Search

Aiming at the crowd in high-definition video motion state of sudden changes in rapid detection of abnormal crowd behavior problem, this paper proposes a kind of abnormal behavior crowd detection method based on motion vector. This algorithm is established upon the Social Force Model, first, extracts the motion vector in the code stream of the high-definition compressed videos, computes the interaction force in the social force model and rapidly draws the characteristics of the moving crowd; then according to the algorithm, we perform the bag of words approach and histogram statistics on the intensity and angle of the interaction force flow; finally we analyze two histograms to distinguish the moving state of the crowd and fulfill the detection of the abnormal crowd movement. The simulation experiment shows the method compared with the traditional social force model, in the 1024 × 768HD video frame processing speed on the average increase of 30% in average, the discrimination of abnormal frame advance 35 frames, the recall to an average increase of 22%.

In face verification applications, precision rate and identifying liveness are two key factors. Traditional methods usually recognize global faces and can not gain good enough results when the faces are captured from different ages, or there are some interference factors, such as facial shade, etc. Besides, the false face attack will pose a great security risk. To solve the above problems, this study examines how to achieve reliable living face verification based on Multi-CNNs (convolutional neural networks) and Bayes probability. Participants are required to make several expressions in random orders and contents, to ensure their liveness. Then, an effective component-based method is proposed for face recognition, and synthesized multiple CNNs can help reflecting intensities of different components. Eventually, the similarity between faces is calculated by integrating the results of each CNN, with the help of Bayes probability. Comparative experiments demonstrate that our algorithm outperforms traditional methods in face recognition accuracy. Moreover, our algorithm has unique preponderance in that it can verify the liveness of users, which can achieve higher security.

The processes of motion target detection and tracking in most of traditional activity recognition methods are usually complicated and the application of these methods is limited. In this paper, we propose a fast violent activity recognition method based on motion vectors. First, we extract the motion vectors from compressed video data directly. Then, we analyze the features of the motion vectors in each frame and between frames, and get Region Motion Vectors descriptor (RMV). Finally, we use the Support Vector Machine (SVM) which takes the radial basis as the kernel function to classify the RMV and determine whether the violent activity exists in the video or not. Experimental results on several datasets have shown that the proposed method can detect 96.1% of the violent activities in videos (false probability is about 5.1%), and the calculation speed is very fast, which means the new method can be used in embedded systems.

Face retrieval is becoming increasingly useful and important for security maintenance operations. In actual applications, face retrieval is usually influenced by some changeable site conditions, such as various postures, expressions, camera angles, illuminations, and so on. In this paper, facial peculiar features are extracted and classified by dynamically integrated deep neural networks (DNNs), in order to enhance the adaptability in actual conditions. Firstly, eight kinds of facial components are detected and located by clustering analysis and Active Shape Model (ASM). Secondly, certain peculiar patterns are defined for each kind of facial component, and eight specialized DNNs are designed to extract features and classify components. Thirdly, the similarity between faces is calculated by dynamically integrating the results of each DNN. Comparative experiments on standard image sets and wild image sets demonstrate that our algorithm outperforms global feature models in retrieval accuracy. Our algorithm is particularly suitable for practical application with regard to natural real videos and images.

Detecting fall down behavior is a meaningful work in the area of public video surveillance and smart home care, as this behavior is often caused by accident but usually trigger serious result. However, the uncertain individual behavior, the difference between different cameras, and the complexity of real application scene make the work absolutely hard. In this paper, a robust fall down behavior recognition algorithm is proposed based on the spatial and temporal analysis of the Key Area of Human Body (KAHB). Firstly, a modified ViBe method is applied to extract motion area. Then a pre-trained human body classifier combined with histogram tracking is used to locate the KAHB and extract its normalized spatial and temporal features. Finally, a SVM classifier is employed to find the fall down behavior.

Single Shot Multibox Detector (SSD) provides a powerful framework for detecting objects using a single deep neural network. The detection framework is one of the top object detection algorithms in both accuracy and speed which processes a large set of object locations sampled across an image. However, this framework does not behave well for the task of pedestrian detection since the images in popular pedestrian datasets have multiple objects occlusion problem and contain lots of small objects. In this paper, we incorporate deconvolution and downsampling unit into the SSD framework allowing detection network to recycle feature maps learned from images. The enhanced performance was obtained by changing the structure of classifier network, e.g., by replacing VGGNet with DenseNet. The contribution of this paper is a one-stage approach to compose a single deep neural network for pedestrian detection task in real-time. This approach addresses the typical difficulty of detecting different scale pedestrian at only one layer by providing a novel channel fusion. To solve small objects problem, base network has been replaced with more powerful one. This approach outperforms competing one-single methods on standard Caltech pedestrian dataset benchmark. It is also faster than all the other methods.

Detecting suspicious behavior from high definition (HD) videos is always a complex and time-consuming process. To solve that problem, a fast suspicious behavior recognition method is proposed based on motion vectors. In this paper, the data format and decoding features of HD videos are analyzed. Then, the characteristics of suspicious activities and the ways of obtaining motion vectors directly from the video stream are concluded. Besides, the motion vectors are normalized by taking the reference frames into account. The feature vectors that display the inter-frame and intra-frame information of the region of interest are extracted. Gaussian radial basis function is employed as the kernel function of the support vector machines (SVM). It also realizes the detection and classification of suspicious behavior in HD videos. Finally, an extensive set of experiments are performed and this method is compared with some of the most recent approaches in the field using publicly available datasets as well as a new annotated human action dataset including actions performed in complex scenarios.

Finger-vein recognition is an increasingly promising biometric identification technology in terms of its high identification accuracy and prominent security performance. The main challenge faced by finger-vein recognition is the low recognition performance caused by segmentation error and local difference. To tackle this challenge, a finger-vein recognition method with modified binary tree (MBT) model is proposed in this paper. MBT model is used to describe the relationship and spatial structure of vein branches quantitatively. Based on the MBT model, four stages including rough selection, model correction, segment matching, and comprehensive judgment are presented to achieve a robust matching for finger-vein. Experiments demonstrate that the proposed method can boost the performance of finger-vein recognition that is degraded by segmentation error and local difference. While maintaining low complexity, the proposed method achieves 0.12 % equal error rate in the introduced dataset with 8,100 finger-vein images from 150 participants, which outperforms the state-of-the-art methods.

This paper presents a new human targets detection algorithm for video surveillance scenes. We propose a new approach for estimation of the existence of human targets, which includes multiple modes. Based on the detection structure, different features of human targets can be extracted. There are mainly two folds of advantage. First, the system integrates the time filtration and proportion of reliability weight, in order to intelligently merge and reinforce the results of different kinds of sensors. Second, it uses a short time video analysis, which can detect the human face by a Hough Transform, in order to confirm human targets obtained by the first step. Furthermore, before face detection, we use the gray similarity for local region clustering from a whole video frame, by which the range of Hough Transform can be reduced. By these processes, the accuracy of human targets detection can be increased effectively, and the video surveillance system can be switched on regions where human bodies are detected. Therefore, it can save system resources, and enhance the whole efficiency.

3D finger vein representation is the premise of 3D finger vein authentication. In this paper, a data organization model called liana model is presented for 3D finger vein representation. Firstly, four strategies including caliber uniformity, bine classification, node partition and spread restriction are proposed to covert the complex structure of 3D finger vein into a simple binary tree structure. Then, the liana model is developed based on the binary tree. Finally, the elements, attributes and mathematical descriptions of the model are described in detail. The new model is helpful for 3D finger vein representation and 3D features extraction.

This article discusses a new algorithm for vein matching based on log-polar transform to address problems that occur with the changing of finger position and from differences between imaging devices for current vein matching algorithms. The new algorithm first extracts the feature area, which contains enough characteristics for image matching, depending on the structure of the finger vein ridge alignment. It then calculates the degree of similarity between the log-polar transform results of the model image feature areas and the sample image, and finally analyzes the result by the degree of similarity and the relationship of relative positions between feature areas. Experiments show that the algorithm is robust for rotating and zooming images of the finger vein.

This paper presents a new optimized method for wireless sensor network, which can be used in video surveillance systems. The new algorithm involves two steps. In the first step, the system integrates the time filtration and proportion of reliability weight, in order to intelligently merge and reinforce the results of kinds of sensors. In the second step, it uses compressive sensing to confirm the alarm sensors. By these processes, the accuracy of target detection can be increased effectively, and the video surveillance system can only be switched on a region where some targets are detected. Therefore, it can save system resources, and enhance overall efficiency.

Face detection is the first step of vision-based driver fatigue detection method. Traditional face detection methods have problems of high false-detection rates and long detection times. A space-time restrained Adaboost method is presented in this paper that resolves these problems. Firstly, the possible position of a driver’s face in a video frame is measured relative to the previous frame. Secondly, a space-time restriction strategy is designed to restrain the detection window and scale of the Adaboost method to reduce time consumption and false-detection of face detection. Finally, a face knowledge restriction strategy is designed to confirm that the faces detected by this Adaboost method. Experiments compare the methods and confirm that a driver’s face can be detected rapidly and precisely.

A joint template matching algorithm is proposed in this paper to reduce the high rate of miss-detection and false-alarm caused by the traditional template matching algorithm during the process of multi-object detection. The proposed algorithm can reduce the influence on each object by matching all objects together according to the correlation information among different objects. Moreover, the rate of miss-detection and false-alarm in the process of single-template matching is also reduced based on the algorithm. In this paper, firstly, joint template is created from the information of relative positions among different objects. Then, matching criterion according to normalized cross correlation is generated for multi-object matching. Finally, the proposed algorithm is applied to the detection of watermarks in bill. The experiments show that the proposed algorithm has lower miss-detection and false-alarm rate comparing to the traditional NCC algorithm during the process of multi-object detection.

Based on Wireless Mesh Network (WMN), combined with the clustering topology features of Wireless Sensor Network (WSN), the structure of WSN can be reconstructed. Where made the heavier traffic sensor nodes or cluster head nodes netting together, and regarded sensor nodes as WMN clients which access Internet through WMN backbone, then reconstructed those topology as wireless mesh sensor network (WMSN). After that, a multi-hop data transmission model was proposed in terms of queuing principle. Then, multi-hop transferring and listening schemes of different nodes in the model were studied, and the minimum effective listening time for different nodes sitting at different hop in WMSN were determined by incorporating with energy efficient. The study results show that the combination of WSN and WMN can achieve mesh wide hybrid network topology. WMSN multi-hop data transmission system can be modeled as a limit multi-service window M/M/n queuing system. Multi-hop data delivery mechanism which is built on queuing system can achieve dynamic listening adjustment of different nodes.

In this Paper, the applicable conditions of Strength Reduction Method (SRM) and Limit Equilibrium Method (LEM) were studied. The stabilities for one side slope with the height of 300 meters under two different schemes of excavation were researched by using LEM and finite difference SRM severally. Dynamic response of this side slope under blasting vibration was also studied by FLAC numerical simulation. The research results show that the plastic failure and the plastic flow of side slope can be simulated more effectively and accurately by FLAC than by LEM; the mining blasting vibration in July 15 2010 has little influence on the stabilities of side slope, but this blasting would enlarge the instantaneous acceleration of the top of the side slope in a ratio of 10.8%.

In this paper, we propose a fast and robust algorithm for fighting behavior detection based on Motion Vectors (MV), in order to solve the problem of low speed and weak robustness in traditional fighting behavior detection. Firstly, we analyze the characteristics of fighting scenes and activities, and then use motion estimation algorithm based on block-matching to calculate MV of motion regions. Secondly, we extract features from magnitudes and directions of MV, and normalize these features by using Joint Gaussian Membership Function, and then fuse these features by using weighted arithmetic average method. Finally, we present the conception of Average Maximum Violence Index (AMVI) to judge the fighting behavior in surveillance scenes. Experiments show that the new algorithm achieves high speed and strong robustness for fighting behavior detection in surveillance scenes.

Vehicle license plate location is the key of the vehicle plate recognition system. This paper proposes a new method that sufficiently utilizes the color feature such as fixed color collocation, color centers and distribution in the plate region. In this method, firstly, the input image is convolved with 1-D Gaussian filter, and the regions which have the structure and texture are similar to the license plate are extracted. Then improved fuzzy C-means (FCM) is used to color clustering in the RGB color space. Finally, according to the clustering center and color distribution, the plate region was chosen from the candidate regions. The experimental results of more than 250 images collected show that the method is more accurate and robust to environment variation, and the rate of location is above 98%.

Large terrain dataset visualization is very difficult inflight simulation, because the size of terrain dataset is orders of magnitude larger than the main memory of system in flight simulation. The dataset are usually stored out-ofcore. The relatively slow access speed of out-ofcore makes data retrieval the bottleneck of the terrain visualization in flight simulation. The previous works on large terrain dataset visualization focus on two aspects: simplifying scene and organizing data. They always organize the terrain dataset on the disks, then load them into main memory part by part and render them efficiently. However, organization of terrain data on disk is quite difficult because the error, the triangulation dependency and the spatial location of each vertex all need to be considered, and the frame deadlock or skip phenomenon will occur probably. This paper proposes a new method for large dataset loading and rending in flight simulation. This method presents a new concept called sub-dataset, which are the all data to render the terrain in aflame. By loading the sub-dataset into main memory and real-time render them, we can realize the large terrain dataset visualization in flight simulation. Virtually, the terrain dataset refer to the longitude and latitude (L/L) coordinates of the flight position. By means of the index of L/L information, we can find the sub-dataset we need to load and render. The new method is very simple, and it costs much less time, provides visual continuity, and obtains fast visualization and animation of large scale terrain inflight simulation.

In this paper, an algorithm for smoke region of focus (ROF) detection based on surveillance video is proposed in order to solve the problem of limited application in scenes range and imaging environment of the traditional smoke detection algorithm. The frog vision perception model is used in this algorithm. First the suspect regions are detected, and then the static and the dynamic features of the regions are chosen for the smoke ROF detection. Experimental results show that the algorithm is efficient and significant for improving the operational rate of the detection.