Search

Sparse representation of signals have become an important tool in computer vision. In many computer vision applications such as image denoising, image super-resolution and object recognition, sparse representations have produced remarkable performances. Sparse representation models often contain two stages: sparse coding and dictionary learning. In this paper, we propose a non-linear non-negative sparse representation model: NNK-KSVD. In the sparse coding stage, a non-linear update rule is proposed to obtain the sparse matrix. In the dictionary learning stage, the proposed model extends the kernel KSVD by embedding the non-negative sparse coding. The proposed non-negative kernel sparse representation model was evaluated on several public image datasets for the task of classification. Experimental results show that by exploiting the non-linear structure in images and utilizing the ‘additive’ nature of non-negative sparse coding, promising classification performance can be obtained. Moreover, the proposed sparse representation method was also evaluated in image retrieval tasks, competitive results were obtained.

During the recent years, fashion landmark detection has attracted great attention of many researchers. The main task of Fashion landmark detection is to recognize the components of fashion in images, therefore it is essential for other fashion applications such as fashion classification, retrieval, design and recommendation. By utilizing various techniques in deep learning, the recently proposed landmark detection methods for fashion achieve notable performance, this motivates us to review the fashion landmark detection works. In this paper, we aim at roughly categorizing these recently proposed methods, discussing how these methods can improve detection performance. The fashion benchmark datasets used for landmark detection are reviewed in this paper as well. To our best knowledge, there is no a single review in the current literature that has discussed and reviewed fashion landmark detection.

Images are possibly degraded by various reasons, the typical forms of degradation are: blur, noise, low resolution, and etc. Image restoration techniques try to recover the degraded images to the original images with maximum fidelity. Image restoration is a challenging task and also an import research area in image processing. During the decades, researchers have proposed many restoration methods such as inverse filter, Weiner filter, wavelet analysis, support vector machine, and etc. Recently, deep learning has been increasingly popular among researchers and has obtained remarkable results. In this paper, we briefly review the approaches based on generative adversarial networks (GANs) for image restoration. The typical GANs based restoration methods for image super-resolution, image denoising, image inpainting and image deblurring are introduced and discussed.

Gearboxes are widely used in engineering machinery, but tough operation environments often make them subject to failure. And the emergence of periodic impact components is generally associated with gear failure in vibration analysis. However, effective extraction of weak impact features submerged in strong noise has remained a major challenge. Therefore, the paper presents a new adaptive cascaded stochastic resonance (SR) method for impact features extraction in gear fault diagnosis. Through the multi-filtered procession of cascaded SR, the weak impact features can be further enhanced to be more evident in the time domain. By analyzing the characteristics of non-dimensional index for impact signal detection, new measurement indexes are constructed, and can further promote the extraction capability of SR for impact features by combining the data segmentation algorithm via sliding window. Simulation and application have confirmed the effectiveness and superiority of the proposed method in gear fault diagnosis.

Traditional intelligent fault diagnosis techniques based on artificially selected features fail to make the most of the raw data information, and are short of the capabilities of feature self-learning. Moreover, the most informative and distinguished parts of the different faults signals only account for a small portion in the time domain and frequency domain signals. Therefore, in order to learn the discriminative features from the raw data adaptively, this paper proposes a multiscale local feature learning method based on back-propagation neural network (BPNN) for rolling bearings fault diagnosis. Based on the local characteristics of the fault features in the time domain and the frequency domain, the BPNN is used to locally learn meaningful and dissimilar features from signals of different scales, thus improving the fault diagnosis accuracy. Two sets of rolling bearing datasets are adopted to verify the validity and superiority of the proposed method by comparing with other methods.

This paper investigated a potential instrument for carbon disulfide in situ measurement with high precision and strong anti-interference capability. A compact and automated carbon disulfide detection system was developed using a fiber opto-electronic sensing device. A custom software interface based on LabVIEW was developed. The multi-wavelength least-squares method based on differential optical absorption spectroscopy was employed for improved detection and the anti-interference capabilities of the system. The detection limit of the system (signal-to-noise ratio=3) was determined to be 1 ppb per meter. Using this scheme, the reliability of a liquid-seal was verified to have carbon disulfide leakage. Although the liquid level of the liquid-sealed carbon disulfide showed no significant change over 24 h, a residue concentration of over 30 ppm remained detectable on the surface.

Image super-resolution is an important research field in image analysis. The techniques of image super-resolution has been widely used in many computer vision applications. In recent years, the success of deep learning methods in image super-resolution have attracted more and more researchers. This paper gives a brief review of recent deep learning based methods for single image super-resolution (SISR), in terms of network type, network structure, and training methods. The advantages and disadvantages of these methods are analyzed as well.

Many types of deep neural networks have been proposed to address the problem of human biometric identification, especially in the areas of face detection and recognition. Local deep neural networks have been recently used in face-based age and gender classification, despite their improvement in performance, their costs on model training is rather expensive. In this paper, we propose to construct a local deep neural network for age and gender classification. In our proposed model, local image patches are selected based on the detected facial landmarks; the selected patches are then used for the network training. A holistical edge map for an entire image is also used for training a “global” network. The age and gender classification results are obtained by combining both the outputs from both the “global” and the local networks. Our proposed model is tested on two face image benchmark datasets; competitive performance is obtained compared to the state-of-the-art methods.

In image super-resolution area, the effectiveness of sparse learning methods have been demonstrated, especially in single image super-resolution applications. A novel sparse learning method is proposed in this paper for the task of single image super-resolution. The proposed non-linear and nonnegative sparsity-based learning model is used to capture the nonlinear data distributions in images. Moreover, a boosting ensemble strategy is used to improve the learned ‘weak’ sparse model. In each round of boosting, the training images are chosen according to the reconstruction errors of the learned models from previous rounds. The obtained dictionary ensemble contains dictionaries which have different reconstruction capability on different types of image patches. Then the support vector regression is applied for obtaining the final result image. The proposed non-negative kernel sparse model was tested on popular benchmark images, the experimental results illustrate the competitiveness of the proposed method.

Sparse representations are widely used tools in image super-resolution (SR) tasks. In the sparsity-based SR methods, linear sparse representations are often used for image description. However, the non-linear data distributions in images might not be well represented by linear sparse models. Moreover, many sparsity-based SR methods require the image patch self-similarity assumption; however, the assumption may not always hold. In this paper, we propose a novel method for single image super-resolution (SISR). Unlike most prior sparsity-based SR methods, the proposed method uses non-linear sparse representation to enhance the description of the non-linear information in images, and the proposed framework does not need to assume the self-similarity of image patches. Based on the minimum reconstruction errors, support vector regression (SVR) is applied for predicting the SR image. The proposed method was evaluated on various benchmark images, and promising results were obtained.

a b s t r a c t A precise method for the measurement of diffuse reflectance by using gas absorption spectroscopy technique with an integrating sphere is demonstrated. A quantitative relationship between the diffuse reflectance and the gas absorption spectrum is formulated, which has been further validated by exper- iments. The precision of the reflectivity measurement depends on the diameter and port fraction of the integrating sphere as well as gas concentration, and it increases linearly with the magnitude of the reflectivity. A high precision of 0.005% was achieved at the reflectivity of 0.98844(5).

As an important data structure model, ontology has become one of the core contents in information science. Multi-dividing ontology algorithm combines the advantages of graph structure and learning algorithms proved to have high efficiency. In this paper, we investigate some theoretical problems of ontology algorithm in multi-dividing setting. The relationship between two versions of low noise assumptions is established. The risk excess and Lq-error are given. Specifically, the upper bound and lower bound minimax learning rate are obtained based on assumptions we describe in Section 2.

A novel Ho:LuAG ceramic laser Q-switched by a polycrystalline Cr2+:ZnS saturable absorber was reported for the first time in this paper. We took a diode-pumped Tm:YLF laser emitting at 1.9075 μm as the pump source. The laser operated in both continuous wave mode and passively Q-switching (PQS) mode. The maximum PQS output power of 2.67 W was obtained with a slope efficiency of 26.4%. When the absorbed pump power increased from 4.78 to 10.8 W, with three output couplers of T = 2%, T = 10% and T = 25%, the pulse widths decreased as the pump power increased, from 102.9 to 89.2 ns, from 147.1 to 127.6 ns, and from 173 to 150 ns, respectively, and the repetition frequency varied from 10.2 to 20.1 kHz, from 9.3 to 18.3 kHz, and from 8.45 to 16.66 kHz as well. The central wavelength remained constant 2100.64 nm with the change of output couplers and operation modes. Furthermore, the output laser had a beam quality factor M 2 of 1.1.

In this study, a sensitive system for simultaneous sensing of sulfur dioxide and carbon disulfide was developed based on absorption spectroscopy in the deep ultraviolet. An effective spectrum-unfolding approach is proposed to examine the overlapping spectral characteristics. Direct proportional relations with determination coefficients of 0.999 were obtained. The detection limit of sulfur dioxide was determined to be 42 ppb, and a detection limit of 5 ppb for carbon disulfide was achieved with an optical length of 20 cm. The interplay between the measurement results of the two components was investigated. Interference close to the detection limits was confirmed for both sulfur dioxide and carbon disulfide measurements. An automatic and reliable simultaneous sensing system for sulfur dioxide and carbon disulfide was constructed.

A simple cubic-shaped cavity with a high-diffuse-reflectivity inner coating as a novel gas detection cell was developed. The effective optical path length (EOPL) was evaluated by comparing the oxygen absorption signal in the cavity and in air based on tunable diode laser absorption spectroscopy. The law for a spherical cavity was applied and modified to a cubic cavity as a function of reflectivity ρ, port fraction f, and the side length. Single-pass average path length of the cubic cavity was 0.723(7) times the side length. EOPL can be modified conveniently by adjusting the parameters of the cavity.

Tunable multi-mode diode laser absorption spectroscopy for methane detection was demonstrated. The exclusive dependence of the 1318 nm laser modes distribution on the input current and temperature was justified. Stable absorption signals related to the methane concentrations were obtained based on second-harmonic detection technique. A real-time data recording and analyzing software program was developed to realize the on-line gas concentration monitoring. A measurement sensitivity of 25 ppm m and an accuracy of 0.27% of this system were achieved.

A method of integrating sphere effective optical path length (EOPL) evaluation using tunable diode laser absorption spectroscopy for gas detection was demonstrated. Oxygen was used as a sample gas for an 8.38 cm diameter integrating sphere calibration; 393.7 ± 1.3 cm EOPL was obtained from the wavelength modulation spectroscopy with second harmonic calibration by measuring oxygen P11 line at 764 nm, which is in agreement with that of 393 cm by using direct absorption spectroscopy calibration. The EOPL calibration accuracy of this method can reach 0.33 %. It has been justified that the EOPL of an integrating sphere is independent of the incident light intensity.

Sparse representations of signals have become an important tool in computer vision. In this paper, we propose a non-linear non-negative sparse representation model: NNK-KSVD. In the sparse coding stage, a non-linear update rule is proposed to obtain the sparse matrix. In the dictionary learning stage, the proposed model extended the kernel KSVD by embedding the non-negative sparse coding. The proposed non-negative kernel sparse representation model was evaluated on two public image datasets for image retrieval, promising image retrieval performance was obtained.

PurposeThe task of internet intrusion detection is to detect anomalous network connections caused by intrusive activities. There have been many intrusion detection schemes proposed, most of which apply both normal and intrusion data to construct classifiers. However, normal data and intrusion data are often seriously imbalanced because intrusive connection data are usually difficult to collect. Internet intrusion detection can be considered as a novelty detection problem, which is the identification of new or unknown data, to which a learning system has not been exposed during training. This paper aims to address this issue.Design/methodology/approachIn this paper, a novelty detection‐based intrusion detection system is proposed by combining the self‐organizing map (SOM) and the kernel auto‐associator (KAA) model proposed earlier by the first author. The KAA model is a generalization of auto‐associative networks by training to recall the inputs through kernel subspace. For anomaly detection, the SOM organizes the prototypes of samples while the KAA provides data description for the normal connection patterns. The hybrid SOM/KAA model can also be applied to classify different types of attacks.FindingsUsing the KDD CUP, 1999 dataset, the performance of the proposed scheme in separating normal connection patterns from intrusive connection patterns was compared with some state‐of‐art novelty detection methods, showing marked improvements in terms of the high intrusion detection accuracy and low false positives. Simulations on the classification of attack categories also demonstrate favorable results of the accuracy, which are comparable to the entries from the KDD CUP, 1999 data mining competition.Originality/valueThe hybrid model of SOM and the KAA model can achieve significant results for intrusion detection.

Accurate and reliable classification of microscopic biopsy images is an important issue in computer assisted breast cancer diagnosis. In this paper, a new cascade Random Subspace ensembles scheme with reject options is proposed for microscopic biopsy image classification. The classification system is built as a serial fusion of two different Random Subspace classifier ensembles with rejection options to enhance the classification reliability. The first ensemble consists of a set of Support Vector Machine classifiers that converts the original $$K$$ -class classification problem into a number of $$K$$ 2-class problems. The second ensemble consists of a Multi-Layer Perceptron ensemble, that focuses on the rejected samples from the first ensemble. For both of the ensembles, the reject option is implemented by relating the consensus degree from majority voting to a confidence measure, and abstaining to classify ambiguous samples if the consensus degree is lower than some threshold. We also investigated the effectiveness of a feature description approach by combining Local Binary Pattern (LBP) texture analysis, statistics derived using the Gray Level Co-occurrence Matrix (GLCM) and the Curvelet Transform. While the LBP analysis efficiently describes local texture properties and the GLCM reflects global texture statistics, the Curvelet Transform is particularly appropriate for the representation of piece-wise smooth images with rich edge information. The combined feature description thus provides a comprehensive biopsy image characterization by taking advantages of their complementary strengths. Using a benchmark microscopic biopsy image dataset, obtained from the Israel Institute of Technology, a high classification accuracy of $$99.25 \%$$ was obtained (with a rejection rate of $$1.94 \%$$ ) using the proposed system.

A model test of the portals of two parallel tunnels is carried out to learn about the dynamic response of tunnel liner and the interaction between surrounding rock and liner in earthquakes. The experiment results show that: first, when the seismic acceleration traverses the model material, the low-frequency segment of seismic acceleration is magnified and the high-frequency segment of seismic acceleration is attenuated; second, the horizontal shear failure of the surrounding rock is caused by the interaction between the surrounding rock and the tunnel liner, and the cracks in the surrounding rock grow nearly in the same direction, however, because of the different constraints on the tunnel liner by the surrounding rock outside the tunnel, the destruction degree is different; third, the liner cracks of the left tunnel with short length appear mainly at the left tunnel entrance, the cracks of right tunnel with large length appear mainly at the right tunnel entrance and the tunnel cross-section nearly which is in the same vertical plane with the left tunnel portal, and the liner cracks are distributed mainly on the closer side of two liners between the two holes; finally, in the same vertical testing cross-section, the liner maximal strain at the inner sides between two tunnels is greater than outer sides. In addition, the cross-section maximal strain on the right tunnel decreases with the increasing distance between the tested cross-section and a reference vertical plane containing the left tunnel portal.

Wavelet-based image denoising is an important technique in the area of image noise reduction. Wavelets have their natural ability to represent images in a very sparse form which is the foundation of wavelet-based denoising through thresholding. This paper explores properties of several representative thresholding techniques in wavelets denoising, such as VisuShrink, SureShrink, BayesShrink and Feature Adaptive Shrinkage. A quantitative comparison between these techniques through PSNR (Peak Signal-to-Noise Ratio) is also given.

A new sensor — transverse electric (TE) polarized excite surface plasmon resonance (SPR) fiber optic biosensor is proposed. The graphene is the plasma layer. The transfer matrix method and the finite difference time domain method are applied to conduct the numerical simulation of the four layers (fiber core/photonic crystals/graphene/sample) of fiber optic biosensor. The results show that the relationship between refractive index and resonant wavelength is linear and the sensitivity of the fiber optic biosensor reaches 1942 nm/RIU.

We derive the dispersion relation of coupled-resonator optical waveguides (CROWs) without approximation. And, we use the exact dispersion relation of CROWs established to calculate the group delay of microring CROWs and obtain a result similar to the experimental result reported by Poon et al. Further, through numerical simulation with the parameters used to simulate the experimental result, we found that the output group delay of microring CROWs could be adjusted continuously by changing the argument of the coupling coefficient θ resulting from the shift of the dispersion band. But, the adjustment of output group delay was not linear and meticulous control of θ could lead to a more favorable adjustment of the output group delay. The continuous adjustment of group delay is of great significance for applications of microring CROWs in delay lines and optical buffers of future all-optical communication systems.

Recently, many researchers had reported their work about sub- and superluminal propagation. And several experiments had demonstrated the signal evolution of slow and fast light. In this letter, the authors described a simple experiment for the tracing of the light signal in saturated absorption (SA) and reverse saturated absorption (RSA) media, firstly. We had, directly, observed the evolution of slow and fast signals with different waveforms that traveled in a ruby crystal and C60. Through tracking the signal when slow light (vg≪c) and fast light (vg

Source separation of musical signals is an appealing but difficult problem, especially in the single-channel case. In this paper, an unsupervised single-channel music source separation algorithm based on average harmonic structure modeling is proposed. Under the assumption of playing in narrow pitch ranges, different harmonic instrumental sources in a piece of music often have different but stable harmonic structures; thus, sources can be characterized uniquely by harmonic structure models. Given the number of instrumental sources, the proposed algorithm learns these models directly from the mixed signal by clustering the harmonic structures extracted from different frames. The corresponding sources are then extracted from the mixed signal using the models. Experiments on several mixed signals, including synthesized instrumental sources, real instrumental sources, and singing voices, show that this algorithm outperforms the general nonnegative matrix factorization (NMF)-based source separation algorithm, and yields good subjective listening quality. As a side effect, this algorithm estimates the pitches of the harmonic instrumental sources. The number of concurrent sounds in each frame is also computed, which is a difficult task for general multipitch estimation (MPE) algorithms.

The anaplastic lymphoma kinase (ALK) rearrangements, mostly EML4-ALK fusion, occur in 3-7% of lung cancer patients and define a patient population that could respond to receptor tyrosine kinase inh...

Sparse representation of signals have become an important tool in computer vision. In many applications in computer vision, such as image denoising, image super-resolution and object recognition, sparse representations have produced remarkable performance. In this paper, we propose a non-linear non-negative sparse coding model NNK-KSVD. The proposed model extended the kernel KSVD by embedding the non-negative sparse coding. Experimental results show that by exploiting the non-linear structure in images and utilizing the ‘additive’ nature of non-negative sparse coding, promising classification performance can be obtained.

An optical fiber surface plasmon resonance biosensor is presented that allows to numerically demonstrate, using transfer matrix method and the finite difference time domain method, the detection range is very wide. Two different structures of graphene photonic crystal multilayer (i.e. sensor I and sensor II) are constructed in the cladding region of single-mode fiber. Graphene is used as the plasma layer instead of the traditional metal. According to the analysis, the properties of graphene can be changed by adjusting the chemical potential µ c . In the spectral region of 1.667| µ c | ћω µ c |, the imaginary part of conductivity σ ″ becomes negative. Thus the weakly bounded low-less TE-SPR is supported by graphene. The results of the numerical simulation show that the relationship between refractive index and resonant wavelength is linear. The sensor I can detect the refractive index range of 1.33–1.4, and the sensitivity is 1942 nm/RIU. The sensor II can detect the refractive index range of 1.41–1.67, and the sensitivity is up to 2315.4 nm/RIU. Therefore, the detection of wide refractive index range of 1.33–1.67 or simultaneous detection of different biological medium concentration is realized by the sensor.

A method for measuring diffuse reflectivity using cubic cavity based on the variable port fraction method was developed by measuring oxygen P11 line at 762 nm using tunable diode laser absorption spectroscopy. An experimental method to determine the additional path length l0 was presented. We measured the diffuse reflectivity of a cubic cavity with scattering coatings of different thickness. The error of diffuse reflectivity was reduced from 0.004 to 0.0003 when the diffuse reflectivity increased from 0.867(4) to 0.9887(3). A simulation result manifests that the error of diffuse reflectivity has the potential to be further reduced at higher diffuse reflectivity.

To investigate microvessel density (MVD), vascular endothelial growth factor (VEGF), endothelial nitric oxide synthase (eNOS), and interleukin-6 (IL-6) mRNA expression in peritoneal tissues, and their relationships with baseline peritoneal transport in uremia. Thirty uremic patients with a peritoneal dialysis catheter were selected in the Department of Nephrology in Dalian Central Hospital, Liaoning, China between 2010 and 2012. Peritoneal specimens were harvested for assessment of MVD, VEGF, eNOS, and IL-6 mRNA expression. One month after continuous ambulatory peritoneal dialysis, a peritoneal equilibration test was conducted. According to the 4-h peritoneal dialysate and plasma creatinine ratio (D/P Cr), patients were divided into high (n = 16) and low (n = 14) transport groups. General clinical data of high and low transport groups were similar (P > 0.05). The MVD in peritoneal tissues was significantly higher in the high than in the low transport group (P

Histological image is important for diagnosis of breast cancer. In this paper, we present a novel automatic breast cancer classification scheme based on histological images. The image features are extracted using the Curvelet Transform, statistics of Gray Level Co-occurrence Matrix (GLCM) and the Completed Local Binary Patterns (CLBP), respectively. The three different features are combined together and used for classification. A classifier ensemble approach, called Random Subspace Ensemble (RSE), are used to select and aggregate a set of base neural network classifiers for classification. The proposed multiple features and random subspace ensemble offer the classification rate 95.22% on a publically available breast cancer image dataset, which compares favorably with the previously published result 93.4%.

All non-small cell lung cancer (NSCLC) patients are recommended to be screened for anaplastic lymphoma kinase (ALK)-rearrangement despite its low occurrence (< 7%). This is due to recent advances in treatment for patients with ALK-positive NSCLC with receptor tyrosine kinase inhibitors Crizotinib and Ceritini. Current rearranged-ALK testing includes fluorescence in situ hybridization (FISH), reverse transcription polymerase chain reaction (RT-PCR), and immunohistochemistry (IHC). Despite the high sensitivities and specificities, the limitations of the first two assays include time-consuming, infeasible, and specialized techniques that make them unsuitable for the routine screening for which IHC has been proposed. However, the biggest obstacle to detecting rearranged-ALK in lung cancer patients by IHC is the lack of highly sensitive ALK antibodies that detect the extremely low abundance of ALK in patients. To overcome this limitation, we have developed an ALK mouse monoclonal antibody (clone OTI1A4) with higher sensitivity in comparison to a rabbit ALK antibody (clone D5F3). Moreover, OTI1A4 identified ALK-positive NSCLC samples (16/16 cases) and showed negative staining for ALK-negative samples (11/11 cases) previously validated by FISH (10/16 ALK-positive cases) or qPCR (10/16 ALK-positive cases). This indicated the potential advantage of using clone OTI1A4 over FISH or qPCR to detect ALK rearrangements. Lastly, there was 100% concordance between OTI1A4 and D5F3 in detecting ALK-rearranged NSCLC samples previously confirmed by FISH (76 ALK-positive and 438 ALK-negative samples). These results support the possibility that ALK clone OTI1A4 could be used for routine screening of patients with ALK-positive NSCLC by IHC. Citation Format: Hsiangmin Lu, Rachel Gonzalez, Yi Shen, Mu-lan Jin, Yipan Wu, Yungang Zhang, Kehu Yuan, Boyang Chu, Lili Qi, Huibo Liu, Chenlin Wang, Guangli Wang, Youmin Shu, Julie McDowell, Donghui Ma, Wei-wu He, Jian Chen, Ray Lin. A new, highly sensitive ALK antibody improves the screening of rearranged-ALK by IHC. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 415.

Objective To evaluate whether saw palmetto extract (SPE) relaxes corpus cavernosum and explore the underlying mechanisms. Methods Forty Sprague-Dawley rats and 30 New Zealand rabbits were randomly allocated into 3 SPE-treated groups (low-, middle-, and high-dose) and 1 saline-treated control group. SPE was administered intragastrically for 7 consecutive days. Another 23 rats treated with sildenafil were used to appraise the erectile response to electrical stimulation of nerves in the corpus cavernosum. The erectile functions of rats and rabbits were evaluated 24 hours after the last SPE administration or 15 minutes after intragastric sildenafil. Outcome measures included corpus cavernosum electrical activity recording, phosphodiesterase 5 (PDE5) activity detected by the colorimetric quantitative method, and messenger ribonucleic acid (mRNA) expression level for PDE5 and inducible nitric oxide synthase (iNOS) determined using real-time polymerase chain reaction. Results In the SPE-treated animals, the relaxant response to electrical stimulation of nerves in the corpus cavernosum, reflected by the amplitude of the electrical activity within the cavernosum, was significantly and dose-dependently augmented. Similar effects were observed in the sildenafil-treated rats. PDE5 activity in rat and rabbit corpus cavernosum tissues was significantly and dose-dependently inhibited in SPE-treated animals, whereas the iNOS mRNA level increased compared with the saline group. PDE5 mRNA, however, was only significantly enhanced in the rats treated with the middle dose of SPE. Conclusion The results suggest that SPE may have potential application value for the prevention or treatment of erectile dysfunction through an increase in iNOS mRNA expression and inhibition of PDE5 activity in corpus cavernosum smooth muscles.

Accuracy and reliability are two important issues in computer assisted breast cancer diagnosis. In this paper, a new cascade Random Subspace ensembles scheme with reject options is proposed for automatic breast cancer diagnosis. The diagnosis system is built as a serial fusion of two different Random Subspace classifier ensembles with rejection options to enhance the classification reliability. The first ensemble consists of a set of Support Vector Machine (SVM) classifiers that converts the original K-class classification problem into a number of K 2-class problems. The second ensemble consists of a Multi-Layer Perceptron (MLP) ensemble, that focuses on the rejected samples from the first ensemble. For both of the ensembles, the reject option is implemented by relating the consensus degree from majority voting to a confidence measure, and abstaining to classify ambiguous samples if the consensus degree is lower than some threshold. Using a microscopic breast biopsy image dataset from Israel Institute of Technology and benchmark datasets from UCI, promising results are obtained using the proposed system.

Color and shape information have been two important image descriptors in Content Based Image Retrieval (CBIR) systems. The focus of this research is to find a method representing images with color and shape information in the way of human visual perception. The image retrieval approach proposed here depends on the color and shape features extracted by color Vector Quantization (VQ) and the Digital Curvelet Transform (DCT), respectively. The extracted color and shape features were combined and weighted by Genetic Algorithm (GA), then used for image similarity measurement. Experimental results show that the GA combined features can bring about improved image retrieval performance.

Histological image is important for diagnosis of breast cancer. In this paper, we present a novel automatic breaset cancer classification scheme based on histological images. The image features are extracted using the Curvelet Transform, statistics of Gray Level Co‐occurence Matrix (GLCM) and Completed Local Binary Patterns (CLBP), respectively. The three different features are combined together and used for classification. A classifier ensemble approach, called Random Subspace Ensemble (RSE), are used to select and aggregate a set of base neural network classifiers for classification. The proposed multiple features and random subspace ensemble offer the classification rate 95.22% on a publically available breast cancer image dataset, which compares favorably with the previously published result 93.4%.

Color and shape descriptions of an image are the most widely used visual features in content-based image retrieval systems. Feature vectors for shape and color can be combined to improve the performance of the content-based image retrieval systems. In this paper, a novel image retrieval method integrating HSV color quantization and curve let transform is proposed. By analyzing properties of HSV(Hue, Saturation, Value) color space, a new dividing method to quantize the HSV color space into 24 non-uniform bins based on HSV soft decision is introduced and used for color histogram generation. Digital curve let transform is employed for extracting shape features in images, as it has been proved that the curve let transform is an almost optimal sparse representation of objects with edges. The generated HSV color histogram and the curve let feature are then combined and weighted for image retrieval, using Manhattan distance metric as the similiarity measure. Experiments on an image database of 565 images show that the combined feature performs well in precision and adaptability.

Wavelet thresholding is an effective way of image denoising and enhancement. The most important issue in wavelet thresholding is how to find an optimal threshold. In this paper, an adaptive threshold selection technique is proposed and morphological operations to improve the denoised result are discussed. An image denoising and enhancement scheme based on the adaptive wavelet shrinkage and mathematical morphology is described. Compared with some existing denoising methods such as VisuShrinkage, BayesShrinkage, the experimental result shows the proposed method outperforms these techniques in terms of PSNR (Peak Signal to Noise Ratio) and MSE (Mean Square Error).

An object based image retrieval scheme is described. In this scheme, to get objects in an image is the first step, which indeed is an image split and merge process. To control image segmentation, definition of LPD is given. After the split, merge the sub-blocks to get some homogeneous regions called objects. In objects detection, the concepts of inside block and outside block are proposed, a method to choose the seed of regions and proper growth criterion is also presented. Based on the objects get from the image, some important features can be extracted to measure similarity between images.

We report on a monitoring technique for nitrogen dioxide based on broadband absorption spectroscopy using a blue light-emitting diode (LED) operating around 465 nm. The technique is suited for real-time measurements of nitrogen dioxide due to the use of a straightforward data evaluation method, limited interference from other gases, and a low degree of complexity compared with other real-time optical detection techniques having the same precision. Additionally, the use of a LED can reduce the cost of nitrogen dioxide monitoring. Real-time measurements of nitrogen dioxide concentration were demonstrated at atmospheric pressure, which is of great interest for industrial nitrogen dioxide emission monitoring; a detection limit of about 3 ppm using a 50-cm-long gas cell with 2 s integration time was achieved.

A spectroscopic system for continuous real-time monitoring of SO(2) concentrations in industrial emissions was developed. The sensor is well suited for field applications due to simple and compact instrumental design, and robust data evaluation based on ultraviolet broadband absorption without the use of any calibration cell. The sensor has a detection limit of 1 ppm, and was employed both for gas-flow simulations with and without suspended particles, and for in situ measurement of SO(2) concentrations in the flue gas emitted from an industrial coal-fired boiler. The price/performance ratio of the instrument is expected to be superior to other comparable real-time monitoring systems.

Cluster analysis is a fundamental technique in pattern recognition. It is difficult to cluster data on complex data sets. This paper presents a new algorithm for clustering. There are three key ideas in the algorithm: using mutual neighborhood graphs to discover knowledge and cluster data; using eigenvalues of local covariance matrixes to express knowledge and form a knowledge embedded space; and using a denoising trick in knowledge embedded space to implement clustering. Essentially, it learns a new distance metric by knowledge embedding and makes clustering become easier under this distance metric. The experiment results show that the algorithm can construct a quality neighborhood graph from a complex and noisy data set and well solve clustering problems.

Character segmentation is an important step in License Plate Recognition (LPR) system. There are many difficulties in this step, such as the influence of image noise, plate frame, rivet, space mark, and so on. This paper presents a new algorithm for character segmentation, using Hough transformation and the prior knowledge in horizontal and vertical segmentation respectively. Furthermore, a new object enhancement technique is used for image preprocessing. The experiment results show a good performance of this new segmentation algorithm.

Sodium bismuth titanate Na0.5Bi0.5TiO3 (NBT) single crystal doped with Nd3+ was grown by a top-seeded solution growth method. Powder x-ray diffraction revealed a pure perovskite structure with the rhombohedral phase. We found that the dielectric and ferroelectric properties were enhanced by the Nd3+ dopant. After poling along the [1?1?1] direction, transmittance was enhanced dramatically. The Sellmeier dispersion equation and energy band gaps were obtained. The absorption band around 808?nm has high full-width at half-maximum and large absorption cross-section, which is suitable for AlGaAs diode-laser pumping. A strong emission transition band of Nd3+ at around 1066?nm was observed; a long radiation lifetime 324??s shows a low quenching effect. These results indicate that Nd3+-doped NBT crystal could be applied in photonic or integrated optoelectronic devices as a multi-functional crystal.

Classification of medical images is an important issue in computer-assisted diagnosis. In this paper, a classification scheme based on a one-class kernel principle component analysis (KPCA) model ensemble has been proposed for the classification of medical images. The ensemble consists of one-class KPCA models trained using different image features from each image class, and a proposed product combining rule was used for combining the KPCA models to produce classification confidence scores for assigning an image to each class. The effectiveness of the proposed classification scheme was verified using a breast cancer biopsy image dataset and a 3D optical coherence tomography (OCT) retinal image set. The combination of different image features exploits the complementary strengths of these different feature extractors. The proposed classification scheme obtained promising results on the two medical image sets. The proposed method was also evaluated on the UCI breast cancer dataset (diagnostic), and a competitive result was obtained.

After publication of our work [1] we noticed that there were three references and an acknowledgment missing. These are given below: Figure 5 and Figure 6 should be referenced with reference number 53 from the reference list in [1]. The first sentence in section 4.1.2 ‘The 3D OCT retinal image set was collected at the Royal Hospital of University of Liverpool’ should be referenced with [2]. The Acknowledgments should also state that ‘The authors appreciate Dr. Yalin Zheng and Mr. Albarrark A. for their support on the 3D OCT retinal image set used in the paper’.