Graph regularized multiset canonical correlations with applications to joint feature extraction

Multiset canonical correlation analysis (MCCA) is a powerful technique for analyzing linear correlations among multiple representation data. However, it usually fails to discover the intrinsic geometrical and discriminating structure of multiple data spaces in real-world applications. In this paper, we thus propose a novel algorithm, called graph regularized multiset canonical correlations (GrMCCs), which explicitly considers both discriminative and intrinsic geometrical structure in multiple representation data. GrMCC not only maximizes between-set cumulative correlations, but also minimizes local intraclass scatter and simultaneously maximizes local interclass separability by using the nearest neighbor graphs on within-set data. Thus, it can leverage the power of both MCCA and discriminative graph Laplacian regularization. Extensive experimental results on the AR, CMU PIE, Yale-B, AT&T, and ETH-80 datasets show that GrMCC has more discriminating power and can provide encouraging recognition results in contrast with the state-of-the-art algorithms.     

Fractional-order embedding canonical correlation analysis and its applications to multi-view dimensionality reduction and recognition

Due to the noise disturbance and limited number of training samples, within-set and between-set sample covariance matrices in canonical correlation analysis (CCA) usually deviate from the true ones. In this paper, we re-estimate within-set and between-set covariance matrices to reduce the negative effect of this deviation. Specifically, we use the idea of fractional order to respectively correct the eigenvalues and singular values in the corresponding sample covariance matrices, and then construct fractional-order within-set and between-set scatter matrices which can obviously alleviate the problem of the deviation. On this basis, a new approach is proposed to reduce the dimensionality of multi-view data for classification tasks, called fractional-order embedding canonical correlation analysis (FECCA). The proposed method is evaluated on various handwritten numeral, face and object recognition problems. Extensive experimental results on the CENPARMI, UCI, AT&T, AR, and COIL-20 databases show that FECCA is very effective and obviously outperforms the existing joint dimensionality reduction or feature extraction methods in terms of classification accuracy. Moreover, its improvements for recognition rates are statistically significant on most cases below the significance level 0.05.     

加权型曲率保持 PDE 图像滤波方法

提出了一种加权型曲率保持偏微分方程(Partial differential equation, PDE)滤波方法.传统曲率保持PDE 滤波方法未考虑各积分曲线可能经历不同的图像结构,如此影响了其对图像边缘的保持能力.在此基础上, 利用局部图像方向信息为不同积分曲线设计了相应的权重,得到了一种张量驱动的加权型曲率保持PDE滤波方法. 实验结果表明该方法在滤波的同时能较好地保持图像中边缘与曲率结构,且对图像具有一定增强能力.     

Learning multi-kernel multi-view canonical correlations for image recognition

Computational Visual Media - canonical correlations (M2CCs) framework for subspace learning. In the proposed framework, the input data of each original view are mapped into multiple higher...     

基于 MTF 和变分的全色与多光谱图像融合模型

Pan-sharpening将高分辨率图像全色(Panchromatic, Pan)波段的空间细节注入多光谱(Multispectral, MS)波段, 以生成同时具有高光谱和高空间分辨率的多光谱图像. 为改善融合效果, 需要考虑多光谱和全色波段的调制传输函数(Modulation transfer function, MTF). 本文提出了一个新的基于MTF和变分的Pan-sharpening模型. 该模型的能量泛函包括两项, 第1项为细节注入项, 基于高通滤波器从Pan波段中提取细节信息并注入融合图像;第2项为光谱保真项, 基于MTF设计多孔小波的低通滤波器以保持MS波段的多光谱信息. 在QuickBird、IKONOS和GeoEye数据集上的融合结果表明, 该模型可以生成同时具有高空间和高光谱质量的融合图像, 融合效果优于AWLP、IHS_BT、HPM-CC-PSF、NAWL、快速变分等算法.     

Feature Extraction Using Fractional-Order Embedding Direct Linear Discriminant Analysis

In this paper, a novel LDA-based dimensionality reduction method called fractional-order embedding direct LDA (FEDLDA) is proposed. More specifically, we redefine the fractional-order between-class and within-class scatter matrices which can significantly reduce the deviation of sample covariance matrices caused by the noise disturbance and limited number of training samples; then the novel feature extraction criterion based on the direct LDA (DLDA) and the idea of fractional-order embedding is applied. Experiments on AT&T, Yale and AR face image databases are performed to test and evaluate the effectiveness of the proposed algorithms. Extensive experimental results show that FEDLDA outperforms DLDA and other closely related methods in terms of classification accuracy and efficiency.     

结合EFD与NCut的彩色图像分割方法

为了克服边缘流引导的各向异性扩散(EFD)方法过分割和归一化分割(NCut)方法计算复杂度高的缺点,提出结合EFD和NCut的彩色图像分割方法。首先利用EFD对图像进行预分割,然后将分割区域作为节点构建带权无向图G,用NCut对图进行全局最优化分类,并进行相应后处理,得到最终结果。由于图G是基于过分割区域而非像素点的,所以算法效率得到较大提高。另外,EFD方法可有效利用图像的局部信息,NCut方法则考虑到图像的全局特征,因此文中方法综合两者的优点。实验结果表明,文中方法能够取得较好的分割效果。     

稀疏保持典型相关分析及在特征融合中的应用

稀疏保持投影(Sparsity preserving projections, SPP)由于保持了数据间的稀疏重构性, 因而获取的投影向量满足旋转、尺度和平移的不变性, 并能够在无标签的情况下提取样本的自然鉴别信息, 在人脸识别领域取得了较为成功的应用. 本文在典型相关分析(Canonical correlation analysis, CCA)的基础上引入稀疏保持项, 提出一种稀疏保持典型相关分析(Sparsity preserving canonical correlation analysis, SPCCA). 该方法不仅实现了两组特征集鉴别信息的有效融合, 同时对提取特征间的稀疏重构性加以约束, 增强了特征的表示和鉴别能力. 在多特征手写体字符集与人脸数据集上的实验结果表明, SPCCA比CCA具有更优的识别性能.     

典型相关分析的理论及其在特征融合中的应用

利用典型相关分析的思想，提出了一种基于特征级融合的组合特征抽取新方法．首先，探讨了将典型分析用于模式识别的理论构架，给出了其合理的描述．即先抽取同一模式的两组特征矢量，建立描述两组特征矢量之间相关性的判据准则函数，然后依此准则求取两组典型投影矢量集，通过给定的特征融合策略抽取组合的典型相关特征并用于分类．其次，解决了当两组特征矢量构成的总体协方差矩阵奇异时，典型投影矢量集的求解问题，使之适合于高维小样本的情形，推广了典型相关分析的适用范围．最后，从理论上进一步剖析了该方法之所以能有效地用于识别的内在本质．该方法巧妙地将两组特征矢量之间的相关性特征作为有效判别信息，既达到了信息融合之目的，又消除了特征之间的信息冗余，为两组特征融合用于分类识别提出了新的思路．在肯考迪亚大学CENPARMI手写体阿拉伯数字数据库和FERET人脸图像数据库上的实验结果证实了该方法的有效性和稳定性，而且识别结果优于已有的特征融合方法及基于单一特征进行识别的方法．