Near-infrared finger vein patterns for personal identification

We have demonstrated a personal identification system that is based on near-infrared finger vein patterns. Finger vein patterns of 678 volunteers are acquired by transmitting near-infrared light through a finger and capturing the image with a CCD camera. These vein patterns are enhanced by a background-reduction filter. The similarity between two patterns is then quantified by use of the normalized maximum of the cross correlation of the two images after correction of the tilt angle. The enhanced finger vein pattern enabled 678 persons to be successfully identified.     

Profile offsetting for feature extraction and feature tool mapping in sheet metal

It is widely accepted that automatic process planning is the key solution to bridging the gap between design and manufacturing. In order to realize this, it is important that the CAD system should not only allow the designer to work in terms of features which relate to manufacturing operations but also be able to extract features and recognize them. Extraction of features internal to the contour of the sheet metal involves identifying groups that are connected to members within itself but not connected to any other group. However the same rationale cannot be adopted for features on the contour due to their lack of connectivity. This paper presents a new approach towards extracting contour features by generating the offset of the contour. It is important that once the features have been extracted and recognized, appropriate tools for creating the feature be identified. This paper proposes a tool mapping algorithm based on the medial axis transformation (MAT). Since stamping is assumed to be the method of manufacturing, feature-tool mapping involves identifying the punch which can stamp the feature.     

A novel feature extraction algorithm for asymmetric classification

A linear feature extraction technique for asymmetric distributions is introduced, the asymmetric class projection (ACP). By asymmetric classification is understood discrimination among distributions with different covariance matrices. Two distributions with unequal covariance matrices do not, in general, have a symmetry plane, a fact that makes the analysis more difficult compared to the symmetric case. The ACP is similar to linear discriminant analysis (LDA) in the respect that both aim at extracting discriminating features (linear combinations or projections) from many variables. However, the drawback of the well-known LDA is the assumption of symmetric classes with separated centroids. The ACP, in contrast, works on (two) possibly concentric distributions with unequal covariance matrices. The ACP is tested on data from an array of semiconductor gas sensors with the purpose of distinguish bad grain from good.     

深度神经网络的语音深度特征提取方法

为了提升连续语音识别系统性能,将深度自编码器神经网络应用于语音信号特征提取。通过堆叠稀疏自编码器组成深度自编码器(Deep Auto-Encoding,DAE),经过预训练和微调两个步骤提取语音信号的本质特征,使用与上下文相关的三音素模型,以音素错误率大小为系统性能的评判标准。仿真结果表明相对于传统梅尔频率倒谱系数(Mel-Frequency Cepstral Coefficient,MFCC)特征以及优化后的MFCC特征,基于深度自编码器提取的深度特征更具优越性。     

基于Gabor小波变换的汉字识别方法

基于Gabor滤波器时频局域性和方向选择性对应在汉字图像上即是对笔划宽度和方向的选择这一认识,考虑到笔划宽度多峰值、笔划是方向的特点,利用二维Gabor小波的多分辨率特性,对汉字图像进行了多尺度多方向分析,提取多个子平面的滤波器输出系数作为统计特征,实现了字符图像的高性能识别,为开展低质量、低分辨率的汉字字符图像识别研究提供了新方法。实验结果表明,用这种方法识别低分辨率的字符图像,效果优于其他方法。     

时频图像特征用于声场景分类

为解决根据音频流识别声场景的问题,对音频信号进行恒Q变换,得到其时频表达图像,然后进行滤波平滑等处理,随之提取能够表述信号谱能量变化方向信息的梯度直方图特征,以及能够捕捉信号谱纹理信息的局部二值模式特征,输入具有线性核函数的支持向量机分类器,对不同声场景数据进行分类实验。结果表明,相对于传统的时频域特征和梅尔频率倒谱系数特征,所提出的特征基本能够捕捉到给定声场景具有区分度的信息,所得分类率更高,且两者的互补作用使得联合特征分类效果达到最优,该方法为声信号特征提取贡献了一种新思路。     

Optimal feature extraction for segmentation of Diesel spray images

A one-dimensional simplification, based on optimal feature extraction, of the algorithm based on the likelihood-ratio test method (LRT) for segmentation in colored Diesel spray images is presented. If the pixel values of the Diesel spray and the combustion images are represented in RGB space, in most cases they are distributed in an area with a given so-called privileged direction. It is demonstrated that this direction permits optimal feature extraction for one-dimensional segmentation in the Diesel spray images, and some of its advantages compared with more-conventional one-dimensional simplification methods, including considerably reduced computational cost while accuracy is maintained within more than reasonable limits, are presented. The method has been successfully applied to images of Diesel sprays injected at room temperature as well as to images of sprays with evaporation and combustion. It has proved to be valid for several cameras and experimental arrangements.     

基于时频图像特征提取的状态识别方法研究与应用

介绍一种基于时频图像进行设备状态识别的新方法。论述了对振动信号应用Hilbert谱构建二维时频图像,研究时频图像三维重心与信息熵的特征提取方法构造特征向量,采用支持向量机进行设备的状态识别。并以滚动轴承不同状态的识别为例,验证方法的有效性。为提高滚动轴承振动信号特征提取的可靠性,采用多循环平均方法减小循环波动性影响。研究表明此方法能够提高设备故障诊断与状态识别的准确性。     

民机驾驶舱舱音小波包分析及特征提取

摘 要：针对民航事故调查中的舱音背景声分析,利用小波包分析方法,对不同种类开关手柄声进行了三层分解,提取了不同开关手柄声的各频带归一化能量特征用于对开关手柄进行识别。最后,以B737客机的常用开关手柄的声音为研究对象,采集了大量舱音信号,利用该方法,对舱音信号进行了分析识别,并对结果进行统计验证。
     

一种模糊红外目标边缘和不变特征提取方法

针对红外图像中模糊目标难以分割和识别的情况,提出以小波多尺度滤波分解后的逼近图像作为自适应域值进行图像二值化的方法,由此分割出模糊图像中的目标,提取目标的轮廓边缘.然后,提出一种新的矩计算方法提取目标的不变性特征.实验结果表明,该方法具有很强的抗噪和抗扰性能,能有效提取复杂背景中模糊目标的平移、缩放和旋转不变量,极大提高了运动模糊目标识别的可靠性.     

Theory of spatiochromatic image encoding and feature extraction

We consider how to interpret, filter, and cross-correlate complex-value color (hue and saturation) images by using a single discrete Fourier transform: the spatiochromatic discrete Fourier transform. The model defines new types of spatiochromatic oriented sinusoidal gratings, termed rainbow gratings, which encode the variation of color over space. We demonstrate how color-opponent detectors observed within the vertebrate visual system can be easily defined by linear filters within this representation. This model also allows us to filter and detect both spatial and chromatic patterns in images by using a single cross-correlation procedure. In doing so, we explore a new form of the Cauchy-Schwartz inequality applied to complex-valued scalar products. Results demonstrate the power of this form of spatiochromatic matched filtering in detecting signals embedded in such a significant amount of noise that they are not visible to the unaided human eye.     

基于特征端元提取的像元分解方法

通过对端元概念的引申,提出了更具一般性的特征端元的概念,从而将端元方法的适用范围从高光谱数据拓展到了多光谱数据和全色遥感数据.同时改进了端元提取算法,并提出了基于特征端元提取的像元分解方法.实验结果表明,基于特征端元提取的混合像元分解方法可以得到较好的像元分解结果,并且该方法与传统的端元概念及混合像元分解结果具有良好的可比性.     

噪声协助的EMD-1.5维谱信号抗混分解与特征提取

针对大型动力装备核心部件微弱故障特征信息提取问题,提出了一种噪声协助的EMD-1.5维谱故障诊断方法。经验模式分解(EMD)方法中,信号极值点间隔特性影响模式混淆现象的出现,针对此状况提出信号极值点间隔特性评价方法,分析高斯白噪声有助于信号抗混分解原理,通过对原始信号加入高斯白噪声得到噪声协助的EMD方法,提高信号抗混分解能力。将1.5维谱与噪声协助的EMD方法结合,得到一种新的故障特征提取方法,该方法具有对信号进行有效抗混分解、提取非线性耦合等特性,并以此来提取故障的微弱特征信息。通过仿真研究与电力机车滚动轴承的故障诊断工程实例,验证了该方法的有效性。     

基于函数变换的水下图像目标分割和特征提取

针对海水信道的特殊性以及成像环境的复杂性,对视觉系统的图像处理和特征提取带来的影响,提出了一种基于模糊变换的图像分割和基于函数变换的特征提取方法,以克服水下不确定因素给目标识别带来的困难,并对其进行了仿真试验.试验结果表明,此方法在对深海烟囱图像的分割和特征提取上能够取得好的效果,可有效地克服水下图像灰度分布不均匀和环境不确定因素的干扰,实现了难于分类判别的深海热液喷口目标的区分.     

非线性小波变换在故障特征提取中的应用

经典小波变换在不同尺度采用一种小波基 ,不能很好地匹配信号的局部特征 ,因而造成降噪信号丢失了原始信号中部分的有用信息。为了克服上述缺陷 ,提出了一种基于第二代小波变换的非线性小波变换振动信号预处理方法 .应用第二代小波变换的预测器和更新器相互独立的特点 ,根据预测方差最小的选取原则 ,确定每个变换样本的最佳预测器 ,使预测器能够适应信号的局部特征。模拟数据和振动信号的分析表明 ,该方法克服了传统小波降噪方法局部信息丢失的缺陷 ,不仅可以有效地去除信号中的噪声 ,而且能够保留信号的局部特征。作为一种预处理方法 ,在某发电厂的故障诊断中有效地从振动信号中提取了故障特征。     

Statistical feature extraction based iris recognition system

Iris recognition systems have been proposed by numerous researchers using different feature extraction techniques for accurate and reliable biometric authentication. In this paper, a statistical feature extraction technique based on correlation between adjacent pixels has been proposed and implemented. Hamming distance based metric has been used for matching. Performance of the proposed iris recognition system (IRS) has been measured by recording false acceptance rate (FAR) and false rejection rate (FRR) at different thresholds in the distance metric. System performance has been evaluated by computing statistical features along two directions, namely, radial direction of circular iris region and angular direction extending from pupil to sclera. Experiments have also been conducted to study the effect of number of statistical parameters on FAR and FRR. Results obtained from the experiments based on different set of statistical features of iris images show that there is a significant improvement in equal error rate (EER) when number of statistical parameters for feature extraction is increased from three to six. Further, it has also been found that increasing radial/angular resolution, with normalization in place, improves EER for proposed iris recognition system.     

基于IVMD-RPE的船舶噪声信号去噪及特征提取方法

针对由于复杂环境条件下的船舶噪声信号识别度低的问题,提出了一种改进的变分模式分解(Improved Variational Mode Decomposition,IVMD)、反向排列熵(Reverse Permutation Entropy,RPE)、加权排列熵(Weighted Permutation Entropy,WPE)和能量比相结合的船舶噪声信号识别分类方法。该方法利用IVMD将信号分解为若干本征模态函数(Intrinsic Mode Function,IMF);再利用RPE对IMF进行筛选,得到敏感IMF,实现去噪过程;最后计算IMF的WPE并与各个IMF的能量比构建特征向量,建立长短期记忆网络(Long Short-Term Memory,LSTM)进行识别分类。实验结果表明,与经验模态分解相比,文中提出的基于IVMD-RPE的船舶噪声信号去噪及特征提取方法能有效减少环境噪声的影响,提高信噪比,对船舶噪声目标信号识别分类的准确率更高。     

A staged approach for feature extraction from sheet metal part models

This paper proposes a staged approach of feature identification and extraction for sheet metal parts from a solid model. The identification and extraction process is divided into several stages. Each stage is automatically processed and the results after each stage can be interactively corrected and modified for the next stage. With this approach, the features of sheet metal parts can be automatically extracted from their well-defined solid models. At the same time, for ill-defined or inaccurate solid models of sheet metal parts, their features can be extracted through certain user interaction at a suitable stage. An attribute is also introduced in the paper to measure the relationship between sheet surfaces and other surfaces, and simplify the identification process. The approach has been implemented in the AutoCAD® environment and used by an industrial collaborator for the design of progressive dies.     

Motion feature extraction for stepped frequency radar based on Hough transform

The high resolution range profile can be obtained through coherent pulse compression processing in every coherent processing interval (CPI) for stepped frequency radars (SFR). The radar observation time can be divided into multiple CPIs. By utilising the temporal information inherent to a sequence of these range profiles, a time-range image containing the target motion information can be synthesised. Through analyses of the point scatterer targets with the radial uniform motion, radial uniformly accelerated motion, rotation and vibration, this study indicates that there is a certain correlation between the target motion features and the time-range image features in the limited observation time. Based on this result, we propose a motion feature extraction method for SFR without any assistant information using the conventional and well-known feature extraction tool, Hough transform. The authors give detailed analyses and explanation of this method, and verify them by outfield experiments, demonstrate the motion feature extraction results using real radar data. The proposed method enables the motion feature extraction through space non-coherent integration, which fully utilises the radar data information under low signal-to-noise ratio conditions.