基于概率模型运动检测技术的视频对象分割技术研究

MPEG-4最重要的特点是用音频-视频对象来描述内容和进行编码。这就需要按照视频对象进行视频图像分割,而视频对象具有时域和空域这两大特点,而时域分割又占主要的地位。研究了一种时域分割算法,即研究了一种运动检测技术方案,采用Til Aach,Andre Kaup提出的基于概率模型的运动检测技术,用Gauss模型描述噪声引起的像素值变化,将运动目标和有噪声的背景区别开来。最后,采用二值开闭重建滤波器来去除前景和背景中的“小洞”,从而获取了最终的运动检测掩码。     

Spatio-temporal pattern recognition using hidden Markov models

A spatio-temporal method for identifying objects contained in an image sequence is presented. The Hidden Markov Model (HMM) technique is used as the classification algorithm, making classification decisions based on a spatio-temporal sequence of observed object features. A five class problem is considered. Classification accuracies of 100% and 99.7%, are obtained for sequences of images generated over two separate regions of viewing positions. HMMs trained on image sequences of the objects moving in opposite directions showed a 98.1% successful classification rate by class and direction of movement. The HMM technique proved robust to image corruption with additive correlated noise and had a higher accuracy than a single-look nearest neighbor method. A real image sequence of one of the objects used was successfully recognized with the HMMs trained on synthetic data. This study shows the temporal changes that observed feature vectors undergo due to object motion hold information that can yield superior classification accuracy when compared with single-frame techniques     

一种基于背景自适应的运动目标检测与跟踪算法

背景图像差分法是运动目标实时检测中常用的方法,但缺乏背景图像随监视场景光照变化而及时更新的合理方法,限制了该方法的适应性。对此,文章首先提出了一种自适应背景更新方法;然后利用最大类间方差法实现运动目标的自适应阈值分割,并利用基于形态学方法的连通区检测算法检测运动目标;最后以Kalman滤波为运动模型实现对运动目标的连续跟踪。实验结果表明:所提方法可随着光照条件的变化,实时、准确地检测出运动目标并实现稳定跟踪。     

Noise effects on centroid tracker aim point estimation

An investigation of errors due to noise in centroid tracker aim-point estimation is presented. The centroid tracker discussed is similar to the tracker described by A.L. Gilbert et al. (1980). Simplifications to this algorithm were made so that the derived models would be consistent with the actual tracker algorithm. Two statistical models are derived which relate image noise effects to computation of the target centroid. The first model, the simplified aim-point error model, is derived by assuming that the probabilities of incorrectly classifying target and background pixels are equal. The second model, the extended aim-point error model, is derived by assuming that the probability of incorrectly classifying a target pixel can differ from the probability of incorrectly classifying a background pixel. These models are described and their mathematical implication is discussed. Simulation results which verify the models are presented     

基于运动特征和中心环绕模型的空对地车辆检测方法

平台的移动基于空基车辆检测以其视野广、速度快、适应性强等优点,近些年成为了智能交通和航空应急救援等领域内的一个热点课题。然而,在城市背景的交通环境下,由于复杂背景造成的干扰、车辆运动的不规律性以及光照等条件变化带来的噪声,有效地获取运动对象特征是空对地视觉监视方法面临的一个难题。为有效解决这一问题,提出了结合运动特征的中心环绕模型进行空对地车辆检测的算法。现有的空对地车辆检测算法,大多是针对固定的摄像头的情况下,基于运动特征进行检测,例如光流法、背景差法等。这些方法在对高速公路等背景相对较为简单的环境下获得了良好的检测结果,但当面对城市复杂背景条件或摄像头自身运动的状况时,难以有效地检测出运动目标。而在对颜色、强度等静态特征进行显著度分析的基础上,加入了对运动特征的分析,可以解决在复杂环境下准确提取车辆目标有效特征的问题。试验表明,相较于基于图像相减的检测算法和单一显著度算法,该算法可以获得更高的检测率与更低的误报率。     

An algorithm for detection of moving optical targets

An algorithm is described for detecting moving optical targets against spatially nonstationary Poisson background and noise. The algorithm has applications in optical detection of objects such as meteors, asteroids, and satellites against a stellar background. A maximum-likelihood approach is used which results in reducing interference from stars. It is shown that by choosing a detection threshold to provide a constant false alarm rate, the resulting algorithm is independent of the signal strength of the target. An analysis of this algorithm is presented, showing the probability of detection for several false-alarm rates     

Efficient target tracking using dynamic programming

A dynamic programming (DP) algorithm has been developed for the detection and tracking of subpixel-sized, low signal-to-noise ratio (SNR) targets observed by side-or forward-looking imaging sensors. A distinguishing feature of this approach is that target detection and tracking are combined into a single optimization procedure that takes into account statistical models of target motion, background noise, and clutter. Current work has led to a number of technical innovations that improve the performance and efficiency of the DP tracking algorithm, including the development of a new track scoring function, and an extension to the basic DP algorithm that reduces computation requirements by over an order of magnitude. A prototype infrared (IR) target tracking system incorporating these enhancements has been implemented for a step-starting IR camera application. Sensitivity improvements of several decibels over conventional sequential detection and tracking algorithms were realized     

结合核相关滤波器和深度学习的运动相机中无人机目标检测

针对无人机与相机快速相对运动造成的运动模糊问题，以及小型无人机外观信息缺失和背景复杂造成漏警和虚警问题，提出了一种新的无人机检测-跟踪方法。针对成像尺寸小于32像素×32像素的无人机目标，提出改进的多层特征金字塔的分类和目标框回归器作为目标检测器，克服漏警。利用检测结果初始化基于核相关滤波的目标跟踪器，并持续修正跟踪结果，跟踪结果为剔除检测器虚警提供依据。在跟踪过程中，引入对观测场景纹理自适应的相机运动补偿策略实现目标重定位。多场景下的实验结果表明：提出的方法在对高速运动小目标的检测和跟踪指标上显著优于传统方法，且运动补偿机制的引入进一步增强了方法在极端复杂场景下的鲁棒性。     

Identification and velocity computation of multiple moving objectsin images

Analysis for identifying and computing the velocity of multiple moving objects in sequences of images is presented. While previously published analyses addressed sequences with one or two moving objects, this analysis is extended to sequences with multiple moving objects. simulation is performed to demonstrate the applicability of this technique. Multiple moving objects appear in the spectral domain as groups of individual peaks which identify the objects and are used to compute their velocities. Algorithms which evaluate the spectral peaks and calculate their velocity are also presented     

Moving Object Detecting Method in Video

A method of moving object detection that integrates image gradient information with motion information is described herein. One multi-scale morphological gradient operator is used to detect gradient information. This operator has a very good effect on the detail of the edge location. The result is superior to the traditional method in edge continuity and isotropy. Past approaches have used methods of detection based on motion information and edge information, but this is the first to use a multi-scale morphological gradient operator. The implementation described can operate on lower resolution images under difficult conditions. Novel contributions include: 1. Based on multi-scale morphological gradient operator, this algorithm is robust to noise, object shadow, and illumination. 2. Using interframe difference and only the first two frames, this detection method uses very little computation and is suitable for real-time applications. 3. This method can detect rigid-body or non-rigid-body moving objects among video sequences with complicated backgrounds.     

复杂条件下的运动目标检测方法研究综述

运动目标检测是当前研究热点之一,被广泛地应用于计算机视觉、视频处理等领域。将运动目标检测的三种常用方法进行对比,总结其各自的适用性及局限性。针对光照变化、背景存在干扰运动、前景色与背景色接近等复杂条件,综述了相应的多种解决方案,并指出了方案算法的局限性,对复杂条件下运动目标检测方法的发展方向进行展望,从而为相关工作的进一步研究指出了方向。     

Video surveillance applications using multiple views of a scene

Automatic video surveillance techniques are used to detect intruders within a scene. This task is mostly reduced to the problem of detecting moving objects evaluating image sequences of a monocular camera. An essential problem of this monocular approach is its inability to measure the 3D-size and 3D-position of objects reliably, as object size and velocity are estimated within the 2D-image plane. To include 3D-information about the scene an approach using a second camera is proposed in this paper which combines the evaluation of the measurement data of the two cameras using an efficient 3D-scene model. Here, two cameras are used with an overlapping field of view, which represents an installation often used in existing video surveillance applications. It is shown that using the combined evaluation of the two cameras, the false detection rate in the case of moving shadows, leaves, birds and insects or blindings can be further reduced compared to a pure monocular evaluation     

Modified Difference Change Detector for Small Targets in SAR Imagery

One of the most straightforward techniques for detecting changes in an image involves forming the difference between a test image and a reference image. Unfortunately, such a technique can give rise to a large number of false alarms due to the statistical variability of the underlying pixel values, as has been well established within the radar community over the years. One method for dealing with this large number of false alarms involves forming the ratio, rather than the difference, of two synthetic aperture radar (SAR) images. We introduce a modified version of the standard differencing technique to overcome problems associated with pixel value variability. The new (modified) differencing approach utilizes assumptions about the statistics of the image background and the object being sought (target) to reduce the number of false alarms due to highly variable background (clutter) regions, and it includes the standard ratio test as a special case. In fact, we find that the modified difference approach can also be viewed as a modified version of the ratio test with a threshold that varies as a function of the background clutter radar cross section (RCS). We also present an abridged, albeit suboptimal, version of this approach that eliminates assumptions regarding the target's probability distribution, and we analyze both of the approaches. We then compare these results with those obtained with a standard ratio test, and illustrate how the modified difference test reduces to the ratio test under certain operating conditions. The abridged version of the modified approach is applied to high resolution synthetic aperture radar imagery and compared with results obtained with the classical differencing technique, and following this, the modified difference technique is compared with the standard ratio test. Results suggest that under appropriate conditions the abridged, modified technique can successfully detect changes without the need for any image segmentation.     

Use of measurements from an imaging sensor for precision targettracking

The extraction of measurements for precision tracking of the centroid of a target from a forward-looking infrared imaging sensor is presented. The size of the image of the target is assumed to be small, i.e. around 10 pixels. The statistical characterization of the centroid of the target is obtained. Similarly, the statistical properties of the image correlation of two frames, which measures the target offset, are derived. Explicit expressions that map the video noise statistics into measurement noise statistics are obtained. The offset measurement noise is shown to be autocorrelated. State variable models for tracking the target centroid with these measurements are then presented. Simulation results and quantitative conclusions about achievable subpixel tracking accuracy are given. It is shown that the filter that models the autocorrelated measurement noise provides the best performance     

基于属性形态学分析的图像显著极值检测及应用

由于图像噪声的存在,使得利用传统的极值检测算法通常会使要提取的显著极值淹没在大量的噪声极值中;同时由于先验知识的缺乏,采用普通滤波技术也往往不能很好的滤除噪声,反而会破坏图像的关键结构。本文提出了一种基于属性形态学分析的图像显著极值检测算法。该算法可以在不需要对图像进行滤波的前提下,从数学形态学的角度对图像极值的显著性进行计算和评估,从而能够较好地提取出显著的极值。在沉浸模拟算法的基础上,给出了基于属性形态学分析显著极值检测的快速算法实现,并将其成功应用在视觉注意选择和独立运动目标检测上。实践证明,该算法不仅具有较好的抗噪声特性,而且快速实用,具有广泛的应用价值。     

用RTR技术研究固体发动机燃烧室中粒子运动轨迹(Ⅱ)图象处理技术应用

针对采用RTR技术获取固体火箭发动机（SRM）燃烧室中凝相粒子运动图象的特点，应用了背景减影、图象增强、图象分割、噪声过滤、图象叠加等图象处理手段对其进行处理，获得了较清晰的粒子运动轨迹图象。结果表明图象处理技术是开展利用RTR技术进行SRM二相流研究的关键环节之一。     

Family of constrained signal detectors for hyperspectral imagery

Generalized likelihood ratio tests (GLRTs) are derived for the problem of detecting targets in hyperspectral images. These detectors are derived under the assumptions that the signals from the materials in the image mix linearly and that the noise in the system is Gaussian. It is also assumed that the abundances of the signals from the various materials in a pixel must sum to one. This constraint models the fact that the material abundances in a pixel are just the fraction of the pixel that they occupy. Under these assumptions, detectors are derived which outperform the detectors derived without the sum-to-one constraint.     

A Model for Generating Synthetic VHF SAR Forest Clutter Images

We propose a model for generating low-frequency synthetic aperture radar (SAR) clutter that relates model parameters to physical characteristics of the scene. The model includes both distributed scattering and large-amplitude discrete clutter responses. The model also incorporates the SAR imaging process, which introduces correlation among image pixels. The model may be used to generate synthetic clutter for a range of environmental operating conditions for use in target detection performance evaluation of the radar and automatic target detection/recognition algorithms. We derive a statistical representation of the proposed clutter model's pixel amplitudes and compare with measured data from the CARABAS-II SAR. Simulated clutter images capture the structure and amplitude responses seen in the measured data. A statistical analysis shows an order of magnitude improvement in model fit error compared with standard maximum-likelihood (ML) density fitting methods.     

Adaptive Rectification Filter for Detecting Small IR Targets

Clutter suppression is one of the most important subjects in the field of small target detection under infrared (IR) strong clutter background. While removing the clutter background, however, such methods may reverse the relative energy distribution of target and noise in the clutter suppressed image, and disturb the subsequent target segmentation and detection. This paper analyzes the causation of such problems, does research on the relationship between target energy characteristics and detection probability, and presents a novel filter of energy distribution adaptive rectification (EDARF). Based on the EDARF, an improved framework of dim small target detection is proposed to rectify the energy distribution in the clutter-suppressed images by conventional adaptive filters. The proposed EDARF's performance is estimated by experimental comparisons of three linear/nonlinear filters before and after using EDARF. Extensive experimental results show that the proposed EDARF improves efficiently the performance of detecting dim small targets against strong undulant cloud-cluttered backgrounds.