Bidirectional Background Modeling for Video Surveillance

Traditional background model methods often require complicated computations, and are sensitive to illumination and shadow. In this paper, we propose a block-based background modeling method, and use our proposed method to combine color and texture characteristics. Suppression and relaxation are the two key strategies to resist illumination changes and shadow disturbance. The proposed method is quite efficient and is capable of resisting illumination changes. Experimental results show that our method is suitable for real-word scenes and real-time applications.

Diversity Sampling Based Kernel Density Estimation for Background Modeling

A novel diversity-sampling based nonparametric multi-modal background model is proposed. Using the samples having more popular and various intensity values in the training sequence, a nonparametric model is built for background subtraction. According to the related intensifies, different weights are given to the distinct samples in kernel density estimation. This avoids repeated computation using all samples, and makes computation more efficient in the evaluation phase. Experimental results show the validity of the diversity- sampling scheme and robustness of the proposed model in moving objects segmentation. The proposed algorithm can be used in outdoor surveillance systems.

Neural network based method for background modeling and detecting moving objects

This paper proposes a novel method, primarily based on the fuzzy adaptive resonance theory （ART） neural network with forgetting procedure, for moving object detection and background modeling in natural scenes. With the ability, inheriting from the ART neural network, of extracting patterns from arbitrary sequences, the background model based on the proposed method can learn new scenes quickly and accurately. To guarantee that a long-life model can derived from the proposed mothed, a forgetting procedure is employed to find the neuron that needs to be discarded and reconstructed, and the finding procedure is based on a neural network which can find the extreme value quickly. The results of a suite of quantitative and qualitative experiments conducted verify that for processes of modeling background and detecting moving objects our method is more effective than five other proven methods with which it is compared.

A temporal-spatial background modeling of dynamic scenes

Moving object detection in dynamic scenes is a basic task in a surveillance system for sensor data collection. In this paper, we present a powerful back- ground subtraction algorithm called Gaussian-kernel density estimator （G-KDE） that improves the accuracy and reduces the computational load. The main innovation is that we divide the changes of background into continuous and stable changes to deal with dynamic scenes and moving objects that first merge into the background, and separately model background using both KDE model and Gaussian models. To get a temporal- spatial background model, the sample selection is based on the concept of region average at the update stage. In the detection stage, neighborhood information content （NIC） is implemented which suppresses the false detection due to small and un-modeled movements in the scene. The experimental results which are generated on three separate sequences indicate that this method is well suited for precise detection of moving objects in complex scenes and it can be efficiently used in various detection systems.

Adaptive learning algorithm based on mixture Gaussian background

The key problem of the adaptive mixture background model is that the parameters can adaptively change according to the input data. To address the problem, a new method is proposed. Firstly, the recursive equations are inferred based on the maximum likelihood rule. Secondly, the forgetting factor and learning rate factor are redefined, and their still more general formulations are obtained by analyzing their practical functions. Lastly, the convergence of the proposed algorithm is proved to enable the estimation converge to a local maximum of the data likelihood function according to the stochastic approximation theory. The experiments show that the proposed learning algorithm excels the formers both in converging rate and accuracy.

A numerical model study on multi-species harmful algal blooms coupled with background ecological fields

Based on systematized physical, chemical, and biological modules, a multi-species harmful algal bloom （HAB） model coupled with background ecological fields was established. This model schematically embod-ied that HAB causative algal species and the background ecological system, quantified as total biomass, were significantly different in terms of the chemical and biological processes during a HAB while the inter-action between the two was present. The model also included a competition and interaction mechanism between the HAB algal species or populations. The Droop equation was optimized by considering tempera-ture, salinity, and suspended material impact factors in the parameterization of algal growth rate with the nutrient threshold. Two HAB processes in the springs of 2004 and 2005 were simulated using this model. Both simulation results showed consistent trends with corresponding HAB processes observed in the East China Sea, which indicated the rationality of the model. This study made certain progress in modeling HABs, which has great application potential for HAB diagnosis, prediction, and prevention.

Adaptive learning rate GMM for moving object detection in outdoor surveillance for sudden illumination changes

A dynamic learning rate Gaussian mixture model（GMM）algorithm is proposed to deal with the problem of slow adaption of GMM in the case of moving object detection in the outdoor surveillance,especially in the presence of sudden illumination changes.The GMM is mostly used for detecting objects in complex scenes for intelligent monitoring systems.To solve this problem,a mixture Gaussian model has been built for each pixel in the video frame,and according to the scene change from the frame difference,the learning rate of GMM can be dynamically adjusted.The experiments show that the proposed method gives good results with an adaptive GMM learning rate when we compare it with GMM method with a fixed learning rate.The method was tested on a certain dataset,and tests in the case of sudden natural light changes show that our method has a better accuracy and lower false alarm rate.

Real-time moving object detection for video monitoring systems

Moving object detection is one of the challenging problems in video monitoring systems, especially when the illumination changes and shadow exists. Amethod for real-time moving object detection is described. Anew background model is proposed to handle the illumination varition problem. With optical flow technology and background subtraction, a moving object is extracted quickly and accurately. An effective shadow elimination algorithm based on color features is used to refine the moving obj ects. Experimental results demonstrate that the proposed method can update the background exactly and quickly along with the varition of illumination, and the shadow can be eliminated effectively. The proposed algorithm is a real-time one which the foundation for further object recognition and understanding of video mum＇toting systems.

Crustal Electrical Structure and Deep Metallogenic Potential in Northern Wuyi Area(South China),based on Magnetotelluric Data

The northern Wuyi area,which is located in the northern Wuyi metallogenic belt,has superior mineralization conditions.The Pingxiang-Guangfeng-Jiangshan-Shaoxing fault(PSF)extends across the whole region regardless of whether or how the PSF relates to the near-surface mineralization.We carried out an MT survey in the region and obtained a reliable 2D model of the crustal electrical structure to a depth of 30 km.In the resistivity model,we inferred that a continuous high conductivity belt that ranges from the shallow to deep crust is a part of the PSF.Then,we estimated the fluid content and pressure gradient to identify the deep sources of fluid as well as its pattern of motion pattern.Finally,we proposed a model for the deep metallogenic migration processes that combines geological data,fluid content data,pressure gradient data,and the subsurface resistivity model.The model analysis showed that the Jiangnan orogenic belt and the Cathaysia block formed the PSF during the process of com.The deep fluid migrated upward through the PSF to the shallow crust.Therefore,we believe that the PSF is an ore-forming fluid migration channel and that it laid the material basis for large-scale mineralization in the shallow crust.

High efficient moving object extraction and classification in traffic video surveillance

Moving object extraction and classification are important problems in automated video surveillance systems. A background model based on region segmentation is proposed. An adaptive single Gaussian background model is used in the stable region with gradual changes, and a nonparametric model is used in the variable region with jumping changes. A generalized agglomerative scheme is used to merge the pixels in the variable region and fill in the small interspaces. A two-threshold sequential algorithmic scheme is used to group the background samples of the variable region into distinct Gaussian distributions to accelerate the kernel density computation speed of the nonparametric model. In the feature-based object classification phase, the surveillance scene is first partitioned according to the road boundaries of different traffic directions and then re-segmented according to their scene localities. The method improves the discriminability of the features in each partition. AdaBoost method is applied to evaluate the relative importance of the features in each partition respectively and distinguish whether an object is a vehicle, a single human, a human group, or a bike. Experimental results show that the proposed method achieves higher performance in comparison with the existing method.

EMOTIONAL SPEECH RECOGNITION BASED ON SVM WITH GMM SUPERVECTOR

Emotion recognition from speech is an important field of research in human computer interaction. In this letter the framework of Support Vector Machines (SVM) with Gaussian Mixture Model (GMM) supervector is introduced for emotional speech recognition. Because of the importance of variance in reflecting the distribution of speech, the normalized mean vectors potential to exploit the information from the variance are adopted to form the GMM supervector. Comparative experiments from five aspects are conducted to study their corresponding effect to system performance. The experiment results, which indicate that the influence of number of mixtures is strong as well as influence of duration is weak, provide basis for the train set selection of Universal Background Model (UBM).

Surveillance Video Defogging Algorithm Optimized by Background Extraction

To reduce the flicker artifacts caused by video defogging,a surveillance video defogging algorithm based on the background extraction and consistent constraints is proposed.First,an inter frame consistency constraint is constructed and applied to background modeling.Second,the extracted background is defogged with an improved static defogging approach.Third,the foreground is extracted using the extracted background and further defogged using constraints of the consistency between the foreground and background.Experimental results show that our algorithm can remove fog effectively and preserve the temporal coherence well.

Emission inventory evaluation using observations of regional atmospheric background stations of China

Any accurate simulation of regional air quality by numerical models entails accurate and up-to-date emissions data for that region.The INTEX-B2006 （I06）,one of the newest emission inventories recently popularly used in China and East Asia,has been assessed using the Community Multiscale Air Quality model and observations from regional atmospheric background stations of China.Comparisons of the model results with the observations for the species SO2,NO 2,O 3 and CO from the three regional atmospheric background stations of Shangdianzi,Longfengshan and Linan show that the model can basically capture the temporal characteristics of observations such as the monthly,seasonal and diurnal variance trends.Compared to the other three species,the simulated CO values were grossly underestimated by about two-third or one-half of the observed values,related to the uncertainty in CO emissions.Compared to the other two stations,Shangdianzi had poorer simulations,especially for SO2 and CO,which partly resulted from the site location close to local emission sources from the Beijing area;and the regional inventory used was not capable of capturing the influencing factors of strong regional sources on stations.Generally,the fact that summer gave poor simulation,especially for SO2 and O 3,might partly relate to poor simulations of meteorological fields such as temperature and wind.

Research of whispered speech vocal tract system conversion based on universal background model and effective Gaussian components

Directing to the weakness of the present fixed values mapping methods （method_F）, a vocal tract system conversion method based on the universal background model （UBM） is proposed for improving the performance of the speech conversion system from Chinese whis- pered speech to normal speech. For the numerous components of UBM, the errors produced by the acoustical probability density statistical model can＇t be ignored. Thus an effective Gaus- sian mixture components chosen method based on the posterior probability summation of the minimum spectral distortion is developed to optimizing the system performance. The proposed method （method_U） is analyzed and compared using the performance index （PI） based on Itakura-Saito spectral distortion measure. It is shown experimentally that the performance of method_U is more stability for different speakers and different phonemes than that of method_F. The average PI of method_U is better than method_F. It is shown that by selecting effective Gaussian mixture components, the PI of method_U can be further improved 5.11%. Subjective auditory tests also show that the proposed method can improve the definition and intelligibility of conversion speech.

A primary-secondary background model with sliding window PCA algorithm

Rain and snow seriously degrade outdoor video quality.In this work,a primary-secondary background model for removal of rain and snow is built.First,we analyze video noise and use a sliding window sequence principal component analysis de-nosing algorithm to reduce white noise in the video.Next,we apply the Gaussian mixture model(GMM)to model the video and segment all foreground objects primarily.After that,we calculate von Mises distribution of the velocity vectors and ratio of the overlapped region with referring to the result of the primary segmentation and extract the interesting object.Finally,rain and snow streaks are inpainted using the background to improve the quality of the video data.Experiments show that the proposed method can effectively suppress noise and extract interesting targets.

Spatially Adaptive Subsampling for Motion Detection

Many video surveillance applications rely on efficient motion detection. However, the algorithms are usually costly since they compute a background model at every pixel of the frame. This paper shows that, in the case of a planar scene with a fixed calibrated camera, a set of pixels can be selected to compute the background model while ignoring the other pixels for accurate but less costly motion detection. The cali- bration is used to first define a volume of interest in the real world and to project the volume of interest onto the image, and to define a spatial adaptive subsampling of this region of interest with a subsampling density that depends on the camera distance. Indeed, farther objects need to be analyzed with more precision than closer objects. Tests on many video sequences have integrated this adaptive subsampling to various motion detection techniques.

Exemplar-based video inpainting with large patches

Inpainting is the process of reconstructing damaged regions of images and video frames.This study deals with weaknesses of the current video inpainting techniques,when an object is totally damaged,and a framework for video inpainting is proposed.Using this framework,the moving object is separated from the background.A large mosaic image is constructed using the moving object and then a patch-based method with large patches is used to fill holes.In each frame,the inpainted foreground is obtained by placing the object in its location.Missing areas of the stationary background are also filled separately and the final video is produced by composing the inpainted background and object frames.Results for three video sequences with an occluded object show that this approach represents the object in the missing region better than other approaches.

Individual pig object detection algorithm based on Gaussian mixture model

The background models are crucially important for the object extraction for moving objects detection in a video.The Gaussian mixture model(GMM)is one of popular methods in the background models.Gaussian mixture model which applied to the pig target detection has some shortcomings such as low efficiency of algorithm,misjudgment points and ghosts.This study proposed an improved algorithm based on adaptive Gaussian mixture model,to overcome the deficiencies of the traditional Gaussian mixture model in pig object detection.Based on Gaussian mixture background model,this paper introduced two new parameters of video frames m and T0.The Gaussian distribution was scanned once every m frames,the excessive Gaussian distribution was deleted to improve the convergence speed of the model.Meanwhile,using different learning rates to suppress ghosts,a higher decreasing learning rate was adopted to accelerate the background modeling before T_(0),the background model would become stable as the time continued and a smaller learning rate could be used.In order to maintain a stable background and reduce noise interference,a fixed learning rate after T_(0) was used.Results of experiments indicated that this algorithm could quickly build the initial background model,detect the moving target pigs,and extract the complete contours of the target pigs’.The algorithm is characterized by good robustness and adaptability.

A hierarchical clustering of features approach for vehicle tracking in traffic environments

Purpose-Developing algorithms for automated detection and tracking of multiple objects is one challenge in the field of object tracking.Especially in a traffic video monitoring system,vehicle detection is an essential and challenging task.In the previous studies,many vehicle detection methods have been presented.These proposed approaches mostly used either motion information or characteristic information to detect vehicles.Although these methods are effective in detecting vehicles,their detection accuracy still needs to be improved.Moreover,the headlights and windshields,which are used as the vehicle features for detection in these methods,are easily obscured in some traffic conditions.The paper aims to discuss these issues.Design/methodology/approach-First,each frame will be captured from a video sequence and then the background subtraction is performed by using the Mixture-of-Gaussians background model.Next,the Shi-Tomasi corner detection method is employed to extract the feature points from objects of interest in each foreground scene and the hierarchical clustering approach is then applied to cluster and form them into feature blocks.These feature blocks will be used to track the moving objects frame by frame.Findings-Using the proposed method,it is possible to detect the vehicles in both day-time and night-time scenarios with a 95 percent accuracy rate and can cope with irrelevant movement(waving trees),which has to be deemed as background.In addition,the proposed method is able to deal with different vehicle shapes such as cars,vans,and motorcycles.Originality/value-This paper presents a hierarchical clustering of features approach for multiple vehicles tracking in traffic environments to improve the capability of detection and tracking in case that the vehicle features are obscured in some traffic conditions.

Tail Distortion Risk Measure for Portfolio with Multivariate Regularly Variation

For the multiplicative background risk model,a distortion-type risk measure is used to measure the tail risk of the portfolio under a scenario probability measure with multivariate regular variation.In this paper,we investigate the tail asymptotics of the portfolio loss ∑_(i=1)^(d)R_(i)S,where the stand-alone risk vector R=(R_(1),...,R_(d))follows a multivariate regular variation and is independent of the background risk factor S.An explicit asymptotic formula is established for the tail distortion risk measure,and an example is given to illustrate our obtained results.