一种基于词袋模型的图像优化分类方法

该文应用词袋模型对图像进行分类,并针对传统词袋模型存在的不足进行改进,提出了一种基于兴趣区域(Region Of Interest,ROI)提取以及金字塔匹配原理的优化方法。首先对训练图像进行ROI提取,对得到的ROI区域进行密集尺度不变特征变换(Scale-Invariant Feature Transform,SIFT)特征的抽取和描述并生成视觉词典,由此产生的视觉词典更能精确的描述图像的特征,且能够抵抗多变的位置信息及背景信息的影响。其次应用金字塔匹配原理对图像进行基于视觉词典的直方图表示,代入支持向量机(Support Vector Machine,SVM)分类器进行分类。通过对Caltech 101和Caltech 256两个数据库进行实验,结果表明该方法较传统方法提高了分类的正确率,且能够在训练图像较少的情况下达到良好的分类效果。最后通过与现有同类方法的比较验证了该方法的优越性。

一种改进的k-means聚类视觉词典构造方法

传统词袋(bag of words,BoW)模型在构造视觉词典时一般采用k-means聚类方法实现,但k-means聚类方法的性能在很大程度上依赖于初始点的选择,从而导致生成的视觉词典鲁棒性较差,此外,每次迭代都要计算数据点与中心点的距离,计算复杂度高。针对上述问题,提出了一种改进的k-means聚类视觉词典构造方法,该方法首先对初始值的选取进行了优化,克服了随机选取初始值对聚类性能的影响,其次基于三角形不等式对计算进行了简化,使生成的视觉词典更加稳定,计算复杂度更低,最后引入权值分布对图像进行基于视觉词典的表示,并将基于改进的视觉词典的词袋模型应用于图像分类,提高了分类性能。通过在Caltech 101和Caltech 256两个数据库进行实验,验证了本文方法的有效性,并分析了词典库大小对分类性能的影响。从实验结果可以看出,采用本文方法所得到的分类正确率提高了5%～8%。

Methodical Analysis of Western-Caucasian and East-Asian Basic Facial Expressions of Emotions Based on Specific Facial Regions

Facial expressions are the straight link for showing human emotions. Psychologists have established the universality of six prototypic basic facial expressions of emotions which they believe are consistent among cultures and races. However, some recent cross-cultural studies have questioned and to some degree refuted this cultural universality. Therefore, in order to contribute to the theory of cultural specificity of basic expressions, from a composite viewpoint of psychology and HCI (Human Computer Interaction), this paper presents a methodical analysis of Western-Caucasian and East-Asian prototypic expressions focused on four facial regions: forehead, eyes-eyebrows, mouth and nose. Our analysis is based on facial expression recognition and visual analysis of facial expression images of two datasets composed by four standard databases CK+, JAFFE, TFEID and JACFEE. A hybrid feature extraction method based on Fourier coefficients is proposed for the recognition analysis. In addition, we present a cross-cultural human study applied to 40 subjects as a baseline, as well as one comparison of facial expression recognition performance between the previous cross-cultural tests from the literature. With this work, it is possible to clarify the prior considerations for working with multicultural facial expression recognition and contribute to identifying the specific facial expression differences between Western-Caucasian and East-Asian basic expressions of emotions.

Facial Expression Recognition Based on Local Fourier Coefficients and Facial Fourier Descriptors

The recent boom of mass media communication (such as social media and mobiles) has boosted more applications of automatic facial expression recognition (FER). Thus, human facial expressions have to be encoded and recognized through digital devices. However, this process has to be done under recurrent problems of image illumination changes and partial occlusions. Therefore, in this paper, we propose a fully automated FER system based on Local Fourier Coefficients and Facial Fourier Descriptors. The combined power of appearance and geometric features is used for describing the specific facial regions of eyes-eyebrows, nose and mouth. All based on the attributes of the Fourier Transform and Support Vector Machines. Hence, our proposal overcomes FER problems such as illumination changes, partial occlusion, image rotation, redundancy and dimensionality reduction. Several tests were performed in order to demonstrate the efficiency of our proposal, which were evaluated using three standard databases: CK+, MUG and TFEID. In addition, evaluation results showed that the average recognition rate of each database reaches higher performance than most of the state-of-the-art techniques surveyed in this paper.

Simultaneous Children Recognition and Tracking for Childcare Assisting System by Using Kinect Sensors

Developing a childcare assisting system is highly necessary due to the lack of nursery teachers, and it will make an important progress on effective utilization of nursery teacher resources. In this paper, we proposed simultaneous children recognition and tracking system by using Kinect sensors for the childcare assisting system to provide information for the nursery teachers. Each of the children is recognized by integrating his/her personal information of color, face and motion. The tracking problem is modeled as finding the MAP solution of a posterior probability, and is solved by using Markov Chain Monte Carlo (MCMC) particle filter. Our system can recognize and robustly track each child during class activities. Trajectories, motion ranges and relative distances information can be provided for the nursery teachers to assist their childcare work. The effectiveness of our system is proved through continuous monitoring of the children in a nursery school.