基于Bag-of-words和Hash编码的近似重复图像检测算法

针对近似重复图像检测的传统算法存在检测效率和准确率不够高的缺点,提出了基于Bag-of-words和哈希编码的近似重复图像检测算法。该算法首先利用Bag-of-words把一幅图像表示成一个500维的特征向量;然后,利用主成分分析(PCA)和尺度不变特征转换(SIFT)进行特征降维,并利用Hash编码技术对特征进行编码;最后,利用动态距离度量技术实现近似重复图像的检测。实验结果表明,利用该算法进行近似重复图像检测是完全可行的,在准确度和查全率之间做到了较好的平衡,查准率可达90%～95%,查全率可达70%～80%。

Sequential Bag-of-Words model for human action classification

Recently, approaches utilizing spatial-temporal features to form Bag-of-Words （BoWs） models have achieved great success due to their simplicity and effectiveness. But they still have difficulties when distinguishing between actions with high inter-ambiguity. The main reason is that they describe actions by orderless bag of features, and ignore the spatial and temporal structure information of visual words. In order to improve classification performance, we present a novel approach called sequential Bag-of-Words. It captures temporal sequential structure by segmenting the entire action into sub-actions. Meanwhile, we pay more attention to the distinguishing parts of an action by classifying sub- actions separately, which is then employed to vote for the final result. Extensive experiments are conducted on challenging datasets and real scenes to evaluate our method. Concretely, we compare our results to some state-of-the-art classification approaches and confirm the advantages of our approach to distinguish similar actions. Results show that our approach is robust and outperforms most existing BoWs based classification approaches, especially on complex datasets with interactive activities, cluttered backgrounds and inter-class action ambiguities.

An improved Bag-of-Words framework for remote sensing image retrieval in large-scale image databases

Due to advances in satellite and sensor technology,the number and size of Remote Sensing(RS)images continue to grow at a rapid pace.The continuous stream of sensor data from satellites poses major challenges for the retrieval of relevant information from those satellite datastreams.The Bag-of-Words(BoW)framework is a leading image search approach and has been successfully applied in a broad range of computer vision problems and hence has received much attention from the RS community.However,the recognition performance of a typical BoW framework becomes very poor when the framework is applied to application scenarios where the appearance and texture of images are very similar.In this paper,we propose a simple method to improve recognition performance of a typical BoW framework by representing images with local features extracted from base images.In addition,we propose a similarity measure for RS images by counting the number of same words assigned to images.We compare the performance of these methods with a typical BoW framework.Our experiments show that the proposed method has better recognition performance than that of the BoW and requires less storage space for saving local invariant features.

Improved Bag-of-Words Model for Person Re-identification

Person re-identification(person re-id) aims to match observations on pedestrians from different cameras.It is a challenging task in real word surveillance systems and draws extensive attention from the community.Most existing methods are based on supervised learning which requires a large number of labeled data. In this paper, we develop a robust unsupervised learning approach for person re-id. We propose an improved Bag-of-Words(i Bo W) model to describe and match pedestrians under different camera views. The proposed descriptor does not require any re-id labels, and is robust against pedestrian variations. Experiments show the proposed i Bo W descriptor outperforms other unsupervised methods. By combination with efficient metric learning algorithms, we obtained competitive accuracy compared to existing state-of-the-art methods on person re-identification benchmarks, including VIPe R, PRID450 S, and Market1501.

基于多头注意力机制的BM-Linear信用贷款评估模型

信贷评估模型可加快放贷效率、缩减放贷时间。利用Pytorch深度学习框架,组合Bag-of-Words及Bert中多头注意力机制得到BM-Linear评估模型,同时在引入多组信贷训练集的前提下,创造性地构建参数独立训练及参数共用训练的对比实验,探究BM-Linear的优异性。研究表明:BM-Linear首先弱化与信贷训练集的对应关系,解决信贷模型受限于信贷场景问题,减少因反复训练模型所造成的放贷效率低下现象;其次,忽略缺失特征并将离散特征转为信贷文本,降低特征处理造成的信贷干扰,提高信贷评估效率;最后,克服因词袋与信贷词语对应关系所带来的词向量固化问题,实现动态词向量过程,进而提高评估准确率。所提出的BM-Linear模型,可为信贷机构高效评估快速放贷提供支持。

An Intelligent Deep Neural Sentiment Classification Network

A Deep Neural Sentiment Classification Network(DNSCN)is devel-oped in this work to classify the Twitter data unambiguously.It attempts to extract the negative and positive sentiments in the Twitter database.The main goal of the system is tofind the sentiment behavior of tweets with minimum ambiguity.A well-defined neural network extracts deep features from the tweets automatically.Before extracting features deeper and deeper,the text in each tweet is represented by Bag-of-Words(BoW)and Word Embeddings(WE)models.The effectiveness of DNSCN architecture is analyzed using Twitter-Sanders-Apple2(TSA2),Twit-ter-Sanders-Apple3(TSA3),and Twitter-DataSet(TDS).TSA2 and TDS consist of positive and negative tweets,whereas TSA3 has neutral tweets also.Thus,the proposed DNSCN acts as a binary classifier for TSA2 and TDS databases and a multiclass classifier for TSA3.The performances of DNSCN architecture are evaluated by F1 score,precision,and recall rates using 5-fold and 10-fold cross-validation.Results show that the DNSCN-WE model provides more accuracy than the DNSCN-BoW model for representing the tweets in the feature encoding.The F1 score of the DNSCN-BW based system on the TSA2 database is 0.98(binary classification)and 0.97(three-class classification)for the TSA3 database.This system provides better a F1 score of 0.99 for the TDS database.

基于Bag-of-phrases的图像表示方法

在过去的几年,将图像内容表示为特定"视觉词"出现次数直方图的Bag-of-words模型,展示了其在图像内容分类方面的强大优势.然而,在这种统计特定"视觉词"出现次数直方图的模型中,"视觉词"之间的相互位置关系几乎被完全丢弃了.本文从分析Bag-of-words模型在文本分类和图像内容分类领域的对应关系的角度出发,提出一种加入"视觉词"之间的相互位置关系的图像表示方法—Bag-of-phrases模型.在标准数据集上验证了该图像表示方法对图像内容分类性能的影响.实验结果显示,本文提出的方法相对于传统的Bag-of-words模型可以达到更好的分类性能.

基于改进PLSA分类器的目标分类算法

通过SIFT描述目标特征,利用Bag-of-words模型将目标特征构建为codebook,通过PLSA分类器对目标进行分类,根据PLSA分类学习过程中存在迭代复杂的问题,将贝叶斯分类器中的直接统计方法替换PLSA中最大似然估计,为PLSA提供足够的先验知识,减少学习过程中迭代次数,实验结果表明,相比于传统PLSA分类算法,本文方法检测结果较为准确,算法切实可行。

基于几何信息的特征匹配改进算法

现在常用的SIFT(Scale Invariant Feature Transform)特征匹配算法忽视了特征点之间的几何信息,而BP-SIFT算法(Belief Propagation,置信传播)虽然采用几何信息进行特征匹配,但是特征点之间并不总是满足距离相等的约束条件,根据空间距离来选择临近节点只能代表图像部分区域的几何信息,而且算法的时间开销太大.提出以特征点空间距离和所有邻近节点空间距离的均值比值作为约束条件.并且提出了根据BOW(Bag-of-Words)模型确定邻近节点.实验表明,采用新的约束条件和邻近节点的选择方法不仅可以提高特征匹配的准确率,而且提高了算法速度.

基于SIFT-SVM的北冰洋海冰识别研究

本文提出以SIFT特征为海冰识别关键技术的复杂场景SAR图像海冰识别方法,首先从SAR图像中选取不同类别的图像作为样本集,提取图像的尺度不变特征,然后利用K-means聚类方法将样本特征集构建为构造一个包含K个词汇的词典,找到词典中与样本特征相匹配的单词,统计不同类别特征集中每个单词的出现次数,从而将图像表示为K为特征向量。最后设计一种基于支持向量机的分类器,实现对海冰SAR图像中不同类型海冰区域的识别。

Image Classification Based on the Fusion of Complementary Features

Image classification based on bag-of-words（BOW）has a broad application prospect in pattern recognition field but the shortcomings such as single feature and low classification accuracy are apparent.To deal with this problem,this paper proposes to combine two ingredients：（i）Three features with functions of mutual complementation are adopted to describe the images,including pyramid histogram of words（PHOW）,pyramid histogram of color（PHOC）and pyramid histogram of orientated gradients（PHOG）.（ii）An adaptive feature-weight adjusted image categorization algorithm based on the SVM and the decision level fusion of multiple features are employed.Experiments are carried out on the Caltech101 database,which confirms the validity of the proposed approach.The experimental results show that the classification accuracy rate of the proposed method is improved by 7%-14%higher than that of the traditional BOW methods.With full utilization of global,local and spatial information,the algorithm is much more complete and flexible to describe the feature information of the image through the multi-feature fusion and the pyramid structure composed by image spatial multi-resolution decomposition.Significant improvements to the classification accuracy are achieved as the result.

Human Action Recognition Based on Dense Trajectories Analysis and Random Forest

This paper presents a human action recognition method. It analyzes the spatio-temporal grids along the dense trajectories and generates the histogram of oriented gradients （HOG） and histogram of optical flow （HOF） to describe the appearance and motion of the human object. Then, HOG combined with HOF is converted to bag-of-words （BoWs） by the vocabulary tree. Finally, it applies random forest to recognize the type of human action. In the experiments, KTH database and URADL database are tested for the performance evaluation. Comparing with the other approaches, we show that our approach has a better performance for the action videos with high inter-class and low inter-class variabilities.

Bag-of-visual-words model for artificial pornographic images recognition

It is illegal to spread and transmit pornographic images over internet,either in real or in artificial format.The traditional methods are designed to identify real pornographic images and they are less efficient in dealing with artificial images.Therefore,criminals turn to release artificial pornographic images in some specific scenes,e.g.,in social networks.To efficiently identify artificial pornographic images,a novel bag-of-visual-words based approach is proposed in the work.In the bag-of-words(Bo W)framework,speeded-up robust feature(SURF)is adopted for feature extraction at first,then a visual vocabulary is constructed through K-means clustering and images are represented by an improved Bo W encoding method,and finally the visual words are fed into a learning machine for training and classification.Different from the traditional BoW method,the proposed method sets a weight on each visual word according to the number of features that each cluster contains.Moreover,a non-binary encoding method and cross-matching strategy are utilized to improve the discriminative power of the visual words.Experimental results indicate that the proposed method outperforms the traditional method.

QDCT Encoding-Based Retrieval for Encrypted JPEG Images

A privacy-preserving search model for JPEG images is proposed in paper,which uses the bag-of-encrypted-words based on QDCT(Quaternion Discrete Cosine Transform)encoding.The JPEG image is obtained by a series of steps such as DCT(Discrete Cosine Transform)transformation,quantization,entropy coding,etc.In this paper,we firstly transform the images from spatial domain into quaternion domain.By analyzing the algebraic relationship between QDCT and DCT,a QDCT quantization table and QDTC coding for color images are proposed.Then the compressed image data is encrypted after the steps of block permutation,intra-block permutation,single table substitution and stream cipher.At last,the similarity between original image and query image can be measured by the Manhattan distance,which is calculated by two feature vectors with the model of bag-of-words on the cloud server side.The outcome shows good performance in security attack and retrieval accuracy.

JudPriNet: Video transition detection based on semantic relationship and Monte Carlo sampling

Video understanding and content boundary detection are vital stages in video recommendation.However,previous content boundary detection methods require collecting information,including location,cast,action,and audio,and if any of these elements are missing,the results may be adversely affected.To address this issue and effectively detect transitions in video content,in this paper,we introduce a video classification and boundary detection method named JudPriNet.The focus of this paper is on objects in videos along with their labels,enabling automatic scene detection in video clips and establishing semantic connections among local objects in the images.As a significant contribution,JudPriNet presents a framework that maps labels to“Continuous Bag of Visual Words Model”to cluster labels and generates new standardized labels as video-type tags.This facilitates automatic classification of video clips.Furthermore,JudPriNet employs Monte Carlo sampling method to classify video clips,the features of video clips as elements within the framework.This proposed method seamlessly integrates video and textual components without compromising training and inference speed.Through experimentation,we have demonstrated that JudPriNet,with its semantic connections,is able to effectively classify videos alongside textual content.Our results indicate that,compared with several other detection approaches,JudPriNet excels in high-level content detection without disrupting the integrity of the video content,outperforming existing methods.

微信公众平台的转基因新闻报道框架偏向性研究

转基因议题一直受到媒体与公众的广泛关注,近年来微信公众平台成为新闻消费重要的来源,微信公众平台的转基因新闻报道具有突出的研究价值。在大数据和信息爆炸的时代背景下,本文提出新闻议题框架研究的新思路,主要采用语义网络和bag-of-words模型对微信公众平台转基因新闻报道框架偏向性进行研究。研究发现,中央主流媒体、生活服务媒体与科普类媒体对转基因不同议题框架明显呈现不同的报道偏向性。本文针对媒体报道现状存在的问题提出了建议。

Beyond bag of latent topics: spatial pyramid matching for scene category recognition

We propose a heterogeneous, mid-level feature based method for recognizing natural scene categories. The proposed feature introduces spatial information among the latent topics by means of spatial pyramid, while the latent topics are obtained by using probabilistic latent semantic analysis （pLSA） based on the bag-of-words representation. The proposed feature always performs better than standard pLSA because the performance of pLSA is adversely affected in many cases due to the loss of spatial information. By combining various interest point detectors and local region descriptors used in the bag-of-words model, the proposed feature can make further improvement for diverse scene category recognition tasks. We also propose a two-stage framework for multi-class classification. In the first stage, for each of possible detector/descriptor pairs, adaptive boosting classifiers are employed to select the most discriminative topics and further compute posterior probabilities of an unknown image from those selected topics. The second stage uses the prod-max rule to combine information coming from multiple sources and assigns the unknown image to the scene category with the highest ＇final＇ posterior probability. Experimental results on three benchmark scene datasets show that the proposed method exceeds most state-of-the-art methods.

Visual polysemy and synonymy:toward near-duplicate image retrieval

that are duplicate or near duplicate to a query image.One of the most popular and practical methods in near-duplicate image retrieval is based on bag-of-words(BoW)model.However,the fundamental deficiency of current BoW method is the gap between visual word and image’s semantic meaning.Similar problem also plagues existing text retrieval.A prevalent method against such issue in text retrieval is to eliminate text synonymy and polysemy and therefore improve the whole performance.Our proposed approach borrows ideas from text retrieval and tries to overcome these deficiencies of BoW model by treating the semantic gap problem as visual synonymy and polysemy issues.We use visual synonymy in a very general sense to describe the fact that there are many different visual words referring to the same visual meaning.By visual polysemy,we refer to the general fact that most visual words have more than one distinct meaning.To eliminate visual synonymy,we present an extended similarity function to implicitly extend query visual words.To eliminate visual polysemy,we use visual pattern and prove that the most efficient way of using visual pattern is merging visual word vector together with visual pattern vector and obtain the similarity score by cosine function.In addition,we observe that there is a high possibility that duplicates visual words occur in an adjacent area.Therefore,we modify traditional Apriori algorithm to mine quantitative pattern that can be defined as patterns containing duplicate items.Experiments prove quantitative patterns improving mean average precision(MAP)significantly.