运动背景中结合特征位移矢量场模糊分割与 OTSU法的运动检测

运动背景中的运动检测难度较大,背景运动补偿后差分以及分割光流场可实现动目标和背景的分离,差分前需进行鲁棒的背景估计,且差分后易出现空洞,而光流估计在噪声以及目标运动速度较大时并不准确,尤其在光照变化时,两种方法均易失效。本文提出一种特征点位移矢量场模糊分割与图像自适应阈值化相结合的运动检测方法,实现在无任何关于运动目标或者运动背景先验信息条件下的动目标检测。通过改进的 SIFT匹配方法生成鲁棒的特征位移矢量场,采用模糊 C均值聚类算法对 SIFT位移矢量场进行无监督分类,实现动目标与背景特征的自适应分离。 OTSU法和形态学操作实现图像的自适应分割,用以修正特征点凸包,最终分割出动目标区域。与鲁棒的背景运动补偿后差分以及光流估计的对比实验表明,在目标运动速度较大、光照变化以及噪声情况下,本文方法均能够检测出运动目标,且在光照变化下的优势明显。     

Nonlinear correlation for estimating the motion of multiple objects in image sequences

A nonlinear correlation algorithm is proposed for estimating the motion of objects from an image pair. This algorithm requires no a priori information on the number, size, or shape of the moving objects and does not require feature extraction or segmentation of either image. The algorithm directly yields information on the number of moving objects, the motion of the objects, and the size of the objects. Additional processing can be performed to yield the centroid of the objects in either frame. The utility of the resulting algorithm is demonstrated by application to a pair of example image sequences.     

Adaptive vs. non‐adaptive strategies for the computation of optical flow

Confident adaptive algorithms are described, evaluated, and compared with other algorithms that implement the estimation of motion. A Galerkin finite element adaptive approach is described for computing optical flow, which uses an adaptive triangular mesh in which the resolution increases where motion is found to occur. The mesh facilitates a reduction in computational effort by enabling processing to focus on particular objects of interest in a scene. Compared with other state‐of‐the‐art methods in the literature our adaptive methods show only motion where main movement is known to occur, indicating a methodological improvement. The mesh refinement, based on detected motion, gives an alternative to methods reported in the literature, where the adaptation is usually based on a gradient intensity measure. A confidence is calculated for the detected motion and if this measure passes the threshold then the motion is used in the adaptive mesh refinement process. The idea of using the reliability hypothesis test is straightforward. The incorporation of the confidence serves the purpose of increasing the optical flow determination reliability. Generally, the confident flow seems most consistent, accurate and efficient, and focuses on the main moving objects within the image. © 2006 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 16, 35–50, 2006     

一种快速运动目标检测与跟踪算法

运动目标的检测与跟踪是计算机视觉和图象编码研究的主要内容,有着广泛的应用领域。基于光流场的检测与跟踪是其常用的方法之一,但其计算结果和效率是不能令人满意的。提出了基于帧间差阈值法和光流场相结合的快速运动目标检测与跟踪算法,诸结果表明该算法简单实用、运算速度快,克服了单纯光流场方法的不足。     

一种特征点跟踪的运动目标检测

提出一种基于特征点跟踪的运动目标分割算法。在角点跟踪过程中,8等分匹配点邻域,构造方向子邻域提高匹配精度;提出一种新的聚类准则抽取最优特征子集估计运动参数;结合统计方法消除残差噪声;采用时域滤波滤除孤立区域检测出运动目标。对200帧可见光视频序列和100帧红外图像序列中运动汽车进行检测分割实验,检出率分别达到96%和94%。     

Moving/motionless foreground object detection using fast statistical background updating

Abstract

In video surveillance, the detection of foreground objects in an image sequence from a still camera is very important for object tracking, activity recognition and behaviour understanding. The conventional background subtraction cannot respond promptly to dynamic changes in the background, and temporal difference cannot accurately extract the object shapes and detect motionless objects. In this paper, we propose a fast statistical process control scheme for foreground segmentation. The proposed method can promptly calculate the exact grey-level mean and standard deviation of individual pixels in both short- and long-term image sequences by simply deleting the earliest one among the set of images and adding the current image scene in the image sequence. A short-term updating process can be highly responsive to dynamic changes of the environment, and a long-term updating process can well extract the shape of a moving object. The detection results from both the short- and long-term processes are incorporated to detect motionless objects and eliminate non-stationary background objects. Experimental results have shown that the proposed scheme can be well applied to both indoor and outdoor environments. It can effectively extract foreground objects with various moving speeds or without motion at a high process frame rate.     

序列图像中运动目标检测

提出动态背景下序列图像中的运动目标检测算法。利用像素邻域的各向同性对图像进行归一化,消除亮度变化等因素的影响;利用光流信息并结合小波变换由粗及精计算速度场来配准图像;用当前帧作参考图像,通过时域积分校正背景图像。当前帧与校正后背景图像作差得到差分图像。假设该差分图像中噪声分布为高斯分布,由高斯分布的3σ特性滤除差分图像中的噪声,则粗定位出目标;最后以聚类方法确定运动目标区域。分别对200帧可见光和200帧红外图像序列进行实验,检测率分别为95%和94%。     

一种基于马尔可夫随机场的运动目标检测算法

本文提出了一种基于马尔可夫随机场的运动目标检测算法.针对传统时间分割使用主观固定阁值的缺点,使用马尔可夫随机场模型对差分图像建模,并提出了一种新的模型阶次选择算法,以及一种可以加速收敛过程的随机场迭代算法.采用期望最大算法获取高斯分布参数并检测运动变化区域,利用形态学运算修正时间分割模板;空间分割部分提出了基于人眼视觉特征的改进分水岭算法,有效地解决了过分割问题;最后对时、空间分割结果进行信息融合处理,从而得到完整的运动目标.仿真实验结果证明了本文算法可以有效地分割视频运动目标.     

Optical compressive change and motion detection

Localization information of moving and changing objects, as commonly extracted from video sequences, is typically very sparse with respect to the full data frames, thus fulfilling one of the basic conditions of compressive sensing theory. Motivated by this observation, we developed an optical compressive change and motion-sensing technique that detects the location of moving objects by using a significantly fewer samples than conventionally taken. We present examples of motion detection and motion tracking with over two orders of magnitude fewer samples than required with conventional systems.     

一种视频图像序列中运动对象的分割与跟踪算法

本文提出了一种视频图像序列中运动对象的分割与跟踪算法。该算法通过Canny算子检测出差帧图像的边缘信息,并结合当前帧与背景帧的边缘图像,提取出运动对象。在后续帧中通过建立前帧感兴趣运动对象与当前帧中各运动对象的帧间向量来跟踪当前帧中感兴趣的视频对象。实验结果表明,该算法可行,而且由于该算法简单、计算复杂度小,能很好地满足实时监控系统中对感兴趣运动对象的提取与跟踪。     

Noninvasive OCT imaging of the retinal morphology and microvasculature based on the combination of the phase and amplitude method

To implement the optical coherence tomography (OCT) angiography on the low scanning speed OCT system, we developed a joint phase and amplitude method to generate 3-D angiograms by analysing the frequency distribution of signals from non-moving and moving scatterers and separating the signals from the tissue and blood flow with high-pass filter dynamically. This approach firstly compensates the sample motion between adjacent A-lines. Then according to the corrected phase information, we used a histogram method to determine the bulk non-moving tissue phases dynamically, which is regarded as the cut-off frequency of a high-pass filter, and separated the moving and non-moving scatters using the mentioned high-pass filter. The reconstructed image can visualize the components of moving scatters flowing, and enables volumetric flow mapping combined with the corrected phase information. Furthermore, retinal and choroidal blood vessels can be simultaneously obtained by separating the B-scan into retinal part and choroidal parts using a simple segmentation algorithm along the RPE. After the compensation of axial displacements between neighbouring images, three-dimensional vasculature of ocular vessels has been visualized. Experiments were performed to demonstrate the effectiveness of the proposed method for 3-D vasculature imaging of human retina and choroid. The results revealed depth-resolved vasculatures in retina and choroid, suggesting that our approach can be used for noninvasive and three-dimensional angiography with a low-speed clinical OCT, and it has a great potential for clinic application.     

基于累积光流的运动对象检测

利用解析小波和M估计求解光流约束方程,得到一种精度高的、稳健的光流计算方法,并在此基础上引入扩展累积光流,针对累积光流计算中的几个特殊问题进行了相应的处理,实现了一种稳健的运动对象检测算法.实验结果表明,该算法正确检测运动对象的精度高,能够检测出慢速物体和小目标的运动,并较好地解决了遮挡问题.     

基于差异积累的视频运动对象自动分割

针对视频运动对象的自动分割，本文给出了一种基于差异积累的自动分割算法。与传统的基于运动信息变化检测方法不同，该算法通过累积的帧差信息构建出可靠的背景，与当前帧比较进而提取出视频运动对象。本文提出了一种增强的基于Otsu法的自适应阈值化方法，能更准确地对背景差图像进行阈值化分割，克服了传统Otsu法阈值化容易失效的问题。改进的基于区域生长的定位方法更能避免传统方法的误定位及重定位的问题。实验结果表明，本文算法具有较好的实时性、自适应性以及鲁棒性，可以较为可靠地建立背景模型并进行实时更新，适用于刚体或非刚体存在平缓的光照变化以及摄像头微抖动的视频运动对象的自动分割。     

分块特征匹配与局部差分结合的运动目标检测

针对摄像机在运动情况下运动目标检测问题的研究,提出了一种尺度不变特征变换图像分块特征匹配和局部差分算法相结合的运动目标检测方法。该方法不仅可以减少图像配准参数求解所需要的时间,弥补全局运动补偿对图像完整性的影响,并且可以在摄像机运动的情况下准确地检测出运动目标。 实验证明了该方案的有效性。     

快速背景重建的在线运动目标检测

为了能快速地从视频图像序列中创建可靠的背景图像,进而提取运动目标,文中提出了一种基于反馈信息的运动目标检测算法.首先提出了基于相邻帧信息和背景估计信息相融合的背景重建算法,保证了在视频场景改变时仍能迅速捕捉背景;还提出了基于一种在线Otsu法的运动目标检测,将相邻帧运动目标信息反馈到目标提取算法中,弥补传统Otsu法的不足;最后提出了对光线变化具有一定鲁棒性的背景估计算法.实验表明,该方法的重建速度快,准确率高,能满足实时检测的需要.     

Real-time detection of multiple moving objects in complex image sequences

In this article, a real-time multistage method for detecting multiple objects moving in real scenes is presented. At the first level, a rough focus-of-attention mechanism is used to individuate areas of the input image that show remarkable differences with a real-time updated background image. Binary statistical morphology (BMS) operators are applied to individuate image pixels, which can be associated with real objects moving into the scene. High stability to noise is obtained by tuning the smoothing effects of the BSM operators according to the noise level present in the original image sequence. Then, at the second level, a composition of BSM is applied to eliminate isolated points and to favor dense agglomerate of changed pixels, i.e., blobs. The last level attempts to describe changes in terms of motion of blobs by allowing blobs to merge, split, appear, and vanish. A blob-matching procedure is used for tracking blobs over consecutive frames. Experimental results on real scenes, which demonstrate the advantages of the proposed method with respect to existing change detection methods, are given. © 1999 John Wiley & Sons, Inc. Int J Imaging Syst Technol 10, 305–317, 1999     

Detecting multiple objects under partial occlusion by integrating classification and tracking approaches

A visual‐based framework for detecting in real time multiple objects in real outdoor scenes is presented. The main novelty of the system is its capability to reduce the problems of partial occlusions and/or overlaps that occur very commonly in real scenes containing multiple moving objects. Overlaps and occlusions are dealt with by integrating classification and tracking procedures into a data‐fusion distributed sensory network. Neural tree‐based networks are applied to distinguish among isolated objects and groups of objects on the image plane. Extended Kalman filters are applied to estimate the number of objects in the scene, their position, and the related motion parameters. Experimental results on complex outdoor scenes with multiple moving objects are presented. © 2001 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 11, 263–276, 2000     

流动颗粒显微图像测量技术研究

针对显微环境下流动颗粒测量的需要,构建了测量平台,研制了基于MEMS工艺的测量器件。在分析流动颗粒显微图像特征的基础上,提出了基于颗粒运动特性的图像测量流程,其中提出了基于基准桢差分的目标提取算法,解决了流动颗粒目标分割的问题,根据图像运动模糊的退化模型,研究了图像模糊恢复的问题。最后以润滑油中的磨损颗粒为例,验证了该文的分析方法对流动颗粒检测的有效性。     

基于区域平滑约束的光流场估计方法

从光流的基本约束方程出发，提出了一种基于区域平滑约束的光流场估计方法。     

Camera calibration using a genetic algorithm

An autonomous robot will have to detect moving obstacles online before it can plan its collision-free path, while navigating in a dynamic environment. The robot collects information about the environment with the help of a camera and determines the inputs for its motion planner through image analysis. The present article deals with issues related to camera calibration and online image processing. The problem of camera calibration is treated as an optimization problem and solved using a genetic algorithm so as to achieve minimum distorted image plane error. The calibrated vision system is then utilized for the detection and identification of the objects by analysing the images collected at regular intervals. For image processing, five different operations, such as median filtering, thresholding, perimeter estimation, labelling and size filtering, have been carried out. To show the effectiveness of the developed camera-based vision system, inputs of the motion planner of a navigating robot are calculated for two different cases. It is observed that online detection of the shapes and configurations of the obstacles is possible by using the vision system developed.