Search

Lithium ion batteries (LIBs) are widely used as important energy sources for mobile phones, electric vehicles, and drones. Experts have attempted to replace liquid electrolytes with solid electrolytes that have wider electrochemical window and higher stability due to the potential safety risks, such as electrolyte leakage, flammable solvents, poor thermal stability, and many side reactions caused by liquid electrolytes. However, finding suitable alternative materials using traditional approaches is very difficult due to the incredibly high cost in searching. Machine learning (ML)-based methods are currently introduced and used for material prediction. However, learning tools designed for domain experts to conduct intuitive performance comparison and analysis of ML models are rare. In this case, we propose an interactive visualization system for experts to select suitable ML models and understand and explore the predication results comprehensively. Our system uses a multifaceted visualization scheme designed to support analysis from various perspectives, such as feature distribution, data similarity, model performance, and result presentation. Case studies with actual lab experiments have been conducted by the experts, and the final results confirmed the effectiveness and helpfulness of our system.

Accurate tooth segmentation from CBCT images is a crucial step for specialist to perform quantitative analysis, clinical diagnosis and operation. In this paper, we present a symmetric full convolutional network with residual block and Dense Conditional Random Field (DCRF), which can achieve accurate segmentation automatically for tooth images. The proposed method can not only strengthen feature propagation, but also boost feature reuse, which can be credited to the contracting path and the expanding path that extract and recover pixel cues sufficiently. To this end, we apply special deep bottleneck architectures (DBAs) and summation-based skip connection into our network to ensure accurate segmentation for much deeper neural network. Compared with previous methods which are based on conditional random field for original image intensity, our approach applies DCRF to the posterior probability generated by the proposed network. To avoid the interferences of noises around the tooth, we combine the pixel-level prediction capability of DCRF, which further enhance the segmentation performance. In the experiments, we verify the capabilities of our methods based on four evaluation indicators, which demonstrates the superiority of our method.

Target tracking based on Siamese network has reached the state-of-the-art performance. However is still limited in semantic feature extraction. In this paper, we propose a novel method to distinguish positive and negative samples. Taking deep neural network as the backbone, we fuse the feature maps from different layers and feed it to RPN (Region Proposal Network). In addition, we use a loss term for loss function to achieve self-adjusting and learn more discriminative embedding features of target objects with similar semantics. In the tracking stage, one-shot detection is used as the reference, fix the first frame as the weight of tracking to track the subsequent frames. Our method has achieved outstanding performance on several benchmark data set, such as: OTB2015, VOT2016, VOT2018, and VOT2019 et al.

Modern multistandard wireless systems require reconfigurability for mult iband operat ion; therefore, switchable or reconfigurable frontend filters are in high demand. This article introduces a four-channel switchable filter bank, which was awarded first place in the Four-Channel Switchable/Reconfigurable Filter Bank category of the Student Design Competition (SDC) sponsored by Technical Committee 8 at the 2019 IEEE Microwave Theory and Techniques Society International Microwave Symposium.

Currently, object detectors based on CNN, such as RetinaNet, Faster-RCNN, CornerNet series, can achieve good performance, but have some common drawbacks, like large calculation cost, high model complexity and slow detection speed. In this paper, a new lightweight object detector is proposed, which adopted a density-based approach to merge the real boxes. To reduce calculation cost and improve detection speed, the tactic of multi-scale output is adopted to predict objects of different sizes with features of different scales. Furthermore, a new lightweight network model is proposed, which can show better performance in computation, FPS, and model complexity. Meanwhile, the separation of convolution is used to improve the basic convolution layer, which can achieve better results under the same number of filters. In the experiments, we verified the capability of our methods based on ablation experiment and model evaluation, which demonstrates the superiority of our method. Moreover, we have also conducted deep network and multichannel experiments on MS-COCO2014 datasets and achieved 20.9% mAP performance.

For the actual quantum key distribution (QKD) system, parameter optimization, the selection of the corresponding signal strength and transmission probability are the keys to obtain the best performance, especially when performing low-latency communication. Combining the random forest algorithm with data preprocessing, according to the relative performance parameters of each decision tree, the decision trees with poor performance is eliminated, so that the random forest algorithm is improved. This method is used to predict the best parameters of the QKD system directly, instead of searching for the best parameters. We tested the random forest algorithms on hardware equipment. We find that, compared with the traditional random forest algorithm, we have made great progress and achieved a relatively high optimal key rate. It has broad application prospects in the field of QKD.

Accurate segmentation of rectal cancer from CT images of rectal cancer is the basis of rectal cancer diagnosis. In this article, we propose a new segmentation method based on fully convolutional neural networks. The encoding block consists of a channel concatenation block and a residual block (CesBlock) and a lightweight atrous convolution block (LacBlock) for skip connection. By combining dice loss and cross entropy loss as the new loss function, the obtained model can realize automatic and accurate segmentation of CT images of rectal cancer. The contraction path of the symmetrical convolutional neural network ensures enough feature extraction, and the expansion path ensures the recovery of pixels. In the experiments, the recognized evaluation indicators are utilized to verify the proposed method, which has an improvement of 2.7% on the dice similarity coefficient (DSC).

COVID-19 is a new type of infectious disease that has been massively outbroken in December 2019. So far, with the development and promotion of the vaccines against COVID-19, the epidemic throughout the world appears to be contained. In this paper, we propose a multi-layer prediction method with a SEIRD model which is improved from a classical Susceptible-Exposed-Infected-Recovered (SEIR) model. We expand the classical model by taking the effect of vaccines into account. Furthermore, we use epidemic data accessed from the U.S. government to fit the model and visualize the results, a conclusion that the U.S. will be able to contain its epidemic effectively by practicing the development and injection of vaccines can be inferred.

In this paper, we present a two stages approach to generate the color cartoon image from sketch by generating the grayscale image firstly, then the color cartoon is generated from the grayscale image with colorization methodology. The grayscale cartoon image from sketch is generated with conditional generative adversarial networks (cGAN), the conditions of cGAN will have some effects on the grayscale image such as edge and shade. The final color cartoon image from the grayscale image is generated with Encoder-Decoder UNet, and in this phase, both the prior global and local user hints information will be introduced to control the final color scheme. The benefit of this two phases approach is to generate the high-quality color cartoon image while allowing users to control final effect with more user guide involved. The two phases colorization could be done in almost real-time and user could take tuning for the generated color cartoon image interactively.

Recently, the combination of remote sensing image processing and deep learning methods is an increasingly popular trend. In this paper, we combine the existing instance segmentation model Mask R-CNN and the target detection algorithm Softer-NMS, and propose a model based on the improved Mask R-CNN. The model can capture the edges well, thereby segmenting the target more accurately. The experimental results also verify the effectiveness of our method. Compared with the prototype Mask R-CNN, our method improves the overall segmentation accuracy, and the loss function of the mask branch and the loss function of the candidate frame of the RPN drop faster than the original model. Incidentally, there is also a good improvement in the accuracy of detection.

The 3-D reconstruction methods based on structure from motion (SfM) pipeline mainly use the traditional scale invariant feature detecting methods for feature matching. This type of methods suffers from the accuracy with affine matching in the image-based modeling system. In this paper, we propose an affine-invariant feature detection and matching method which is accurate and fast based on the three types of critical points in Morse theory called precise extreme feature region (PEFR). We also exploit the evaluation method for the reconstruction results without the ground truth. We propose the average descending rate to quantitatively evaluate the accuracy of the reconstruction model. Using PEFR in an SfM pipeline to build 3-D point-cloud, our method presents better results and less runtime compared to the state-of-the-art approaches. In addition, our quality assessment method presents a practical comparison of the reconstruction result based on the geometric accuracy of the reconstruction models. The experimental results prove that our PEFR method is faster and more accurate and works well on the reconstruction pipeline. The assessment method is fit for the quality assessment on 3-D reconstruction.

Constructing a robust appearance model of the visual object is a crucial task for visual object tracking. Recently, more and more studies combine spatial feature with a temporal feature to improve the tracking performance. These methods successfully apply the features from spatial and temporal to address the problem for tracking. This paper presents a novel method for visual object tracking based on spatiotemporal feature combined with correlation filters. In this paper, the visual features of a target object are extracted from a spatial–temporal residual network (STResNet) appearance model with two sub-networks. The STResNet appearance model learns separately spatial feature and temporal feature, respectively, so that we can effectively utilize spatial context around the surrounding of the target object in each frame and the temporal relationship between successive frames to refine the appearance representation of the target object. Finally, our spatiotemporal fusion feature from STResNet appearance model is incorporated into the correlation filter for robust visual object tracking. The experimental results show that our method achieves similar or better performance compared with the other tracking methods based on convolutional neural networks or correlation filter.

Accurate segmentation of entity categories is the critical step for 3D scene understanding. This paper presents a fast deep neural network model with Dense Conditional Random Field (DCRF) as a post-processing method, which can perform accurate semantic segmentation for 3D point cloud scene. On this basis, a compact but flexible framework is introduced for performing segmentation to the semantics of point clouds concurrently, contribute to more precise segmentation. Moreover, based on semantics labels, a novel DCRF model is elaborated to refine the result of segmentation. Besides, without any sacrifice to accuracy, we apply optimization to the original data of the point cloud, allowing the network to handle fewer data. In the experiment, our proposed method is conducted comprehensively through four evaluation indicators, proving the superiority of our method.

Lithium ion batteries (LIBs) are widely used as the important energy sources in our daily life such as mobile phones, electric vehicles, and drones etc. Due to the potential safety risks caused by liquid electrolytes, the experts have tried to replace liquid electrolytes with solid ones. However, it is very difficult to find suitable alternatives materials in traditional ways for its incredible high cost in searching. Machine learning (ML) based methods are currently introduced and used for material prediction. But there is rarely an assisting learning tools designed for domain experts for institutive performance comparison and analysis of ML model. In this case, we propose an interactive visualization system for experts to select suitable ML models, understand and explore the predication results comprehensively. Our system employs a multi-faceted visualization scheme designed to support analysis from the perspective of feature composition, data similarity, model performance, and results presentation. A case study with real experiments in lab has been taken by the expert and the results of confirmed the effectiveness and helpfulness of our system.

Semantic segmentation on medical Computed Tomography (CT) images is of great significance to research and clinical diagnosis. However, methods based on neural network have competitive advantages for segmentation of dental CT images. In this paper, a 3D multi-feature fusion method for tooth segmentation is proposed. In order to obtain the body space of the data, first of all, the dental CT training set is compressed in NII format, and the body space data is processed; then the proposed 3D convolution network is used to train the data, extract the feature vectors, and obtain the probability distribution; to handle the situation that 3D neural network always leads to fuzzy boundary and unclear topology, the new CRF algorithm is used to refine the probability distribution which removes the redundant information generated by the neural network model, and makes the segmentation results more accurate. Compared with diverse contemporary segmentation algorithms, the effectiveness and superiority of our proposed method are verified, proving the conclusion that the supervision mechanism, neural network model components, and optimization proposed methods can improve the accuracy of tooth segmentation is reliable and valid.

The gradient boosting machine (GBM) composed of multiple weak learners is an efficient and widely used machine learning method. As a key factor in the prediction process of the gradient boosting machine, feature affects the performance of the gradient boosting machine when splitting nodes. The idea of GBM gives it a natural advantage to discover a variety of distinguishing features and feature combinations. Once the gradient boosting machine has the correct features, other factors play a relatively weak role. However, the GBM is a complicated and tedious process with diverse structure and attributes of decision tree, leading the model to be less interpretable, especially for high risk areas such as medical diagnosis and financial analytics that require transparent prediction. To tackle this issue, we have proposed an interactive visual analytic system, GBMVis, to help experts quickly analyze and of the gradient boosting machine. In addition to providing information about the features, we have also provided a visualization of the structure of boosting trees, which aims to display the major data flow in the gradient boosting machine. We have demonstrated the effectiveness of our system in a real dataset.

Accurate segmentation of entity categories is the key step for 3D scene understanding. In this paper, we present a novel fast Deep Neural Network (DNN) model with Dense Conditional Random Field (DCRF) as a post-processing step, which can perform accurate semantic segmentation for 3D point cloud scene. On this basis, a compact but flexible framework is introduced for performing segmentation to the semantics of point clouds concurrently, contribute to more precise segmentation. Moreover, based on the labels of semantics, a novel DCRF model is elaborated to refine the result of segmentation. Besides, we apply optimization to the original point cloud, allowing the network to handle fewer points without any sacrifice to accuracy. In the experiment, our proposed method is comprehensively evaluated through four indicators, and achieves state-of-art performance.

The development of the Internet has made social communication increasingly important for maintaining relationships between people. However, advertising and fraud are also growing incredibly fast and seriously affect our daily life, e.g., leading to money and time losses, trash information, and privacy problems. Therefore, it is very important to detect anomalies in social networks. However, existing anomaly detection methods cannot guarantee the correct rate. Besides, due to the lack of labeled data, we also cannot use the detection results directly. In other words, we still need human analysts in the loop to provide enough judgment for decision making. To help experts analyze and explore the results of anomaly detection in social networks more objectively and effectively, we propose a novel visualization system, egoDetect, which can detect the anomalies in social communication networks efficiently. Based on the unsupervised anomaly detection method, the system can detect the anomaly without training and get the overview quickly. Then we explore an ego&rsquo, s topology and the relationship between egos and alters by designing a novel glyph based on the egocentric network. Besides, it also provides rich interactions for experts to quickly navigate to the interested users for further exploration. We use an actual call dataset provided by an operator to evaluate our system. The result proves that our proposed system is effective in the anomaly detection of social networks.

All-flash array is a popular memory device available for use in modern high-performance storage systems. Compared with other types of devices such as DRAM, NVRAM, and EEPROM, flash array combines the best features: shock resistance, low cost, low power consumption, and fast access. Moreover, the ever-increasing density of flash memory has led to a dramatic increase in the capacity, which allows the storage of large volume of data. However, flash memory is not optimal for managing a large number of small files because: 1. the small and random write operation is inefficient in flash memory; 2. massive metadata information occupies a significant portion of the namespace, which is relatively limited or scarce in big data storage systems. This paper introduces a novel approach, hash partitioning-based file compaction (HFC), to improve the efficiency of storing and accessing small files in all-flash storage systems. HFC consists of a file compaction tool and an access interface. The compaction tool merges a group (usually a directory) of small files into a set of "big files" to reduce the metadata required to be maintained in the on-chip memory. The data locality and tree structure of those small files are preserved. The access interface is designed to provide transparent access to the small files in the HFC big files. Experimental results confirm that the proposed method significantly enhances the efficiency of managing massive amounts of small files in flash memory in terms of namespace usage and access speed.

In this paper, a new energy function is proposed that can aggregate the mesh model generated by the point cloud extracted from the UAV and supplement it with contextual semantics to accurately segment the building, which maximizes the consistency of the extracted buildings to restore detail. The semantic information is also used to improve the consistency of the labels between the semantic segments of the extracted input model to ensure the validity of the separation results. A new method of reconstructing polygon and arc models using unstructured models is proposed to improve large scale reconstruction. It can robustly discover the set of adjacency relations and repairs appropriately the non-watertight model due to point cloud loss. The experimental results show that the proposed large scale reconstruction algorithm is suitable for the modeling of complex urban buildings.

Imperceptibility and robustness are the two complementary, but fundamental requirements of any digital image watermarking method. To improve the invisibility and robustness of multiplicative image watermarking, a complex wavelet based watermarking algorithm is proposed by using the human visual texture masking and visual saliency model. First, image blocks with high entropy are selected as the watermark embedding space to achieve imperceptibility. Then, an adaptive multiplicative watermark embedding strength factor is designed by utilizing texture masking and visual saliency to enhance robustness. Furthermore, the complex wavelet coefficients of the low frequency sub-band are modeled by a Gaussian distribution, and a watermark decoding method is proposed based on the maximum likelihood criterion. Finally, the effectiveness of the watermarking is validated by using the peak signal-to-noise ratio (PSNR) and the structural similarity index measure (SSIM) through experiments. Simulation results demonstrate the invisibility of the proposed method and its strong robustness against various attacks, including additive noise, image filtering, JPEG compression, amplitude scaling, rotation attack, and combinational attack.

In this paper, we address the challenge about insufficiency of training set and limited feedback information in each relevance feedback (RF) round during the process of content based image retrieval (CBIR). We propose a novel active learning scheme to utilize the labeled and unlabeled images to build the initial Support Vector Machine (SVM) classifier for image retrieving. In our framework, two main components, a pseudo-label strategy and an improved active learning selection method, are included. Moreover, a feature subspace partition algorithm is proposed to model the retrieval target from users by the analysis from relevance labeled images. Experimental results demonstrate the superiority of the proposed method on a range of databases with respect to the retrieval accuracy.

Commodity object retrieval is a key issue in the application of self-service shopping and so on. In this paper, we propose a saliency object detection-based unsupervised commodity object retrieval scheme. Since most commodity objects are conspicuous and not complicated in commodity images, saliency detection could predict a saliency box that indicates approximate position information of objects. The proposed scheme utilizes the saliency box to filter the proposals extracted by selective search. The reserved proposals have a big overlapping ratio with saliency box to a large extent. This paper composes both the saliency box and the reserved proposals as saliency proposals. Furthermore, we propose a channel weighting generalized mean pooling feature to represent saliency proposals. On one hand, the reduction of proposals’ number after filtering significantly improves the computational efficiency; on the other hand, the new feature more accurately represents the objects to be retrieved, which results in higher retrieval precision. In addition, we built and manually annotated a commodity data set named PRODUCT to evaluate the proposed method. Extensive experiments are also conducted on the databases INSTRE and Flick32. The results demonstrate the superior performance of our scheme compared with the other state-of-the-art methods.

In this paper, we focus on the problem of image feature extraction and similarity measure using region division search. Specifically, we proposed a novel image region division to roughly mimic the location distribution of image color and deal with the color histogram failing to describe spatial information. Furthermore, an image descriptor combining local color histogram and Gabor texture features with reduced feature dimensions are developed for optimizing our region division search method. Moreover, an extended Canberra distance is proposed for images similarity measure to increase the faulttolerant ability of the whole large-scale image search. Extensive experiments on several benchmark image retrieval databases validate the superiority of the proposed approaches.

A novel approach for background and scene activity modelling with spearman correlation analysis and customized deep learning model is introduced in this paper. It detects and gives correlated analytics between casual and temporal regional activities on the basis of similarities and primary dissimilarities in the same scene captured by several cameras. The experiment implement on four overlapped videos that are captured inside the hall from four cameras. Detected and analyzed by our model, 17.32% correlated co-occurrences is actual correlation among all videos. Rest of 82.68% of videos is background that shows similar and repetitive features in spearman rank tied result. Simulation results demonstrate that the proposed method can detect high correlation among all activities during the frame rate with tied features ability.

Face tracking methods are increasingly critical for many expression mapping analysis applications, along its research track, deep convolutional neural network (DCNN-based) search techniques have attracted broad interests due to their high efficiency in 3D feature points. In this paper, we focus on the problem of 3D feature point’s extraction and expression mapping using a light-weight deep convolutional neural network (LW-DCNN) search and data conversion model, respectively. Specifically, we proposed novel light-weight deep convolutional neural network for 3D feature point’s extraction to solve the great initial shape errors in regression cascaded framework and the slow processing speed in traditional CNN. Furthermore, an effective data conversion model is proposed to generate the deformation coefficient to realize the expression mapping. Extensive experiments on several benchmark image databases validate the superiority of the proposed approaches.

As the result of Social media data being unprecedentedly available, we are provided so many substantial opportunities to explore social circle from many perspectives. Many researches have deep understandings of the correlation between social relation online and that on real world. These researches make conclusions from perspectives of statistics only but they cannot explain specific reasons underlying those conclusions. Obviously we can separate an individual’s social circle into several groups such as family group, classmate group and co-worker group etc. Aimed at labeling the friend groups online, in this paper, we implement a visual analytic system to exploring correlation between communities resulted from label propagation algorithm in online social friendship network of a centric user and each friend’s offline POI distribution of such places where the centric user has also visited. To demonstrate the reasonability and utility, we give 3 cases in details based on social media data of the online friendship and offline movement information provided by Tencent (the largest social service platform in China).

With the growth in the elderly population, fall detection methods for the elderly are of great significance. In this paper, we propose a deep learning-based method for real-time fall detection continuous depth maps with Residual Fall Detection Network (RFD-Net). Our method incorporates feature extraction with fall detection. In feature extraction part, seven important features that accurately represent the body posture are extracted from the depth maps to reduce the computation load. In the fall detection part, a novel RFD-Net is proposed to recognize body posture for fall detection. Meanwhile, two other networks are developed to compare with RFD-Net. The experimental results show that the extracted features are good representative of the body posture, and our method delivers performance with a fall detection accuracy of 98.51%, which is higher than other related methods.

this paper describes a deep learning based model for roads detection in Aerial image. In general, standard CNN networks would have less ability for tiny objects detection in remote sensing image. With this regard, we propose a novel fully convolutional network, which utilizes deconvolution layers and feature map fussing to take as input intensity and pixel-wise labeling. Moreover, the class prediction are used as the input to Condition Random Field (CRF) for the final pixel prediction. The Batch Normalization (BN) algorithm and two stages training strategy were used in our model to reduce the time cost of model training. Several experimental results conducted in Massach-useets. Road dataset demonstrate the superiority of our model with respect to accuracy and time cost.

This paper summarizes the recent progress in human motion analysis and its applications. The first part of this paper reviews the motion capture systems and the representations of human’s motion data. Next, the paper sketches the advanced human motion data processing technologies, including motion data filtering, temporal alignment, and segmentation. The following parts overview the state-of-the-art approaches of action recognition and dynamics measuring. The last part discusses the emerging applications of human motion analysis in healthcare and human robot interaction. The promising research topics of human motion analysis in the future are also summarized in the last part.

In the reconstruction surgery of Anterior Cruciate Ligament (ACL), how to locate the anatomical position is a very hard point to clinician occupational therapists. In this paper, we propose an Anatomical Position Locating (APL) approach based on Expectation Maximization (EM) algorithm. Firstly, the proposed intersection set operation algorithm is proposed to compute the attachment region between the injured ACL and femur or tibia. Then, the anatomical position is located by the 3D points cloud with the Gaussian spatial distribution. The last, the attachment spatial distribution and the barycenter, which are also viewed as the candidates of the anatomical position by a proposed EM algorithm, is partition. Experimental results verify our assumption and demonstrate that the located anatomical position has great serviceability.

Newly proposed video standard High Efficiency Video Coding (HEVC) achieves higher compression performance than previous ones. In this paper, we propose a novel algorithm for intra prediction, which scales the complexity of pictures' texture to perform different levels of simplification on Most Probable Mode(MPM) selection. And the proposed algorithm for inter prediction initializes current Coding Unit(CU) depth information with that information of temporally adjacent frame's co-located CU. These two proposed algorithms, utilizing the texture feature and correlation of adjacent frames, reduce the computational complexity to improve the efficiency of encoder. The proposed algorithms decrease more than 30 % of encoding time with nearly negligible increment in bit-rate, especially work well when encoding sequences with high definition.

In this paper, we propose a detection algorithm based on people counting for two special kinds of abnormal crowd behavior, gathering and dispersing. We use an efficient foreground segmentation algorithm for calculating the number of people, which uses an approximate median filter and double background model to obtain a reliable foreground. Further, counting people globally based on potential energy model in crowd scenes. In order to detecting unnormal crowd behavior happened, a crowd distribution curve is proposed, which combines results of counting and crowd entropy to evaluate the spatial distribution of throng, and describes the global distribution as a good feature. Experiments prove that our proposed method is able to detect the abnormal crowd behavior efficiently without camera calibration or supervised training.

Depth estimating from image is a essentially important work in many situations. However, traditional methods always extract depth information from binocular image pairs. Estimating depth information from a single image is much harder because single image lake the relationship between global and local coordinate. This paper proposes a single image depth estimating method by the segmentation convolutional neural network method. Our method aimed at getting the depth map from a single image with high revolution and high speed. The proposed method include three components: segmentation, coarse estimating and high revolution refine. Experiment results show the method can get high quality results. We compare our method with other methods on the accuracy and processing time to show the advantages.

Due to the time and hardware restriction, the amount of feedback information is limited in each image retrieval loop. To solve this problem, this paper proposes an effective relevance feedback (RF) method based on Support Vector Machine (SVM) framework, which increases the amount of feedback information by cluster analysis and utilizing unlabeled images to build SVM classifier. As a result, a pseudo-label strategy, consist of a feature subspace partition algorithm and a cluster analysis scheme, is proposed for unlabeled images selection. Experimental results demonstrate the relative high effectiveness of our proposed active framework.

Fusion-based method for video enhancement has been playing a basic but significant role, which is also proved high-efficiency. Still, there are some open questions, such as lamp-off problem, over-enhanced moving objects and night shadow. To resolve the problems, a novel method--sparse codes fusion (SCF) is proposed. With plenty of samples from daytime videos and nighttime videos of the same scene, we learn and obtain a daytime dictionary and a nighttime dictionary using the proposed mutual coherence learning (MCL) algorithm. These two dictionaries are utilized for fusion and extracting context enhanced background. Moreover, we reconstruct the nighttime dictionary to get nighttime background that would be applied in motion extraction. Then the moving objects are added into the enhanced background. Extensive experimental results show a highly comprehensive description of video frames that leads to improvements over the state of the art on many usual public video datasets.

In this paper, we propose a multi-local means based nearest neighbor classifier (MLMNN). In the MLMNN, k categorical nearest neighbors of a query sample are first found and used to calculate the corresponding k categorical multi-local mean vectors which can represent different local class-specific sample distributions. Then, the query sample is represented by a linear combination of k categorical local mean vectors and the representation coefficient of each local mean vector as the contribution to representing and classifying the query sample is obtained. Finally, the class-specific representation-based distance (i.e. reconstruction residual) between the query sample and k categorical multi-local mean vectors is adopted to determine the class label of the query sample. The experimental results on three popular face databases show that the proposed MLMNN method outperforms the related competitive KNN-based methods.

This paper aims to develop an effective flower classification approach using the technology of feature extraction. With this regard, a fused descriptor based on Pyramid Histogram of Visual Words (PHOW) is used to extract the color, texture and contour information of flower image. Secondly, Dictionary Learning and Locality-constrained Linear Coding (LLC) are operated on PHOW feature and then images are presented by means of Spatial Pyramid Pooling (SPP). Finally, the integrated feature is sent to Support Vector Machines (SVM) classifier for training and testing. By conducting an experiment study in Oxford 17 dataset, our proposed method provided a satisfactory result with an overall 86.17% accuracy rate.

Accurate 3D reconstruction of human tissue is a challenge problem in medical imaging. In this paper, a novel 3D reconstruction method of human brain MRI images is proposed based on the segmentation of human tissue. First, we propose a novel region-based growing algorithm to get points of an MRI image. Then, the moving cubes algorithm is used to reconstruct the accurate 3D object model. Further, in order to display well, a multiple angle observation is provided in our experiment. Experimental results show that the proposed method is better than the traditional methods.

This paper proposes a novel method combining dark channel prior (DCP) and bright channel prior (BCP) for single image dehazing. The proposed method achieves airlight approximations by implementing numerical proximity of atmospheric light, which use the average value of the DCP and BCP. Meanwhile, in order to reduce computational time, a fast guided filtration method is used for the transmission map refinement. Experimental results on a variety of outdoor hazy images show the effectiveness and robustness of the proposed algorithm.

To efficiently locate identical objects in heterogeneous cameras and possibly propagate reliable information between cameras and refine detection, many techniques were used to recognize vehicles. In this paper, we investigate several key problems and present a novel approach for automatic vehicle recognition (AVR) in multiple cameras for video surveillance application. We propose a level-based region comparison algorithm to AVR in multiple cameras. For improving the recognition accuracy, new license plate recognition method is also proposed. Experimental results show that the proposed algorithm is simple and efficient, and the quality of the composed image can be comparable with the results of the state-of-the-art methods.

To resolve video enhancement problems, a novel method of gradient domain fusion wherein gradient domain frames of the background in daytime video are fused with nighttime video frames is proposed. To verify the superiority of the proposed method, it is compared to conventional techniques. The implemented output of our method is shown to offer enhanced visual quality.

In this paper, we propose a novel approach to improve the visual appearance of nighttime video. The proposed method firstly converts the video from RGB color space to YCbCr color space, and then Y component is extracted to enhance illumination using a new image fusion method, which uses wavelet transform. Secondly, low-pass Gaussian filter is used to resolve vague areas problems of the enhanced background and moving regions. Last, the enhanced frames are reconstructed to RGB color space by combining the enhanced intensity component and corresponding color component. The experimental results demonstrate that the proposed approach is effective and competitive.

Soft-tissue segmentation has always been difficult point in the medical research of diagnosis of soft-tissue defects. Especially for Anterior Cruciate Ligament (ACL) rebuilding surgery, ACL segmentation from all soft-tissue inside knee joint, including Posterior Cruciate Ligament (PCL) and meniscus, is a very important task. In this paper, we propose a novel ACL segmentation method: Space Model Contrast Clustering-based (SMC-based) ACL Segmentation. Unlike the widely used processing method, such as segmentation by MRI gray values and Mimics segmentation drawing, the proposed method relies 3D model of knee joint to segment soft tissue by self-adaptive K-means clustering. Extensional experiments demonstrate that the proposed method can be capable of solving the problem of soft-tissue segmentation well and has achieved higher ACL segmentation efficiency.

Contrast enhancement is crucial to the domain of security and surveillance where limitations in dynamic range and lack of lighting sources prevent fine details of the scene from being captured. Here, we propose a method of nighttime video contrast enhancement based on genetic algorithms. Conversion from RGB to HSI and illumination component extraction were done firstly. Illumination-based enhancement which combines chromosome, corresponding operators and genetic algorithm was then applied to enhance the contrast and details of the video according to an objective fitness criterion. Image reconstruction followed previous procedures finally. Comparison of our proposed method with other automatic enhancement techniques such as histogram equalization shows that our method produces natural looking images/videos, especially when the dynamic range of the input image is high. Results obtained, both in terms of subjective and objective evaluation, show the superiority of the proposed method.