Salient Region Extraction provides an alternative methodology to image description in many applications such as adaptive content delivery and image retrieval. In this paper, we propose a robust approach to extracting the salient region based on bottom-up visual attention. The main contributions are twofold: 1) Instead of the feature parallel integration, the proposed saliencies are derived by serial processing between texture and color features. Hence, the proposed approach intrinsically provides an alternative methodology to model attention with low implementation complexity. 2) A constructive approach is proposed for rendering an image by a non-linear intensity mapping, which can efficiently eliminate high contrast noise regions in the image. And then the salient map can be robustly generated for a variety of nature images. Experiments show that the proposed algorithm is effective and can characterize the human perception well.

This paper discusses the issues involved in the design of a complete information retrieval system for DataSpace based on user relevance probabilistic schemes. First, Information Hidden Model (IHM) is constructed taking into account the users' perception of similarity between documents. The system accumulates feedback from the users and employs it to construct user oriented clusters. IHM allows integrating uncertainty over multiple, interdependent classifications and collectively determines the most likely global assignment. Second, Three different learning strategies are proposed, namely query-related UHH, UHB and UHS (User Hidden Habit, User Hidden Background, and User Hidden keyword Semantics) to closely represent the user mind. Finally, the probability ranking principle shows that optimum retrieval quality can be achieved under certain assumptions. An optimization algorithm to improve the effectiveness of the probabilistic process is developed. We first predict the data sources where the query results could be found. Therefor, compared with existing approaches, our precision of retrieval is better and do not depend on the size and the DataSpace heterogeneity.

This paper focuses on the relationship between the number of interest points and the accuracy rate in scene classification. Here, we accept the common belief that more interest points can generate higher accuracy. But, few effort have been done in this field. In order to validate this viewpoint, in our paper, extensive experiments based on bag of words method are implemented. In particular, three different SIFT descriptors and five feature selection methods are adopted to change the number of interest points. As innovation point, we propose a novel dense SIFT descriptor named Octave Dense SIFT, which can generate more interest points and higher accuracy, and a new feature selection method called number mutual information (NMI), which has better robustness than other feature selection methods. Experimental results show that the number of interest points can aggressively affect classification accuracy.

Bag-of-Visual-Words representation has recently become popular for scene classification. However, learning the visual words in an unsupervised manner suffers from the problem when faced these patches with similar appearances corresponding to distinct semantic concepts. This paper proposes a novel supervised learning framework, which aims at taking full advantage of label information to address the problem. Specifically, the Gaussian Mixture Modeling (GMM) is firstly applied to obtain "semantic interpretation" of patches using scene labels. Each scene induces a probability density on the low-level visual features space, and patches are represented as vectors of posterior scene semantic concepts probabilities. And then the Information Bottleneck (IB) algorithm is introduce to cluster the patches into "visual words" via a supervised manner, from the perspective of semantic interpretations. Such operation can maximize the semantic information of the visual words. Once obtained the visual words, the appearing frequency of the corresponding visual words in a given image forms a histogram, which can be subsequently used in the scene categorization task via the Support Vector Machine (SVM) classifier. Experiments on a challenging dataset show that the proposed visual words better perform scene classification task than most existing methods.

Crowd flow prediction in high density urban scenes is involved in a wide range of intelligent transportation and smart city applications, and it has become a significant topic in urban computing. In this letter, a CNN-based framework called Pyramidal Spatio-Temporal Network (PSTNet) for crowd flow prediction is proposed. Spatial encoding is employed for spatial representation of external factors, while prior pyramid enhances feature dependence of spatial scale distances and temporal spans, after that, post pyramid is proposed to fuse the heterogeneous spatio-temporal features of multiple scales. Experimental results based on TaxiBJ and MobileBJ demonstrate that proposed PSTNet outperforms the state-of-the-art methods.

Abnormal behavior detection is now a widely concerned research field, especially for crowded scenes. However, most traditional unsupervised approaches often suffered from the problem when the normal events in the scenario with large visual variety. This paper proposes a self-learning probabilistic Latent Semantic Analysis, which aims at taking full advantage of the high-level abnormal information to solve problems. We select the informative observations to construct the “reference events” from the training sets as a high-level guidance cue. Specifically, the training set is randomly divided into two separate subsets. One is used to learn this model, which is defined as the initialization sequence of “reference events”. The other aims to update this model and the the infrequent samples are chosen into the “reference events”. Finally, we define anomalies using events that are least similar to “reference events”. The experimental result demonstrates that the proposed model can detect anomalies accurately and robustly in the real-world crowd environment.

This paper presents an efficient yet powerful codebook model, named classified codebook model, to categorize natural scene category. The current codebook model typically resorts to large codebook to obtain higher performance for scene categorization, which severely limits the practical applicability of the model. Our model formulates the codebook model with the theory of vector quantization, and thus uses the famous technique of classified vector quantization for scene-category modeling. The significant feature in our model is that it is beneficial for scene categorization, especially at small codebook size, while saving much computation complexity for quantization. We evaluate the proposed model on a well-known challenging scene dataset: 15 Natural Scenes. The experiments have demonstrated that our model can decrease the computation time for codebook generation. What is more, our model can get better performance for scene categorization, and the gain of performance becomes more pronounced at small codebook size.

We present a novel model, named Category Constraint-Latent Dirichlet Allocation (CC-LDA), to learn and recognize natural scene category. Previous work had to resort to additional classifier after obtaining image topic representation. Our model puts the category information in topic inference, so every category is represented in a different topics simplex and topic size, which is consistent with human cognitive habit. The significant feature in our model is that it can do discrimination without combined additional classifier, during the same time of getting topic representation. We investigate the classification performance with variable scene category tasks. The experiments have demonstrated that our learning model can get better performance with less training data.

Localized content-based image retrieval (LCBIR) has emerged as a hot topic more recently because in the scenario of CBIR, the user is interested in a portion of the image and the rest of the image is irrelevant. In this paper, we propose a novel region-level relevance feedback method to solve the LCBIR problem. Firstly, the visual attention model is employed to measure the regional saliency of each image in the feedback image set provided by the user. Secondly, the regions in the image set are constructed to form an affinity matrix and a novel propagation energy function is defined which takes both low-level visual features and regional significance into consideration. After the iteration, regions in the positive images with high confident scores are selected as the candidate query set to conduct the next-round retrieval task until the retrieval results are satisfactory. Experimental results conducted on the SIVAL dataset demonstrate the effectiveness of the proposed approach.

Face verification in the presence of age progression is an important problem that has not been widely addressed. Despite appearance changes for same person due to aging, they are more similar compared to facial images from different individuals. Hence, we design common and adapted vocabularies, where common vocabulary describes contents of general population and adapted vocabulary represents specific characteristics of one of image facial pairs. And the other image is characterized with a concatenation histogram of common and adapted visual words counts, termed as “age-invariant distinctive representation”. The representation describes whether the image content is best modeled by the common vocabulary or the corresponding adapted vocabulary, which is further used to accomplish the face verification. The proposed approach is tested on the FGnet dataset and a collection of real-world facial images from identification card. The experimental results demonstrate the effectiveness of the proposed method for verification of identity at a modest computational cost.

In this paper, an adaptive updating probabilistic model is proposed to track an object in real-world environment that includes motion blur, illumination changes, pose variations, and occlusions. This model adaptively updates tracker with the searching and updating process. The searching process focuses on how to learn appropriate tracker and updating process aims to correct it as a robust and efficient tracker in unconstrained real-world environments. Specifically, according to various changes in an object's appearance and recent probability matrix (TPM), tracker probability is achieved in Expectation-Maximization (EM) manner. When the tracking in each frame is completed, the estimated object's state is obtained and then fed into update current TPM and tracker probability via running EM in a similar manner. The highest tracker probability denotes the object location in every frame. The experimental result demonstrates that our method tracks targets accurately and robustly in the real-world tracking environments.

Appearance changes conform to certain rules for a same person,while for different individuals the changes are uncontrolled. Hence, this paper studies the age progression rules to tackle face verification task. The age progression rules are discovered in the difference space of facial image pairs. For this, we first represent an image pair as a matrix whose elements are the difference of a set of visual words. Thereafter, the age progression rules are trained using Support Vector Machine (SVM) based on this matrix representation. Finally, we use these rules to accomplish the face verification tasks. The proposed approach is tested on the FGnet dataset and a collection of real-world images from identification card. The experimental results demonstrate the effectiveness of the proposed method for verification of identity.

This paper proposes the Extended Bag-of-Visterms (EBOV) to represent semantic scenes. In previous methods, most representations are bag-of-visterms (BOV), where visterms referred to the quantized local texture information. Our new representation is built by introducing global texture information to extend standard bag-of-visterms. In particular we apply the adaptive weight to fuse the local and global information together in order to provide a better visterm representation. Given these representations, scene classification can be performed by pLSA (probabilistic Latent Semantic Analysis) model. The experiment results show that the appropriate use of global information improves the performance of scene classification, as compared with BOV representation that only takes the local information into account.