Background Modeling Algorithm Research Articles

Video big data in smart city: Background construction and optimization for surveillance video processing

Transforming infrastructures, buildings and services with the sensed data from the Internet of Things (IoT) technique has drawn wide attention. Enormous video data from city surveillance cameras poses huge challenges of transmission, storage and analysis, which necessitates new video compression technologies. The fusion of video data generated from smart city could be used to support city management and urban policy. Based on the specific characteristics of surveillance video, which are successive pictures have very strong correlations and each picture can be divided into background and foreground, this work proposes a block-level background modeling (BBM) algorithm to support long-term reference structure for efficient surveillance video coding. A rate–distortion optimization for surveillance source (SRDO) algorithm is also developed to improve the coding performance. Experimental results show that the proposed BBM and SRDO can significantly improve the compression performance, which can effectively support diverse video applications in smart city.

A Multiframes Integration Object Detection Algorithm Based on Time-Domain and Space-Domain

In order to overcome the disadvantages of the commonly used object detection algorithm, this paper proposed a multiframes integration object detection algorithm based on time-domain and space-domain (MFITS). At first, the consecutive multiframes were observed in time-domain. Then the horizontal and vertical four-direction extension neighborhood of each target pixel were selected in space-domain. Transverse and longitudinal sections were formed by fusing of the time-domain and space-domain. The mean and standard deviation of the pixels in transverse and longitudinal section were calculated. We also added an improved median filter to generate a new pixel in each target pixel position, eventually to generate a new image. This method is not only to overcome the RPAC method affected by lights, shadows, and noise, but also to reserve the object information to the maximum compared with the interframe difference method and overcome the difficulty in dealing with the high frequency noise compared with the adaptive background modeling algorithm. The experiment results showed that the proposed algorithm reserved the motion object information well and removed the background to the maximum.

A Background Modeling and Foreground Detection Algorithm Using Scaling Coefficients Defined With a Color Model Called Lightness-Red-Green-Blue.

This paper presents an algorithm for background modeling and foreground detection that uses scaling coefficients, which are defined with a new color model called lightness-red-green-blue (LRGB). They are employed to compare two images by finding pixels with scaled lightness. Three backgrounds are used: 1) verified background with pixels that are considered as background; 2) testing background with pixels that are tested several times to check if they belong to the background; and 3) final background that is a combination of the testing and verified background (the testing background is used in places, where the verified background is not defined). If a testing background pixel matches pixels from previous frames (the match is tested using scaling coefficients), it is copied to the verified background, otherwise the pixel is set as the weighted average of the corresponding pixels of the last input images. After the background is computed, foreground objects are detected by using the scaling coefficients and additional criteria. The algorithm was evaluated using the SABS data set, Wallflower data set and a subset of the CDnet 2014 data set. The average F measure and sensitivity with the SABS Data set were 0.7109 and 0.8725, respectively. In the Wallflower data set, the total number of errors was 5280 and the total F-measure was 0.9089. In the CDnet 2014 data set, the F-measure for the baseline test case was 0.8887 and for the shadow test case was 0.8300.

Background modelling of diffraction data in the presence of ice rings.

An algorithm for modelling the background for each Bragg reflection in a series of X-ray diffraction images containing Debye-Scherrer diffraction from ice in the sample is presented. The method involves the use of a global background model which is generated from the complete X-ray diffraction data set. Fitting of this model to the background pixels is then performed for each reflection independently. The algorithm uses a static background model that does not vary over the course of the scan. The greatest improvement can be expected for data where ice rings are present throughout the data set and the local background shape at the size of a spot on the detector does not exhibit large time-dependent variation. However, the algorithm has been applied to data sets whose background showed large pixel variations (variance/mean > 2) and has been shown to improve the results of processing for these data sets. It is shown that the use of a simple flat-background model as in traditional integration programs causes systematic bias in the background determination at ice-ring resolutions, resulting in an overestimation of reflection intensities at the peaks of the ice rings and an underestimation of reflection intensities either side of the ice ring. The new global background-model algorithm presented here corrects for this bias, resulting in a noticeable improvement in R factors following refinement.

A system on chip-based real-time tracking system for amphibious spherical robots

Aiming at vision applications of our amphibious spherical robot, a real-time detection and tracking system adopting Gaussian background model and compressive tracking algorithm was designed and implemented in this article. Considering the narrow load space, the limited power resource and the specialized application scenarios of the robot, a heterogeneous computing architecture combining advanced Reduced Instruction-Set Computer (RISC) machine and field programmable gate array was proposed on the basis of Zynq-7000 system on chip.Under the architecture, main parts of the vision algorithms were implemented as software programs running on the advanced RISC machine-Linux subsystem. And customized image accelerators were deployed on the field programmable gate array subsystem to speed up the time-consuming processes of visual algorithms. Moreover, dynamic reconfiguration was used to switch accelerators online for reducing resource consumption and improving system adaptability. The word length of accelerators was optimized with simulated annealing algorithm to make a compromise between calculation accuracy and resource consumption. Experimental results confirmed the feasibility of the proposed architecture. The single board tracking system was able to provide an image processing rate of up to 89.2 frames per second at the resolution of 320 × 240, which could meet future demands of our robot in biological monitoring and multi-target tracking.

Temporal weighted learning model for background estimation with an automatic re-initialization stage and adaptive parameters update

Abstarct Background initialization and background update are two important stages considered in the design of most background modeling algorithms. Commonly, these algorithms implement strategies in which their parameters have a very high adaptability in the background initialization stage in order to learn all the variations of the background. Contrary, in the background update phase, these parameters adapt slowly, in most cases with an exponential decay. This paper presents the BE-AAPSA method which automatically determines if the background initialization and the background update need to be re–initialized. Re -initialization is triggered if the video scene presents high variations, allowing the background to be defined more accurately. BE-AAPSA is based on a previously developed system, where two adaptive background models based on weight arrays with temporal learning mechanism identify dynamic objects within a video scene. The system implements four independent modules to treat the different factors that affect a correct definition of the dynamic object. In BE-AAPSA, the objective is to create a robust background estimation model where the learning rates for each pixel are calculated according to the results of the two adaptive weight arrays and the module where the video is classified. This approach allows handling different strategies to update learning rates at a pixel resolution. BE-AAPSA is validated with the SBI and SBMnet video databases and with a video created by concatenating scenes of the video categories presented in the CDnet 2014 database. According to the findings, BE-AAPSA produced highly accurate results with SBI and SBMnet and surpassed state-of-the-art methods with the CDnet video. These results demonstrate the importance of using an automatic re-initialization scheme in the background initialization and background update stages when the video scene presents a major change or involves jittering. Furthermore, it shows the benefits of handling in separate modules the analysis of the background estimation results.

Parallelizing Multimodal Background Modeling on a Low-Power Integrated GPU

Background modeling techniques for embedded computer vision applications must balance accuracy, speed, and power. Basic background modeling techniques run quickly, but their accuracy is not sufficient for computer vision problems involving dynamic background. In contrast, adaptive background modeling techniques are more robust, but run more slowly. Due to its high inherent fine-grain parallelism, robust adaptive background modeling has been implemented on GPUs with significant performance improvements over CPUs. However, these implementations are infeasible in embedded applications due to the high power ratings of the targeted general-purpose GPU platforms. This paper focuses on exploiting fine-grain data parallelism and optimizing memory access patterns to target a low-cost adaptive background modeling algorithm multimodal mean (MMM) to a low-power GPU with thermal design power (TDP) of only 12 watts. The algorithm has comparable accuracy with the Gaussian mixture model (GMM) algorithm, but less computational and memory cost. It achieves a frame rate of 392 fps with a full VGA resolution (640x480) frame on the low-power integrated GPU NVIDIA ION. This is a 20x speed-up of the MMM algorithm compared to the embedded CPU platform Intel Atom of comparable TDP. In addition, the MMM algorithm attains a 5-6x speed up over the GMM implementation on the ION GPU platform.

Background Modeling Algorithm Based on Transitions Intensities

Moving objects detection in a scene from a video sequence is a fundamental and crucial problem in many vision systems such as video surveillance or the traffic monitoring. Background modeling is often used to detect moving objects which assumes that the moving objects take less time in the scene with respect to the background. Indeed, this assumption cannot be valid in case of a scene with fast moving objects. To overcome this problem, in this paper we propose a novel background modeling approach based on transitions intensities. In other words, we will focus on intensity changes (transitions) of pixels during a scene instead of using duration information that a pixel pass in a state without any notable change in its intensity. The normalized RGB color is used to describe the appearance under change lighting conditions, and transition intensity is introduced to capture the dynamicity of objects over time. Experimental results on different video sequences have been presented to approve the effectiveness of the proposed algorithm.

The Research of Moving Target Detection and Tracking System Based on DM6437

In order to enhance the rapidity and the accuracy of moving target detection and tracking, and improve the speed of the algorithm on the DSP (digital signal processor), an active visual tracking system was designed based on the gaussian mixture background model and Meanshift algorithm on DM6437. The system use the VLIB library developed by TI, and through the method of gaussian mixture background model to detect the moving objects and use the Meanshift tracking algorithm based on color features to track the target in RGB space. Finally, the system is tested on the hardware platform, and the system is verified to be quickness and accuracy.

Object detection based on randomized background modeling algorithm

运动目标检测是智能视频监控系统中的重要步骤和前提。提出了一种基于随机背景建模的非参数化建模算法，对场景中运动目标进行快速提取跟踪。在初始化阶段，从当前像素的邻域中随机抽取样本值作为背景模型；在模型更新阶段，引入了随机更新策略和背景传播机制，能够较好地抑制环境噪声；在后处理阶段，给出了一种基于积分图的前景滤波优化方法，进一步滤除噪声和填充前景空洞。实验结果表明，在复杂场景条件下，算法的目标检测性能明显优于其他几种同类算法，能够较好地抑制噪声干扰，具有较高的检测正确率。对于360288像素的测试视频，算法的计算速度高达120 f/s，完全可以满足实时应用。

Fast image foreground object segmentation based on block texture for embedded system implementation

In this paper, a novel block-based foreground object detection method based on block texture is presented. It can significantly reduce the memory usage when constructing the background model in dynamic scenes. The proposed background model and detection algorithm are suitable for implementing on embedded system platforms with resource limitations. The experimental results of processing benchmark videos show that our method has outcomes that are very close to ground truth segmentation. In addition, the proposed method requires approximately 23.97% less memory than the latest algorithms. Finally, the proposed approach is implemented on an embedded system platform. The processing speed can achieve a real-time rate of at least 20 fps, which is an improvement of 17.64% as compared to the latest algorithms.

Mismatched feature detection with finer granularity for emotional speaker recognition

The shapes of speakers’ vocal organs change under their different emotional states, which leads to the deviation of the emotional acoustic space of short-time features from the neutral acoustic space and thereby the degradation of the speaker recognition performance. Features deviating greatly from the neutral acoustic space are considered as mismatched features, and they negatively affect speaker recognition systems. Emotion variation produces different feature deformations for different phonemes, so it is reasonable to build a finer model to detect mismatched features under each phoneme. However, given the difficulty of phoneme recognition, three sorts of acoustic class recognition—phoneme classes, Gaussian mixture model (GMM) tokenizer, and probabilistic GMM tokenizer—are proposed to replace phoneme recognition. We propose feature pruning and feature regulation methods to process the mismatched features to improve speaker recognition performance. As for the feature regulation method, a strategy of maximizing the between-class distance and minimizing the within-class distance is adopted to train the transformation matrix to regulate the mismatched features. Experiments conducted on the Mandarin affective speech corpus (MASC) show that our feature pruning and feature regulation methods increase the identification rate (IR) by 3.64% and 6.77%, compared with the baseline GMM-UBM (universal background model) algorithm. Also, corresponding IR increases of 2.09% and 3.32% can be obtained with our methods when applied to the state-of-the-art algorithm i-vector.

Video-based Height Measurements of Multiple Moving Objects

This paper presents a novel video metrology approach based on robust tracking. From videos acquired by an uncalibrated stationary camera, the foreground likelihood map is obtained by using the Codebook background modeling algorithm, and the multiple moving objects are tracked by a combined tracking algorithm. Then, we compute vanishing line of the ground plane and the vertical vanishing point of the scene, and extract the head feature points and the feet feature points in each frame of video sequences. Finally, we apply a single view mensuration algorithm to each of the frames to obtain height measurements and fuse the multi-frame measurements using RANSAC algorithm. Compared with other popular methods, our proposed algorithm does not require calibrating the camera, and can track the multiple moving objects when occlusion occurs. Therefore, it reduces the complexity of calculation and improves the accuracy of measurement simultaneously. The experimental results demonstrate that our method is effective and robust to occlusion.

Research on Algorithm of Video Segmentation Based on Self-Adaptive Single Gaussian Model

Single Gaussian background modeling is often used in the static scene. In order to overcome the deficiency of the traditional single Gaussian background subtraction method, an improved adaptive background modeling algorithm was proposed. First, the number of each pixel in the image determined as foreground that were as the parameters of each pixel update rate were counted respectively. So the original fixed update rate was transformed into the dynamic update rate changing with the parameter. In this way, the phenomena of ghosting and shelling was suppressed effectively in the traditional single model of Gauss. Finally, foreground objects were obtained by using the method of shadow removal based on morphological reconstruction. The experimental results indicate that the algorithm can quickly and accurately establish and update the background model, and extract the moving object effectively. Compared with existing approaches, experimental results with different real scenes demonstrate the robustness of the proposed method.

A low-cost vehicle counter for next-generation ITS

This paper describes a vehicle counter application to be used in low-cost and low-complexity devices to be deployed in next-generation pervasive Intelligent Transport Systems. The first part of the paper introduces the Linesensor theory, which exploits the temporal redundancy of the movement to enable the processing of a 1D images (i.e. lines), thus reducing the complexity for extracting features and understanding the environment. Because of the high speed of the objects to be detected, the proposed application requires a very high frame rate and consequently an optimised design for the whole computer vision pipeline. For these reasons, in the second part of the paper, we propose a low-complexity background modelling algorithm permitting to extract information related to the whole image from a single metric. Our arguments demonstrate that the proposed algorithm has comparable performance in the segmentation operation as other state-of-the-art techniques, but reducing significantly the computational cost.

Adaptive Background Modeling Algorithm Based on Objects Dynamicity

Detecting moving objects by using an adaptive background model is a critical component for many vision-based applications. In this paper, we propose a novel background modeling algorithm for traffic video surveillance based on objects dynamicity. Our algorithm is based on the definition of background as a set of static objects and on the observation that a static object is a set of pixels having the same appearance in long time interval. The normalized RGB color is used to describe the appearance under change lighting conditions, and a binary label is introduced to capture the dynamicity of objects over time. Finally, the cardinality function, calculated in each continued time period, is used to select persistent appearance corresponding to background objects. An update formula is proposed to reconstruct background model periodically based on current frame and last background image. Experimental results from several video sequences validate the effectiveness of the proposed algorithm

Real-Time Human Action Recognition System Using Depth Map Sequences

This work presents a real-time human action recognition system that uses depth map sequence as input. The system contains the segmentation of human, the action modeling based on 3D shape context and the action graph algorithm. We effectively solve the problem of segmenting human from complex and cluttered scenes by combing a novel quadtree split-and-merge method and the codebook background modeling algorithm. We aims at recognizing actions that are used in games and interactions, especially complex actions that contain foot motion and body heave. By expanding the shape context descriptor into 3D space, we obtain translation and scale invariant features and get rid of normalization error, which is a common problem of real-life applications. Experiments in various scenarios demonstrate the high speed and excellent performance of our procedure.

Human recognition system for outdoor videos using Hidden Markov model

Human recognition is an essential requirement for human-centric surveillance, activity recognition, gait recognition etc. Inaccurate recognition of humans in such applications may leads to false alarm and unnecessary computation. In the proposed work a robust background modeling algorithm using fuzzy logic is used to detect foreground objects. Three distinct features are extracted from the contours of detected objects. An unique aggregated feature vector is formed using a fuzzy inference system by aggregating three feature vectors. To minimize computation in recognition using Hidden Markov model (HMM), the length of final feature vector is reduced using vector quantization. The proposed method is explained using five basic phases; background modeling and foreground object detection, features extraction, aggregated feature vector calculation, vector quantization, and recognition using Hidden Markov model.

Research of Background Modeling Algorithm Method Based on Multi-Frame Average Method in Moving Target Detection

In intelligent video surveillance system, there exists the defect that the background is multimodal when extracting the background image, so the extracted foreground objects are not very pure with the method of background subtraction. In this paper, we put forward an algorithm which adopts adaptive threshold to update the background. Firstly, the common background model algorithms and their defect are analysed. Then we propose the strategy of updating background template based on adaptive threshold and then give the specific algorithm steps. Finally, the algorithm is simulated and the results are analyzed.

A Background Modeling Algorithm Based on Improved Adaptive Mixture Gaussian

For better background modeling in scenes with nonstationary background, a background modeling algorithm based on adaptive parameter adjustment of the Mixture Gaussian is proposed. Mixture Gaussians is applied to learn the distribution of per-pixel in the temporal domain and to control the adaptive adjustment of number K of Gaussian components through increasing, deleting or merging similar Gaussian components adaptively. The new parameters C k and φ k are introduced in the adaptive parameter model. According to the actual situation, the adaptive adjustment of ρ can accurate track the real-time changes with the pixel, which improves the robustness and convergence. Experimental results show that the algorithm can rapidly response when the scene changes in the sequence of video with many uncertain factors, and realize adaptive background modeling with accurate target detection.