融合光流速度场与背景差分的自适应背景更新方法

为利用视频监控系统识别铁路客运站的客流,根据车站监控环境的多变性特点,提出将光流速度场算法与背景差分算法相结合的自适应背景更新方法。将光流引入背景建模中,并与背景差分结果进行并运算,再通过"死角"灰度优化处理,实现背景的实时更新。以实录的北京南站视频对给出的自适应背景更新方法进行验证,并与均值背景法和高斯背景法的处理结果进行比较,结果表明,自适应背景更新方法较好地解决了背景的提取、实时更新及运动目标阴影扰动等问题,拟合的背景干净、虚影弱,描述的背景符合实际背景场景,用于动态场景的客流识别取得了较好的效果。

五点约束最小二乘法估计光流速度场

提出一种二维光流场快速计算算法 首先求取当前像素点光流约束线与其 8邻域像素点所对应 8条光流约束线的交点 ;其次从 8个交点中选取几何位置处于中间的 4点 ,即速度处于中间值 ,且相互接近的 4点 ,并以其对应的像素点与当前像素点一起构成 5置信点 ;利用 5置信点的光流约束方程构造一超定的方程组 ;

基于视频分析的奶牛呼吸频率与异常检测

为实现奶牛呼吸状态信息获取的自动化、智能化,在构建奶牛视频实时采集系统的基础上,研究并提出了奶牛呼吸频率与异常检测方法。用光流法计算视频帧图像各像素点的相对运动速度,根据各点速度,对像素点进行循环Otsu处理筛选出呼吸运动点,动态计算速度方向曲线的周期即可检测牛只呼吸频率,并根据单次呼吸耗时检测呼吸是否异常。对72头奶牛共进行360 min检测试验,结果表明,呼吸频率计算准确率为95.68%,异常检测成功率为89.06%,平均异常误检次数为2.53次/min。

一种交通拥堵自动检测方法

目前,高速公路交通拥堵日趋频繁,然而大部分高速公路的管理方式仍然是通过人工查看轮巡监控视频来发现交通拥堵,效率低下。本文提出一种对高速公路轮巡监控视频自动检测交通拥堵的方法。该方法实时检测监控视频场景是否轮巡、切换、变动,自适应提取变动场景的道路边界结构并分类,然后在提取的道路边界结构范围内,计算宏观边缘占有率与加权平均光流速度两个宏观交通状态参数,并构造特征集,根据不同的道路边界类型选择对应支持向量机来实现对交通拥堵的自动检测和判别。实验结果表明,该方法能有效地对轮巡模式工作的高速公路监控视频检测交通拥堵,检测时间不大于30s,检测正确率达91.8%。

一种融合遮挡分割的多目标跟踪算法

复杂环境下的多目标视频跟踪是计算机视觉领域的一个难点,有效处理目标间遮挡是解决多目标跟踪问题的关键。提出了一种融合遮挡分割的多目标跟踪算法,计算每个目标的光流速度概率直方图,反映其运动统计信息;综合使用外观、运动、颜色信息构造新的像素距离表达,借助分阶段分类思想及K均值聚类技术进行遮挡分割,得到准确的运动前景像素;在粒子滤波器跟踪框架下,使用概率外观模型进行多目标跟踪,更好地处理动态遮挡问题。实验表明,所提算法解决了复杂环境下的多目标跟踪问题。

一种用于异常行为检测的运动特征描述子

目前,用于描述视频中人群的运动信息大多是基于光流的速度描述子。事实上,加速度蕴含丰富的运动信息,能够提供速度描述子在描述复杂运动模式时缺失的信息,以更好地表征复杂的运动模式。文中研究了一种运动特征描述子,使用受限玻尔兹曼机模型进行异常行为检测。首先,提取视频中的光流场信息,计算帧间加速度光流;然后,对一个时空块中的加速度信息进行直方图统计,将若干帧的所有时空块直方图特征进行拼接,从而获得加速度描述子;最后,在仅包含正常行为的训练集上建立受限玻尔兹曼机模型,在测试阶段根据测试视频重建特征与原始特征的误差大小进行异常检测。实验表明,所提出的加速度描述子结合速度描述子,在UMN数据集与UCF-Web数据集上,ROC曲线下的面积分别达到了0.984与0.958,相较于其他算法,所提方法取得了更高的异常行为检测准确率。

Experimental Investigation of Single-phase Flow in Structured Packing by LDV

To date, many models have been developed to calculate the flow field in the structured packing by the computational fluid dynamics （CFD） technique, but little experimental work has been carried out to serve the vali-dation of flow simulation. In this work, the velocity profiles of single-phase flow in structured packing are measured at the Reynolds numbers of 20.0, 55.7 and 520.1, using the laser Doppler velocimetry （LDV）. The time-averaged and instantaneous velocities of three components are obtained simultaneously. The CFD simulation is also carried out to numerically predict the velocity distribution within the structured packing. Comparison shows that the flow pattern, velocity distribution and turbulent kinetic energy （for turbulent flow） on the horizontal plane predicted by CFD simulation are in good agreement with the LDV measured data. The values of the x-and z-velocity components are quantitatively well predicted over the plane in the center of the packing, but the predicted y-component is sig-nificantly smaller than the experimental data. It can be concluded that experimental measurement is important for further improvement of CFD model.

A calculation method for low dynamic vehicle velocity based on fusion of optical flow and feature point matching

Aming at the problem of the low accuracy of low dynamic vehicle velocity under the environment of uneven distribution of light intensity,an improved adaptive Kalman filter method for the velocity error estimate by the fusion of optical flow tracking and scale mvaiant feature transform(SIFT)is proposed.The algorithm introduces anonlinear fuzzy membership function and the filter residual for the noise covariance matrix in the adaptive adjustment process.In the process of calculating the velocity of the vehicle,the tracking and matching of the inter-frame displacement a d the vehicle velocity calculation a e carried out by using the optical fow tracing and the SIF'T methods,respectively.Meanwhile,the velocity difference between theoutputs of thesetwo methods is used as the observation of the improved adaptive Kalman filter.Finally,the velocity calculated by the optical fow method is corrected by using the velocity error estimate of the output of the modified adaptive Kalman filter.The results of semi-physical experiments show that the maximum velocityeror of the fusion algorithm is decreased by29%than that of the optical fow method,and the computation time is reduced by80%compared with the SIFT method.

Contributions to Hom-Schunck optical flow equations-part I： Stability and rate of convergence of classical algorithm

Globally exponential stability （which implies convergence and uniqueness） of their classical iterative algorithm is established using methods of heat equations and energy integral after embedding the discrete iteration into a continuous flow. The stability condition depends explicitly on smoothness of the image sequence, size of image domain, value of the regularization parameter, and finally discretization step. Specifically, as the discretization step approaches to zero, stability holds unconditionally. The analysis also clarifies relations among the iterative algorithm, the original variation formulation and the PDE system. The proper regularity of solution and natural images is briefly surveyed and discussed. Experimental results validate the theoretical claims both on convergence and exponential stability.

Study on the flow field and concentration characteristics of the multiple tandem jets in crossflow

The characteristics of single and multiple tandem jets(n=2,3,4) in crossflow have been investigated using the realizable k-ε model.The results of this model agree well with experimental measurements using PIV(Particle Image Velocimetry) or LIF(Laser Induced Fluorescence).We analyzed the calculated results and obtained detailed properties of velocity and concentration of the multiple jets in the pre-merging and post-merging regions.When the velocity ratio is identical,the bending degree of the leading jet is greater than that of the rear jets.The last jet penetrates deeper as the jet number increases,and the shielding effect of the front jet declines with jet spacing increase.Interaction of the jet and crossflow induces formation and development of a counter-rotating vortex pair(CVP).CVP makes the distribution of concentration appear kidney-shaped(except in the merging region),and maximum concentration is at the center of the counter-rotating vortex.Concentration at the CVP center is 1.03-1.4 times that of the local jet trajectory.Post-merging velocity and concentration characteristics differ from those of the single jet because of the shielding effect and mixing of all jets.This paper presents a unified formula of trajectory,concentration half-width and trajectory dilution,by introducing a reduction factor.

UV soaking for enhancing the photocurrent and response speed of Cs_(2)AgBiBr_(6)-based all-inorganic perovskite photodetectors

The response speed of the reported Cs_(2)AgBiBr_(6)-based photodetectors exhibits a wide variation ranging from microseconds to nanoseconds,while the reason is still unclear.Apart from the conventional approaches such as reducing effective area,new regulating approaches for response speed improvement have rarely been reported.On the other hand,it is generally believed that ultraviolet(UV)light has negative impact on perovskite devices resulting in performance degradation.In this work,we demonstrated that the response speed of the photodetector with FTO/Cs_(2)AgBiBr_(6)/Au structure can be effectively regulated by utilizing UV light-soaking effect without reducing the device area.Particularly,the decay time is efficiently modulated from 30.1μs to 340 ns.In addition,the−3 dB bandwidth of the device is extended from 5 to 20 kHz.It is worth mentioning that the light current is remarkably boosted by 15 times instead of any attenuation.Furthermore,we prove the universality of UV soaking treatment on Cs_(2)AgBiBr_(6)-based photodetectors with other all-inorganic structures,i.e.,FTO/TiO_(2)/Cs_(2)AgBiBr_(6)/Au,FTO/Cs_(2)AgBiBr_(6)/TiO_(2)/Au and FTO/TiO_(2)/Cs_(2)AgBiBr_(6)/CuSCN/Au.Our results demonstrate a new method to improve the response speed and light current of Cs_(2)AgBiBr_(6)-based perovskite all-inorganic photodetectors.

An Experimental Study of Air-Solid Two-Phase Flow in a 90° Bend Using LDV System

The measurements of the mean streamwise and radial velocities,the associated turbulence and the relative particle densities were made in an air-solid two-phase flow in a square sectioned (30mm×30mm)90°vertical to horizontal bend using laser Doppler velocimetry.The radius ratio of the bend was 2.0.Glass beads of 100μm in diameter were employed to form the solid phase.The measurements of air and solid phases were performed separately at the same bulk velocity 19.34m/s,corresponding to a Reynolds number of 3.87×10~4.The mass ratio of solid to air was 1.6%.The results indicate that the particle trajectories are very close to straight lines. The streamwise velocity profiles for the gas and the solids cross over near the outer wall with the solids having the higher speed.At θ=30°and 45°,particle-wall collisions happen mostly in the region from θ=30°to θ=75°,and cause a sudden change in solid velocity.The particles tend to move towards the outer wall in 90° bend.The particle concentration near the outer wall is much higher than that near the inner wall in the bend, and there are few particles in the inside of the bend.The bend leads to apparent phase separation:atθ=45°, the solids concentrate in the half of the duct near the outer wall.Afterθ=60°the second peak concentration appears,and goes gradually towards the inner wall.

Large eddy simulation of a round jet into a counterflow

A round jet into a counterflow under different jet-to-current velocity ratios was investigated using large eddy simulation.The results agree well with experimental measurements from laser-Doppler anemometry and laser-induced fluorescence that include velocity and mean concentrations along the centerline and radial direction.Vortex rings appear in the region near the jet exit and large-scale vortex structures still occur near the stagnation point.The flow becomes more chaotic and three-dimensional with the presence of these structures.In particular,their presence near the stagnation point results in large velocity fluctuations that enhance the mixing process and dilution.These fluctuations are described by probability density functions that deviate from Gaussian distribution.The three-dimensional streamlines indicate that the jet not only oscillates in three directions but also rotates about the jet axis and around the vortex.The second and third moments of the velocity or scalar fluctuations identify that the mixing processes are greater in the region before the stagnation point.