光强倒数色度空间的彩色人脸光照预处理

为了解决彩色人脸识别中色彩信息易受光照影响的问题,提出一种基于光强倒数色度空间(IICS)的彩色人脸图像预处理方法。本方法首先将图像均匀地分割成子块;将每个图像块变换为IICS空间中的一个二维数据集,并根据数据集的线性分布特性估计图像块的光照颜色;然后对全部图像块的光照估计进行颜色直方图统计,根据直方图对分块估计的结果进行合并;最后,利用估计得到的光照和对角模型将图像光照校正到标准白光下,用于人脸识别。在AR和FERET人脸库上的实验表明,通过引入本光照预处理,有效增强了彩色人脸识别方法对光照变化的鲁棒性,提高了识别精度。     

Application of Gobor wavelet and SLLE in face recognition

In order to improve the recognition rate of face recognition algorithm, a new algorithm of face recognition is proposed based on Gabor wavelet transform and Supervised Locally Linear Embedding (SLLE). Gabor wavelet is introduced as a method to extract Gabor magnitude features by convolving the normalized face image with multi scale and multi orientation Gabor filters. In the feature extraction module, the dimension of Gabor features is reduced by SLLE. A minimum distance classifier is trained for classification. With the test of the ORL and YALE face database, it is found that 3.5 %~37.8% increase in recognition rate can be achieved compared with other algorithms.     

A face recognition method based on modular PCA and SVD

In order to solve the problem that modular PCA method is sensitive to translation, rotation and other geometric transform, a face recognition method based on modular PCA and singular value decomposition (SVD) is proposed. The PCA features of sub image and SVD features are extracted respectively. The distance measure that fuses information of modular PCA and SVD is obtained. Minimum distance classifier is used to face recognition. Experimental results on ORL human face database show that the proposed method can obtain higher recognition rate.     

A new pattern recognition method of gas sensor array detection

BP neural network based gas sensor array detection pattern recognition has some disadvantages, such as slow convergence and local minimum problem. A modified immune neural network model which combines BP algorithm and immune algorithm is proposed to enhance global search capability and improve the performance of the neural network model. Orthogonal test is adopted to design the study samples of neural network. This ensures the accuracy of neural network while reducing the number of samples. The simulation results show that the proposed pattern recognition method solves the cross sensitivity of gas sensor effectively, overcomes the disadvantages of traditional BP neural network and improves the learning speed and detection accuracy.     

Multi class obstacles recognition for intelligent vehicle in urban traffic scenes

For multi class obstacles recognition for intelligent vehicle in urban traffic scenes, an improved Binary Tree Support Vector Machine (BT SVM) based on ensemble learning is presented. Based on the distributing probability and pattern diversity of each obstacle in urban traffic scenes, a compatible tree structure of BT SVM is designed. An approach based on AdaBoost ensemble learning is applied to reduce the transfer error and improve the accuracy and generalization ability of per node classifier. The proposed method can efficiently recognize 6 kinds of normal obstacle patterns in urban traffic scenes.     

Digital modulation mode recognition based on multi-classclassification of support vector machine

To solve the overfitting, underfitting and local minimum existing in neural networks, a digital modulation mode recognition method based on support vector machine (SVM) is proposed. Seven characteristic parameters are extracted from instantaneous amplitude, instantaneous phase, instantaneous frequency, frequency spectrum, and changes in characteristics of the envelope to train support vector machine. Compared with the existing algorithms, using binary tree theory to design multi-class classifier has the features of simple, high-speed, high-precision. The simulation results indicate that the scheme can achieve 97% recognition accuracy when the signal to noise ratio (SNR) is above 15 dB with the AWGN channel.     

基于颜色和纹理特征的花生仁外观品质检测研究

为了提高对花生仁外观缺陷的在线分类准确率及效率。通过对采集完好、破损、霉变的花生仁RGB图像进行均值位移法、灰度处理以及阈值分割等预处理,研究提取了花生仁HSV颜色空间下的H、S、V各分量的一阶矩和二阶矩共6个颜色特征值,再基于灰度共生矩阵法提取能量、熵、对比度、逆差分矩共4个纹理特征值,构建颜色和纹理结合的特征向量,最后分别采用BP神经网络和SVM分类器对花生仁进行分类识别。结果表明：在花生仁的整体识别准确率上,BP神经网络为96.67%,SVM分类器为97.22%,后者优于前者,在识别时间上BP和SVM分别为2.5 s和1.1 s,识别效率上也是SVM更好,综合识别准确率和效率两方面考虑,优先选择SVM分类器模型来对花生仁进行分类识别。     

Single sample face recognition based on B-spline and image gradient

A single sample face recognition algorithm based on B-spline and image gradient is proposed. Image gradient method for face recognition has advantage of illumination invariant. But the recognition rate will be greatly decreased when the image contains noise which will seriously influence gradient information. Traditional methods to reduce noise smooth image at the same time and image gradient reorganization rate will be reduced. As the B-spline filter has the feature which can adjust the order, B-spline filter with different orders can be selected according to the image noise value to minimize noise while preserve image gradient information. Experiments prove that using B-spline and image gradient algorithm can achieve a better recognition rate than traditional filtering method on single sample face recognition problem.     

基于Adaboost算法的输电线路舞动预警方法

输电线路舞动是目前尚未被全面准确认识的世界性难题,已严重威胁输电系统的安全稳定运行。文章分析影响舞动的外界气象环境因素,并在此基础上提出一种基于Adaboost集成学习算法的输电线舞动预警方法。采用基于Gini指标的决策桩作为弱分类器,通过对多个弱分类器的训练及加权求和,输出舞动预测结果及其置信度,可为电网运维人员提供决策支撑。最后,使用历史数据进行验证性实验,结果证明了所提方法的有效性。     

基于迁移学习的园艺作物叶部病害识别及应用

为给农户提供质优价廉的园艺作物叶部病害识别服务,提出基于迁移学习的模型训练及基于Flask的Web部署方法。对PlantVillage数据集进行预处理,分别使用ResNet18、ResNet50和ResNet152 3种模型进行迁移学习训练,得到3种识别模型。利用Flask将模型部署到服务器上。3种识别模型对苹果等14类园艺作物26种叶部病害的平均识别准确率分别是95.61%、96.63%和97.33%,识别单张图像的时间分别是10.9、17.9、33.7 ms。综合考虑模型特点和用户期望,设计快速、标准和准确3种识别模式,实现深度模型在服务器中稳定运行,具有一定的实用价值。     

A fuzzy fusion approach to gabor transform optimal channels for facial expression recognition

According to its characteristics of multi-direction and multi-scale, Gabor wavelet is divided into 13 channels. Due to different contributions of different channels to the recognition rate, a fuzzy integral fusion approach for facial expression recognition is proposed based on optimal channels. Firstly, three optimal channels are selected according to three proposed principles. And then Gabor features of facial expressional images through those optimal channels are extracted with the dimensions reduction. Finally, each optimal channel is used as a classifier and three classifiers are fused based on multi-classifier combination with fuzzy integral. With the results on JAFFE database, it is found that the recognition rate of the proposed algorithm is 94.41%, which verifies the effectiveness of the proposed algorithm.     

Wrinkle modeling based on shape control functions and adaptive subdivision

Realistic wrinkles are an extremely important contribution for enhancing the realism of three dimensional (3D) face model.For different shapes of the wrinkle in three views, this paper proposes three shape control functions for each view, namely, cross-sectional shape control function (CSCF), depth attenuation function (DAF) and width attenuation function (WAF).CSCF is used to realistically and naturally express the furrows and bulges of wrinkles. Depth attenuation function and width attenuation function are applied to control the attenuation variations of the depth and width. Users can create various wrinkles just defining several intuitive parameters. Meanwhile, an image matching technique based on feature is employed to obtain three dimensional wrinkle lines. And a simple adaptive subdivision approach is proposed in the predefined influence region to adjust the resolution around the wrinkle mesh for better results and enhance the efficiency of wrinkle modeling simultaneously. Comparing experiment results with the corresponding image demonstrates that our method can generate realistic and natural wrinkles and can greatly improve the similarity of face model.     

A Meta-synthetic Approach for Lower-constrained Handwritten Character Recognition

A meta-synthetic network approach for conbination of multiple classifier is proposed aimed at solving the problems of character recognition. It composes of two B-Wavelet neural network classifiers and a Morlet-B-Wavelet neural network classifier.It is shown that the method is effective used in handwritten capital English character recongnizing.     

Human Face Location Based on Motion Information in a Complex Background

This paper proposes an improvement for the traditional Optical Algorithm, and presents a new way to image segmentation in a complex background. In addition, combined with the neural network, the system can locate the possible human faces successfully by means of two-step location model. In our system, the searching and locating of the human face is the most important stage. According to this, the authors adopt the two-step way to run, firstly they take up the segmentation of the candidate human face areas and then the accurate face locating based on the neural network is used. This algorithm is fast and robust. Experimental results with real scene images are given out there, and all these prove that two-step method gains many advantages in the course of human face location with motion information, such as real-time, robustness and practicality. In addition, the proposed system is also the fundamental and important part of the perfect human face recognition system.     

基于改进SVM分类器的动作识别方法

利用智能手机加速度传感器信号,提出一种改进的动作识别方法以降低传统动作识别方法的复杂程度,提高识别率。在特征提取时用盲选法,即用PCA(principal component analysis)进行特征值的降维和去除多维间的干扰,而所选特征没有对应的物理意义;并在分类识别中将遗传算法应用到SVM(support vector machine)分类器参数优化中。通过实验表明,该方法能够对日常的走路、站立、跑及上下楼等动作进行准确的识别。     

Frequency tracking method and simulation for coriolismass flowmter based on a new adaptive notch filter

In order to improve the long-playing frequency tracking ability in the signal processing for coriolis mass flowmeter (CMF) , a new IIR adaptive notch filter (ANF) is developed via steiglitz-mcBride method (SMM) and applied to filter the sensor output signal whose frequency, amplitude and phase are time-varying following the random walk model. The proposed method can detect the signal frequency fleetly and track the frequency variations continuously. The tracking performances of the proposed method and the adaptive lattice notch filter (LANF) method are investigated with computer simulations. Simulations demonstrate the superiority of the proposed method and show it has advantages of fast convergence rate, insensitivity to initial phase variations, and higher long-playing tracking stability and accuracy than LANF method.     

Illumination compensation of color video for driver fatigue monitor

A robust method to compensation illumination of driver video color image to enhance image quality has been proposed. Firstly, the original image is decomposed into high-frequency image and low-frequency image based on spatial image decomposition. The illumination information is completely contained in the low-frequency image. Next, low-frequency image is compensated by using Retinex approach which can brighten the darker area. Then, the compensated low-frequency image and high-frequency image are synthesized. Finally, new image is obtained by illumination balance of synthesized image. Results show that the algorithm has better results compared with the traditional algorithm.     

联合局部和全局稀疏表示的磁共振图像重建方法

针对在压缩传感中独立使用全局或局部稀疏字典所分别导致的图像细节或整体图像结构信息的丢失,提出了一种联合利用局部和全局稀疏约束来捕捉磁共振图像细节和整体结构信息的磁共振图像重建算法。该算法首先从特定的磁共振图像中训练出稀疏字典,然后利用该字典进行局部稀疏编码。其次,利用预定义的全局字典来加强磁共振图像的全局稀疏性。最后,在局部和全局稀疏的共同约束下,利用非线性共轭梯度算法来对重建模型进行求解。整个重建过程可以重复迭代以逐步改善重建质量。实验结果表明：当下采样因子达到10时,相比于字典学习算法（dictionary learning MRI,DLMRI）,提出的算法在重建质量上可以提高1-6 dB。     

云制造资源语义描述和服务匹配策略

针对云制造环境下资源语义描述和服务匹配问题,在充分考虑用户个性化需求与制造资源异构性的基础上,分析了云制造资源描述和服务匹配的难点,利用规范化描述的特点,将制造资源描述问题转化为Web语义描述问题,提出一种通用的制造资源语义描述框架。结合4种相似度算法,设计了阀值和权重,提出了基于过滤器的服务匹配策略。实验结果证明了该策略和算法的有效性和可行性。应用这4种相似度算法能够有效缩小匹配的规模,提高了服务的查找效率与准确度。     

Fingerprint Identification Using Point Pattern Matching

Fingerprint is those lines distributing on the finger surface. It's unique and stable. It has been hundreds years to use fingerprint to identity. AFIS (Auto Fingerprint Identification System) has been widely used. Besides traditional criminal, AFIS has been used to validate user in OS and web access, and embedded AFIS also has been used in ATM, credit card and door-lock. Fingerprint matching is one of the most important problems in AFIS. In general, the minutiae such as ridge endings and ridge bifurcation is to be used to represent a fingerprint and the fingerprint matching through minutiae matching. Based on this observation and by representing the minutiae as a point pattern, an automatic fingerprint verification problem may be reduced to a point pattern matching (minutia matching) problem. Point pattern matching is a famous problem in the field of pattern recognition. For a good point pattern matching approach, it is able to perform the geometrical invariant quantities (translation, rotation, and scaling) efficiently. Various algorithms have been proposed for point pattern matching. For example, the relaxation approach handles translational difference only and the complexity of triangles approach is very high. In this paper, an algorithm for fingerprint identification using point pattern matching based on cluster approach is proposed, which effectively solves the problems of optimal matching between two fingerprint minutiae images under geometrical transformation and minutiae quantity change. Process which bases on the matching of vector pairs is developed to determine the registration parameters. The experimental results show that the proposed matching algorithm is fast and has high accuracy.