Adaptive Gaussian Mixture Model Research Articles

Background Subtraction in Dynamic Environment based on Modified Adaptive GMM with TTD for Moving Object Detection

Background subtraction is the first processing stage in video surveillance. It is a general term for a process which aims to separate foreground objects from a background. The goal is to construct and maintain a statistical representation of the scene that the camera sees. The output of background subtraction will be an input to a higher-level process. Background subtraction under dynamic environment in the video sequences is one such complex task. It is an important research topic in image analysis and computer vision domains. This work deals background modeling based on modified adaptive Gaussian mixture model (GMM) with three temporal differencing (TTD) method in dynamic environment. The results of background subtraction on several sequences in various testing environments show that the proposed method is efficient and robust for the dynamic environment and achieves good accuracy.

Real-Time Vehicle License Plate Detection Based on Background Subtraction and Cascade of Boosted Classifiers

License plate (LP) detection is the most imperative part of an automatic LP recognition (LPR) system. Typical LPR contains two steps, namely LP detection (LPD) and character recognition. In this paper, we propose an efficient Vehicle-to-LP detection framework which combines with an adaptive GMM (Gaussian Mixture Model) and a cascade of boosted classifiers to make a faster vehicle LP detector. To develop a background model by using a GMM is possible in the circumstance of a fixed camera and extracts the motions using background subtraction. Firstly, an adaptive GMM is used to find the region of interest (ROI) on which motion detectors are running to detect the vehicle area as blobs ROIs. Secondly, a cascade of boosted classifiers is executed on the blobs ROIs to detect a LP. The experimental results on our test video with the resolution of 720x576 show that the LPD rate of the proposed system is 99.14% and the average computational time is approximately 42ms.

Detection of Moving Objects Based on Mixture Gaussian Model

The paper proposes an improved adaptive Gaussian mixture model (GMM) approach with online EM algorithms for updating, which solves the video segmentation problems carried by busy environment and illumination change. Different learning rates are set for foreground district and background district respectively, which improves the convergence speed of background model. A shadow removal scheme is also introduced for extracting complete moving objects. It is based on brightness distortion and chromaticity distortion in RGB color space. Morphological filtering and connected components analysis algorithm are also introduced to process the result of background subtraction. The experiment results show that the improved GMM has good accuracy and high adaptability in video segmentation. It can extract a complete and clear moving object when it is incorporated with shadow removal.

Individual Identification Method for Group-Housed Pigs Based on Optimal Feature Extraction

In order to reduce the workload of farmers on individual identification of pigs, reduce interference to pigs, and raise automation and intelligent level of farms, a Matlab-based automatic classification system is set up. Firstly, top view video sequences are captured for group-housed pigs from pig barn, and the adaptive Gaussian mixture model based on prediction mechanism is used to detect the foreground objects. Then with morphological processing and hole filling, the targets are extracted as completely as possible from the background. Secondly, after building the sample image databases for each pig, color, texture and shape features are extracted and combined as the feature vector. Next, LLE (Locally Linear Embedding) algorithm is adopted to reduce the data dimension. Finally, hybrid kernel SVM (Support Vector Machine) classifier is designed to identify the pigs in test frames. Experimental results show that the recognition rate of pigs is 95.2%. For improving the recognition rate, decreasing cost and interference to pigs, the above method is proposed. Moreover, it can lay a foundation for future researches on group-housed pigs in behaviors, emotions, breathings and so on, which has good practical significance.

Moving Object Detection Based on the Fish

The detection and tracking of moving object is the important research of image analysis and understanding as well as in computer vision field, and have extensive application in the traffic monitoring, the military, industrial process control and medical research, but less application in the underwater monitoring of fish. In this paper, in order to be able to real-time detection of the fish in the digital video system moving target, proposed the fish moving target detection algorithm under a camera. With an improved background updating method of adaptive Gaussian mixture model, a method to detect the target fish based on Gaussian mixture model combined with edge detection operator.

An Automatic Method of Stereoscopic Video Synthesis from Monoscopic Video

This paper presents a novel depth-image-based rendering system for the generation of stereoscopic video from a monoscopic video sequence. The key point of the proposed system is that all the conversion procedures are accomplished automatically. First, we use Global Motion Estimation and an Adaptive Gaussian Mixture Model to segment the foreground and background. Second, different depth values are assigned to the foreground and background. Finally, image warping according to the depth map and stereoscopic video is synthesized. Experimental results show that the proposed method can effectively segment the foreground from the background and synthesize a satisfactory 3D effect, so the proposed approach is successful.

Collaborative Nodes Strategy for Target Tracking in Two-tier Wireless Multimedia Sensor Networks

Two-tier network structure is usually used to lighten the energy consumption in Wireless Multimedia Sensor Networks (WMSNs) for large scale surveillance application. Previous methods for target tracking in two-tier WMSNs mostly focused on reducing energy cost to prolong lifetime of the networks without considering the effect of visual target tracking. This paper presents a collaborative multi-tier nodes strategy for visual target tracking in WMSNs. In the proposed method, the target detection and tracking are carried out by the cooperation of lower-tier scalar sensor nodes and upper-tier camera nodes. Each activated camera node uses an adaptive Gaussian mixture model to extract moving target and Mean Shift algorithm for target tracking. A novel function measuring the value of utility of tracking as well as energy cost is used to select optimal camera node. Experimental results show that the method can effectively achieve tradeoff between performance and energy consumption in real scenes.

Health Degradation Detection and Monitoring of Lithium-Ion Battery Based on Adaptive Learning Method

Battery health degradation detection and monitoring are crucial to realize equipment's near-zero downtime and maximum productivity. A big challenge is how to construct an effective monitoring system that can consistently exemplify the degradation propagation of battery health over its whole life. This paper proposes an on-line method based on adaptive learning scheme for monitoring of state-of-health (SOH) of battery. An adaptive Gaussian mixture model (AGMM) is developed for online learning dynamic changes of battery health in its full life. A Bayesian-inference-based probability indication is developed for detection of new health states that are online modeled by deleting and adding of components in AGMM. Furthermore, health changes are quantified by an AGMM-based health index that measures the overlap between two density distributions approximated by the adapted GMM and the historic GMM, respectively. Research results of its application in a lithium-ion battery life test illustrate that AGMM-based SOH monitoring system is effective for detection and adaptive assessment of battery health degradation. This paper provides a guidance for the development of battery SOH monitoring model based on adaptive learning scheme without too much human intervention.

Adaptively Splitted GMM With Feedback Improvement for the Task of Background Subtraction

Per pixel adaptive Gaussian mixture models (GMMs) have become a popular choice for the detection of change in the video surveillance domain because of their ability to cope with many challenges characteristic for surveillance systems in real time with low memory requirements. Since their first introduction in the surveillance domain, GMM has been enhanced in many directions. In this paper, we present a study of some relevant GMM approaches and analyze their underlying assumptions and design decisions. Based on this paper, we show how these systems can be further improved by means of a variance controlling scheme and the incorporation of region analysis-based feedback. The proposed system has been thoroughly evaluated using the extensive data set of the IEEE Workshop on Change Detection, showing an outranking performance in comparison with state-of-the-art methods.

An optimisation of Gaussian mixture models for integer processing units

This paper investigates sub-integer implementations of the adaptive Gaussian mixture model (GMM) for background/foreground segmentation to allow the deployment of the method on low cost/low power processors that lack Floating Point Unit. We propose two novel integer computer arithmetic techniques to update Gaussian parameters. Specifically, the mean value and the variance of each Gaussian are updated by a redefined and generalized “round” operation that emulates the original updating rules for a large set of learning rates. Weights are represented by counters that are updated following stochastic rules to allow a wider range of learning rates and the weight trend is approximated by a line or a staircase. We demonstrate that the memory footprint and computational cost of GMM are significantly reduced, without significantly affecting the performance of background/foreground segmentation.

Adaptive scale fuzzy local Gaussian mixture model for brain MR image segmentation

The Gaussian mixture model (GMM) has been widely used in brain magnetic resonance (MR) image segmentation. However, due to the MR bias field effect, the implied stochastic assumption that the intensities of each tissue type are sampled from an identical distribution may not be valid. In this paper, we propose a novel adaptive scale fuzzy local GMM (AS-FLGMM) algorithm for accurate and robust brain MR image segmentation. We assume that the local image data within the neighborhood of each pixel follow the GMM, in which the difference of variance among Gaussian components can be ignored. Based on this assumption, we develop a local scale estimation method to adaptively calculate the variance in each distribution. The segmentation is then performed under the fuzzy clustering framework and the objective is defined as the integration of the weighted GMM energy of each pixel. The AS-FLGMM algorithm has been compared to five state-of-the-art segmentation approaches in both synthetic and clinical MR images. Our results show that the proposed algorithm can produce more accurate segmentation results and its performance is more robust to initialization.

Image Object Detection Algorithm Based on Improved Gaussian Mixture Model

Aiming at poor adaptability to illumination variation and single learning rate in traditional Gaussian mixture model, an improved moving object detection algorithm based on adaptive Gaussian mixture model is proposed in this paper, so as to achieve the goal of a self-adaptive background updating model. In this paper, we analyze the existed algorithms and put forward the method to make use of color histogram matching algorithm, through introduction of illumination variation factor and update-counter of model parameter and the components number with self-adaptive selection employed to adaptively adjust learning rate, in order to greatly reduce the computation time of the algorithm and improve the real-time performance. The experiment results show that the new method can effectively adapt the scene, and has more good expansibility, robustness and stability than traditional Gaussian mixture model.

Segmentation of sows in farrowing pens

The correct segmentation of a foreground object in video recordings is an important task for many surveillance systems. The development of an effective and practical algorithm to segment sows in grayscale video recordings captured under commercial production conditions is described. The segmentation algorithm combines a modified adaptive Gaussian mixture model for background subtraction with the boundaries of foreground objects, which is obtained by using dyadic wavelet transform. This algorithm can accurately extract the shapes of a sow under complex environments, such as dynamic background and illumination changes as well as motionless foreground objects. About 97% of the segmented binary images in the validation data sets can be used to track sow behaviours, such as position, orientation and movement. The experimental results demonstrate that the proposed algorithm is able to provide a basis to detect and classify the sow behaviours in farrowing pens.

Improved object detection method of adaptive Gaussian mixture model

The deficiency of Gaussian Mixture Model( GMM) is of high computation cost and cannot deal with the shadow and ghosting. An improved foreground detection algorithm based on GMM was proposed in this paper. By analyzing the stability of the background, intermittent or continuous frame updating was chosen to update the parameters of the GMM. It can efficiently reduce the runtime of the algorithm. In the background updating, the updating rate was associated with the weight and this made it change with the weight. The background pixels which appeared after the objects moving were set a larger updating rate. It can improve the stability of the background and solve the problem of ghosting phenomenon and the transformation of background and foreground. After objects detection, the algorithm eliminated the shadow based on the RGB color space distortion model and treated the result by Gauss pyramid filtering and morphological filtering. Through the whole process, a better contour was obtained. The experimental results show that this algorithm improves the calculation efficiency and accurately segments the foreground object.

Background subtraction using finite mixtures of asymmetric Gaussian distributions and shadow detection

Foreground segmentation of moving regions in image sequences is a fundamental step in many vision systems including automated video surveillance, human-machine interface, and optical motion capture. Many models have been introduced to deal with the problems of modeling the background and detecting the moving objects in the scene. One of the successful solutions to these problems is the use of the well-known adaptive Gaussian mixture model. However, this method suffers from some drawbacks. Modeling the background using the Gaussian mixture implies the assumption that the background and foreground distributions are Gaussians which is not always the case for most environments. In addition, it is unable to distinguish between moving shadows and moving objects. In this paper, we try to overcome these problem using a mixture of asymmetric Gaussians to enhance the robustness and flexibility of mixture modeling, and a shadow detection scheme to remove unwanted shadows from the scene. Furthermore, we apply this method to real image sequences of both indoor and outdoor scenes. The results of comparing our method to different state of the art background subtraction methods show the efficiency of our model for real-time segmentation.

Automatic node selection and target tracking in wireless camera sensor networks

Aiming at fulfilling the wide-area video surveillance, this paper presents a cooperative multi-camera target tracking method for wireless camera sensor networks. In the proposed method, target detection is carried out by single-node processing based on background subtraction, whereas target tracking is performed by senor nodes cooperation based on automatic node selection. The main contributions of the proposed method are summarized as follows. First, each camera node uses an adaptive Gaussian mixture model to extract moving targets and an unscented Kalman filter to solve target tracking. Second, the correspondence between the targets in different camera views is established by homography transformation of target positions. Third, a confidence measure based on the size of the detected target blob and the estimate uncertainty of tracking is defined to achieve optimal node selection. Experimental results show that the proposed method can effectively select camera node to implement the accurate tracking in real scenes.

Geometry-Based Object Association and Consistent Labeling in Multi-Camera Surveillance

This paper proposes a multi-camera surveillance framework based on multiple view geometry. We address the problem of object association and consistent labeling through exploring geometrical correspondences of objects, not only in sequential frames from a single camera view but also across multiple camera views. The cameras are geometrically related through joint combination of multi-camera calibration, ground plane homography constraint, and field-of-view lines. Object detection is implemented using an adaptive Gaussian mixture model, and thereafter the information obtained from different cameras is fused so that the same object shown in different views can be assigned a unique label. Meanwhile, a virtual top-view of ground plane is synthesized to explicitly display the corresponding location and label of each detected object within a designated area-of-interest.

Adaptive Moving Object Detection Algorithm Based on Background Subtraction and Motion Estimation

Moving object detection plays an important role in video surveillance and object tracking. With a stationary camera it is still a challenging problem to precisely extract moving object from real-time video with dynamic backgrounds. In this work, we develop an efficient moving object detection framework to tackle this problem. First, we utilize adaptive Gaussian mixture model to cope with steadily varying components of backgrounds. Second, inspired by the motion estimation method used in video coding, we propose a motion estimation technique to deal with unsteadily varying components of backgrounds. Exhaustive experimental evaluations on complex scenes show that the proposed method is fast and stable, achieving more than 15% improvement in accuracy compared over existing state-of-the-art algorithms. Keyword: Dynamic Background, Adaptive Gaussian Mixture Model, Background Subtraction, Motion Estimation

User Behavior Modeling based on Adaptive Gaussian Mixture Model

A remarkable technological development has been witnessed due to the recent advancements in the area of science and technology. This made the users to access the Internet and store the information retrieved in various databases at various servers across the globe, making World Wide Web as an information gateway. This technological growth has thrown a challenging situation to both the user and the business man. The various resources that are available through the Internet made the users to choose different alternatives which in turn makes it necessary for the businessman to bring out new strategies and alternatives so as to attract the users. In order to overcome these challenging situations, web usage mining brings out a solution to the business man by analyzing different user patterns that are available in the web. Many tools are available for this purpose but majority of the tools lag in including the complete details regarding the web log data .To overcome this disadvantage, in this paper a model based on Adaptive Gaussian Mixture Model, an extension of Gaussian Mixture Model (GMM) to interpret the user navigation behavior is brought out. The proposed model is applied on user traffic data taken from msnbc.com.

Extracting robust distribution using adaptive Gaussian Mixture Model and online feature selection

This paper presents a novel method to extract robust distribution using adaptive Gaussian Mixture Model (GMM) and online feature selection to improve the tracking performance. Traditional histogram-based tracking algorithms use all histogram bins to track the object, and some recent work attempt selecting important bins by directly comparing the corresponding probability value between the foreground and background histogram distribution and give higher weight to those important bins. However, the selected bins may not be discriminative enough. In this paper, we use an adaptive GMM to model histogram and select good Gaussian components corresponding to the discriminative bins and the stable part of the foreground, which form the robust distribution. Given a set of seed, for each feature, the foreground and surrounding background are modeled by an adaptive GMM, respectively. Then, we use the Gaussian Component Separability measure to find good components, to extract the robust distribution of the foreground and to evaluate the separability of each feature. Finally, after ranking features based on their separability, the K most discriminative features are used to generate a weight image and the CAMSHIFT algorithm is employed to locate the object. Experiments show that the proposed method can extract the robust distribution and improve the tracking performance.