Face recognition by using feature-specific imaging

We present a face-recognition system based on the optical measurement of linear features. We describe a polarization-based optical system that computes linear projections of an incident irradiance distribution. We quantify the fundamental limitations of optical feature measurement. We find that higher feature fidelity can be obtained by feature-specific imaging than by postprocessing a conventional image. We present feature-fidelity results for wavelet, principal component, and Fisher features. We study face recognition by using a k-nearest neighbors classifier and two different feed-forward neural networks. Each image block is reduced to either a one- or a two-dimensional feature space for input to these recognition algorithms. As high as 99% recognition has been achieved with one-dimensional wavelet feature projections and 100% has been achieved with two-dimensional projections. A 95-fold increase in noise tolerance by use of feature-specific imaging has been demonstrated for an example of the face-recognition problem. An optical experiment is performed to validate these results.     

Feature-specific imaging

We analyze the performance of feature-specific imaging systems. We study incoherent optical systems that directly measure linear projects of the optical irradiance distribution. Direct feature measurement exploit, the multiplex advantage, and for small numbers of projections can provide higher feature-fidelity than those systems that postprocess a conventional image. We examine feature-specific imaging using Wavelet, Karhunen-Loeve (KL), Hadamard, and independent-component features, quantifying feature fidelity in Gaussian-, shot-, and quantization-noise environments. An example of feature-specific imaging based on KL projections is analyzed and demonstrates that within a high-noise environment it is possible to improve image fidelity via direct feature measurement. A candidate optical system is presented and a preliminary implementational study is undertaken.     

线搜索下带误差项的Dai-Yuan共轭梯度算法

本文在共轭梯度不能精确计算的情况下,采用Wolfe或Armijo步长规则研究了带误差项的Dai-Yuan(abbr．DY)共轭梯度法,我们的方法的一个很重要的特征就是步长不一定趋于零。这种特征使得我们的分析对许多实际问题很有用。我们在很一般的假设条件下证明了算法的全局收敛性。最后给出了数值算例。     

非平稳信号特征提取在航空发动机故障诊断中的应用

在航空发动机早期故障诊断中,特征提取是早期诊断的重要过程之一.文中以航空发动机转子故障为研究对象,给出了基于经验模式分解、小波分析为核心的故障特征提取方法,并作了针对性的比较研究.在matlab7.0环境下开发了一个故障特征提取软件系统.研究结果表明:基于经验模式分解的时频分析方法可以很有效地提取到非平稳故障特征信号,是一种适合于非线性信号处理的方法.     

A combination of rough-based feature selection and RBF neural network for classification using gene expression data

This paper presents a novel rough-based feature selection method for gene expression data analysis. It can find the relevant features without requiring the number of clusters to be known a priori and identify the centers that approximate to the correct ones. In this paper, we attempt to introduce a prediction scheme that combines the rough-based feature selection method with radial basis function neural network. For further consider the effect of different feature selection methods and classifiers on this prediction process, we use the NaIve Bayes and linear support vector machine as classifiers, and compare the performance with other feature selection methods, including information gain and principle component analysis. We demonstrate the performance by several published datasets and the results show that our proposed method can achieve high classification accuracy rate.     

Content-based searching of multimedia databases by use of approximate digital signatures

We propose the use of approximate digital signatures of selected multimedia feature vectors for fast content-based retrieval in large multimedia databases. We adapt and extend the approximate message authentication code (AMAC), introduced by some of the authors recently in the area of message authentication, to the multimedia searching problem. An AMAC is a binary signature with the ability to reflect changes in the message it represents. The Hamming distance between two AMACs is used to measure the degree of the similarity between multimedia objects. We develop a method to compress AMAC signatures to create a direct look-up table that allows for fast searching of a database. The color histogram is used as the example feature space to show how the signature is applied. Experimental results show that the performance of the proposed method is comparable with existing methods based on other popular metrics, but it significantly decreases search time.     

Face modeling and editing with statistical local feature control models

This article presents a novel method based on statistical facial feature control models for generating realistic controllable face models. The local feature control models are constructed based on the exemplar 3D face scans. We use a three‐step model fitting approach for the 3D registration problem. Once we have a common surface representation for examples, we form feature shape spaces by applying a principal component analysis (PCA) to the data sets of facial feature shapes. We compute a set of anthropometric measurements to parameterize the exemplar shapes of each facial feature in a measurement space. Using PCA coefficients as a compact shape representation, we approach the shape synthesis problem by forming scattered data interpolation functions that are devoted to the generation of desired shape by taking the anthropometric parameters as input. The correspondence among all exemplar face textures is obtained by parameterizing a 3D generic mesh over a 2D image domain. The new feature texture with desired attributes is synthesized by interpolating the exemplar textures. With the exception of an initial tuning of feature point positions and assignment of texture attribute values, our method is fully automated. In the resulting system, users are assisted in automatically generating or editing a face model by controlling the high‐level parameters. © 2008 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 17, 341–358, 2007     

Multiple SVM-RFE for gene selection in cancer classification with expression data

This paper proposes a new feature selection method that uses a backward elimination procedure similar to that implemented in support vector machine recursive feature elimination (SVM-RFE). Unlike the SVM-RFE method, at each step, the proposed approach computes the feature ranking score from a statistical analysis of weight vectors of multiple linear SVMs trained on subsamples of the original training data. We tested the proposed method on four gene expression datasets for cancer classification. The results show that the proposed feature selection method selects better gene subsets than the original SVM-RFE and improves the classification accuracy. A Gene Ontology-based similarity assessment indicates that the selected subsets are functionally diverse, further validating our gene selection method. This investigation also suggests that, for gene expression-based cancer classification, average test error from multiple partitions of training and test sets can be recommended as a reference of performance quality.     

The Munich Feature Film Cluster: The Degree of Global Integration and Explanations for its Relative Success

In this paper we assess the role that a secondary feature film cluster can play in an industrial sector largely dominated by the major Hollywood studios. In order to do this we proceed in two steps. We first evaluate as to how Munich-based firms are integrated into the production, distribution, financing and technological development of globally successful feature films. Then we investigate the factors that explain the relative economic success of this regional industry. Here we argue that major corporate actors that established a core group within the cluster have turned the industry towards specialized niche products and have been significantly supported by public policy measures.     

Multifeature distortion-insensitive constellation detection

Many applications require detection of multiple features that locally remain consistent in shape and intensity characteristics, but may globally change position with respect to one another over time or under different circumstances. We refer to these feature sets, defined by their characteristic relative positioning, as multifeature constellations. We introduce a method of processing in which multiple levels of correlation, using specially designed composite feature detection filters, are used to first detect local features, and then to detect constellations of these local features. We include experimental procedures and results indicating how the use of multifeature constellation detection may be utilized in applications such as sign language recognition and fingerprint matching.     

A novel image analysis method for rotational motion of circular feature based on perspective projection

Abstract

Circular feature has been widely applied in recognising moving object and estimating its position and orientation. Based on the principle of perspective projection, the rotational motion of circular feature was analysed. According to whether the rotational motion occurred in the plane of circular feature, the rotational motion could be classified into two types: in-plane rotation and non-in-plane rotation. For the convenience of computation, we set the camera’s initial position and orientation. The optical axis of camera was set perpendicular to the plane of circular feature. We analysed the perspective distortion of the circular feature. Additionally, based on the geometrical reasoning, we also proposed the method of computing the rotation angles. Finally, some experiments were carried to verify correctness and feasibility of the computing method. The results showed that the measurement of rotation angle could achieve better performance though the error existed. The maximum value of errors was only 0·4°. The average errors and the standard deviations of errors were also small.     

Domain wall trapping and influence of the asymmetry in magnetic nanoring studied by micromagnetic simulations

We have studied the magnetic switching behavior of permalloy asymmetric rings using micromagnetic simulations. The simulation results have revealed that a domain wall trapping feature is present at the narrow arm of the asymmetric ring. This trapping feature is obtained via precise control of the lateral geometric features, the ring asymmetry and the film thickness. Our results show that the trapped domain walls do not annihilate until the magnetization in the wide arm is reversed under a relatively large magnetic field. Furthermore, the magnetic field strength needed to annihilate the domain wall is found to be decreasing with larger asymmetry ratio.     

The analysis of feature-based measurement error in coordinate metrology

Coordinate measurement systems (CMSs) dominate the dimensional control and diagnostics of various manufacturing processes. However, CMSs have inherent errors caused by the lack of a tracing ability for some of the measured part features. This is important for product inspection and process variation reduction in a number of automated manufacturing systems, such as for example the automotive body assembly process. The lack of a feature tracing ability means that instead of measuring a given feature, the CMS may actually measure the area around the selected feature. In this paper, a principle for the part feature tracing ability and the resultant feature-based measurement error analysis are developed to estimate the aforementioned deficiencies in the CMSs. The impact of feature type and part(s) positional variation on the feature-based measurement error is explored. The proposed approach is applicable to both contact and non-contact CMSs including both mechanical and optical coordinate measuring machines An analysis of the error for different measurement algorithms is presented. We show that the developed feature-based measurement error can have a significant impact on the measurement accuracy and hence on process control and the diagnostic algorithms currently used in manufacturing. A feature-based error map and error compensation approach are also developed and presented. Simulations, experimental results and two industrial case studies illustrate the proposed method.     

A feature fatigue supply chain game with cooperative programs and ad-hoc facilitators

In this paper, we analyze a manufacturer's decisions to invest in technology to increase the number of product features and make its business more economically appealing. High number of features can lead to an overspecified product that entails a feature fatigue effect. More specifically, consumers show a decreasing willingness to purchase a good, due to its complexity. The manufacturer sells through a retailer, which in turn invests in some store facilitators efforts. The manufacturer can decrease the feature fatigue effect by implementing either a cooperative program or an ad-hoc facilitator. Our results reveal that the manufacturer, as the chain leader, makes decisions according to the program's efficiency, while considering the revenue sharing rule. When the share is low, the manufacturer's decision is always based on efficiency; otherwise, high sharing parameters sponsor the implementation of an ad-hoc facilitator. We identify a region of parameters in which firms have divergent preferences, resulting that the feature fatigue mitigation is a short term objective. The results also reveal a trade-off between profits and feature fatigue. An ad-hoc facilitator guarantees large profits, even without fully solving the feature fatigue issue. Instead, a support program increases mitigates the feature fatigue effect, while worsening the firms' profits.     

Enhancing multivariate pattern analysis for magnetoencephalography through relevant sensor selection

While the popularity of multivariate pattern classification is growing rapidly in magnetoencephalography (MEG) data analysis, the analysis pipelines used by the neuroscience community are still missing some fundamental machine-learning techniques and principles that would increase their effectiveness. Here, we show that MEG decoding accuracy improves significantly with the addition of feature selection methods to the analysis pipeline. We compare one unsupervised and two supervised feature reduction methods in the current study. Our results show that supervised feature selection methods like statistical dependency and mutual information improve decoding performance and attain higher session-to-session reliability compared to unsupervised dimensionality reduction methods like principal component analysis. Furthermore, we demonstrate that the selected sensors in the data related to a visual task at each time point are consistent with the pattern reflecting the sweep of information in the ventral visual pathway.     

Comparative study of feature extraction and classification methods for recognition of characters taken from vehicle registration plates

ABSTRACT

General Optical Character Recognition system works on the base of several successive steps such as pre-processing, segmentation, feature extraction, classification and post-processing. Feature extraction plays here a major role. In this article, we present an overview and comparison of various methods and approaches for off-line recognition of machine written Latin characters. We assume that individual characters are already segmented in an image. To recognize characters and translate them to text requires that each character must be described by a feature vector, which is then classified into one of the 36 classes corresponding to the uppercase Latin alphabet letters and numbers.     

Enhancement of optical processes in coupled plasmonic nanocavities [Invited

We present detailed experimental and numerical investigations of resonances in deep nanogroove gratings in metallic substrates. These plasmonic nanocavity gratings feature enhanced fields within the grooves that enable a large enhancement of linear and nonlinear optical processes. This enhancement relies on both localized and propagating surface plasmons on the nanopatterned surface. We show that the efficiency of optical processes such as Raman scattering and four-wave mixing is dramatically enhanced by plasmonic nanocavity gratings.     

Augmenting security of biometric with watermarking

This paper presents a new blind biometric watermarking scheme based on discrete wavelet transform for biometric template protection. We are using fingerprint feature and iris feature as multiple watermarks instead of randomly generated Gaussian noise type watermark. Watermark embedding locations are selected by measuring power level of each coefficient which gives sensibility of human eye to local image perturbations. Though biometric features are embedded sequentially, care is taken that most significant features are embedded to those locations whose power level is maximum which provides more robustness. We embed watermark into mid-frequency sub-band by modifying the amplitude of selected coefficient based on comparison between original and estimated values. Estimated value of coefficient is calculated from average sub-band of the same resolution level of the cover image. A reliable watermark extraction scheme is developed for the extraction of watermark from distorted image. Experimental evaluation demonstrates that the proposed scheme is able to withstand a variety of attacks. We show that the proposed scheme also gives adjustability to the user for selecting P percentage of retained wavelet coefficients.     

基于特征金字塔的多尺度特征融合网络

特征金字塔网络（FPN）是CNN网络对图像信息进行表达输出的一种有效方法,在目标检测网络中广泛应用．然而,FPN没有充分地将浅层的细节信息传递到深层的语义特征,存在特征融合不足的缺陷,因而只能依靠深层语义信息来进行预测,从而忽略了网络低层细节信息,对各种视觉学习的效果造成了一定的影响．针对FPN存在的以上问题,本文提出基于特征金字塔的多尺度特征融合网络模型,在FPN主干网络的基础上,设计了混合特征金字塔和金字塔融合模块,并结合注意力机制,对特征金字塔进行了多尺度的深度融合．本文在PASCAL VOC2012和MS COCO2014数据集上,以Faster R-CNN作为基础检测器进行实验,验证了MFPN对特征融合的有效性．