problem of inaccurate identificationFor this kind of situation, this paper put forward to increase the appearance of the ve...
详细信息
ISBN:
(纸本)9781510828087
Now, the highway toll system still uses a single license plate recognition, this method has a problem of inaccurate identificationFor this kind of situation, this paper put forward to increase the appearance of the vehicle feature information and can improve the accuracy of recognitionIn this paper, we adopt the ORB algorithm to extract the exterior feature information of the vehicle and two-way matching、RANSAC algorithms to remove mismatching pointsAt the same time, we continue to iteration the scale parameter of the affine transformation and rotation angle at the matching point as a kind of judgment, which improves the robustness of the algorithm.
In this paper, a novel design of the dual passband lowpass filter based on the Defect Ground Structure (DGS) is proposed. The defect ground structure (DGS) is achieved by opening a set of asymmetric U-shaped slot in t...
详细信息
Various kinds of information in large touch screen system need to be protected by various reasons so we need to hide them. One kind of image hiding selection algorithm is proposed here. The steps are as follows, the c...
详细信息
ISBN:
(纸本)9781509041039
Various kinds of information in large touch screen system need to be protected by various reasons so we need to hide them. One kind of image hiding selection algorithm is proposed here. The steps are as follows, the carrier image and secret message is obtained by pre-processing to get a most hiding matrix suitable for image and message, then secret message is hidden into the carrier image by using Kim's method. The final hidden image and the original carrier image is almost the same from the naked eye. Finally, a few tests are carried out and the test results suggest that the hidden effect of the whole scheme is good.
The finite-difference time-domain (FDTD) method is proposed to model Maxwell-Bloch equations with a four-level quantum system for investigating the nano-structures incorporated with gain materials. To understand the l...
详细信息
The finite-difference time-domain (FDTD) method is proposed to model Maxwell-Bloch equations with a four-level quantum system for investigating the nano-structures incorporated with gain materials. To understand the lasing behavior and loss compensation mechanism of the gain material, a self-consistent computational scheme is presented. Comparisons between optical and homogeneous Pumping are given. Our self-consistent computational scheme can be used to design new loss-compensation structures with metamaterials.
Low-rank tensor factorization(LRTF) provides a useful mathematical tool to reveal and analyze multi-factor structures underlying data in a wide range of practical applications. One challenging issue in LRTF is how to ...
详细信息
Low-rank tensor factorization(LRTF) provides a useful mathematical tool to reveal and analyze multi-factor structures underlying data in a wide range of practical applications. One challenging issue in LRTF is how to recover a low-rank higher-order representation of the given high dimensional data in the presence of outliers and missing entries, i.e., the so-called robust LRTF problem. The L1-norm LRTF is a popular strategy for robust LRTF due to its intrinsic robustness to heavy-tailed noises and outliers. However, few L1-norm LRTF algorithms have been developed due to its non-convexity and non-smoothness, as well as the high order structure of data. In this paper we propose a novel cyclic weighted median(CWM) method to solve the L1-norm LRTF problem. The main idea is to recursively optimize each coordinate involved in the L1-norm LRTF problem with all the others fixed. Each of these single-scalar-parameter sub-problems is convex and can be easily solved by weighted median filter, and thus an effective algorithm can be readily constructed to tackle the original complex problem. Our extensive experiments on synthetic data and real face data demonstrate that the proposed method performs more robust than previous methods in the presence of outliers and/or missing entries.
In recent years, researchers have proposed several methods to transform time series (such as those of fractional Brownian motion) into complex networks. In this paper, we construct horizontal visibility networks (HVNs...
In recent years, researchers have proposed several methods to transform time series (such as those of fractional Brownian motion) into complex networks. In this paper, we construct horizontal visibility networks (HVNs) based on the -stable Lévy motion. We aim to study the relations of multifractal and Laplacian spectrum of transformed networks on the parameters and of the -stable Lévy motion. First, we employ the sandbox algorithm to compute the mass exponents and multifractal spectrum to investigate the multifractality of these HVNs. Then we perform least squares fits to find possible relations of the average fractal dimension , the average information dimension and the average correlation dimension against using several methods of model selection. We also investigate possible dependence relations of eigenvalues and energy on , calculated from the Laplacian and normalized Laplacian operators of the constructed HVNs. All of these constructions and estimates will help us to evaluate the validity and usefulness of the mappings between time series and networks, especially between time series of -stable Lévy motions and HVNs.
We show that regular homogeneous two-weight Zpk -codes where p is odd and k ≥ 2 with dual Hamming distance at least four do not exist. The proof relies on existence conditions for the strongly regular graph built on ...
详细信息
暂无评论