基于帧间颜色梯度的背景建模

在分析了交通场景视频特点的基础上,根据视频流中帧间像素颜色变化情况来确定背景像素值。提出以帧间像素梯度值作为背景像素的判断条件来建立背景模型的方法。在计算中将图像进行分块,通过求子区域颜色梯度变化值来确定其背景,然后根据各子区域的计算结果重建背景模型。实验结果表明:平均经过100帧图像初始化后,即可完成背景建模,初始化帧数比其它方法少,平均每帧图像的处理时间为15.566 ms。该方法背景建模的速度和初始化帧数都满足交通场景视频监控系统的要求。     

改进的基于雾气理论的视频去雾

为了进一步提高有雾视频的可用性,提出了一种改进的基于雾气理论的视频去雾方法。该方法以雾气理论为基础,利用暗原色先验知识以及Retinex方法和图像融合的方式,将从视频背景图像求取的大气光值和介质传播图应用于视频的所有帧以便去除雾气。从主观定性评价、客观定量评价和运算速度3个方面对视频去雾效果进行了评价。结果表明,对分辨率为480×640的视频,本文方法的运算速度为5.45frame/s,不仅获得了较快的处理速度且能有效避免复原视频中出现颜色跳变的现象。由于本文采用区间估计的方式对大气光值进行估计,同时利用图像复原和图像增强的方法求取介质传播图,因此,复原视频的清晰度和对比度比典型的视频去雾方法有所提高,颜色效果也比较好。     

基于视频流的车牌识别研究及系统设计

该文深入研究从视频流中提取运动车辆进行车牌识别的问题,提出了一种车牌识别算法,根据该算法开发了基于MFC的视频流车牌识别可视化系统。算法采用三帧差分与背景消减相结合的方法提取含有运动车辆的关键帧,对关键帧进行灰度化,采用Sobel算子进行边缘检测,融合形态学处理对关键帧进行去噪,从而实现车牌的定位,用投影法结合字符特征对车牌字符进行精确分割,采用BP神经网络算法提取字符特征,输出字符识别结果,由此完成车牌字符识别。通过对50组不同的车辆视频进行检测识别,识别率(包含汉字)达到96%以上。     

MRT letter: Visual attention driven framework for hysteroscopy video abstraction

Diagnostic hysteroscopy is a popular method for investigating the regions in the female reproductive system. The videos generated by hysteroscopy sessions of patients are recurrently archived in medical libraries. Gynecologists often need to browse these libraries in search of similar cases or for reviewing old videos of a patient. Diagnostic hysteroscopy videos contain a lot of information with abundant redundancy. Key frame extraction‐based video summarization can be used to reduce this huge amount of data. Moreover, key frames can be used for browsing and indexing of hysteroscopy videos. In this article, a domain specific visual attention driven framework for summarization of hysteroscopy videos is proposed. The visual attention model is materialized by computing saliency based on color, texture, and motion. The experimental results, in comparison with other techniques, demonstrate the efficacy of the proposed framework. Microsc. Res. Tech. 76:559–563, 2013. © 2013 Wiley Periodicals, Inc.     

改进隐马氏模型的运动人体模型学习

基于人体模型的跟踪方法易于实现视频的运动人体跟踪,而且利用较少的视频帧数即可学习获得人体模型。本文针对给出的视频提出了学习人体模型的学习算法。利用片图模型表示未经学习的人体,改进的隐马尔可夫模型(HMM)模拟人体在视频序列各帧间的运动,并使用机器学习方法对该改进的HMM进行推理,获取改进HMM的参数,从而获得所需的人体模型。学习得到的人体模型由包含颜色信息的各人体肢体模板组成。实验显示只用80~90帧包含有人体运动的序列图像,便可学习得到该运动人体的人体模型。结果表明,该学习框架效果明显,可用于快速学习视频序列中的运动人体模型,且可用于学习一人或多人的人体模型。     

共光轴实时彩色融合系统

长波红外与可见光图像的自然感彩色融合有助于观察者快速感知环境和发现目标,而共光轴结构对于提高融合图像质量和简化配准算法具有独特的优势。为满足战场侦查、监视等需求,设计了一种基于TMS320DM642数字信号处理器和ATmega32单片机的共光轴实时彩色融合系统。融合系统的长波红外和可见光成像系统平行放置,利用分光镜和反射镜实现光路分离,通过视场匹配和光轴校准实现双波段视频图像完全配准。图像处理程序基于DSP/BIOS嵌入式操作系统开发,根据视频图像的特点,采用基于CbCr查找表的自然感彩色融合算法。针对硬件特点,对融合算法进行优化,使720 pixel×576 pixel的单帧图像融合时间为9.18 ms,满足25 frame/s的实时融合要求;融合图像呈现自然色彩,场景中的热目标易于辨识。文中对共光轴融合系统设计、实现过程中相关问题的研究具有参考借鉴价值。     

基于分区灰度投影的稳像算法在卫星装配中的应用

针对传统电子稳像算法无法快速有效地消除视频图像随机抖动的问题,采用一种基于分区灰度投影的稳像算法,以确保机器人系统能够输出稳定连贯的卫星装配画面。对前后两帧视频图像进行划分并删除对比度低的子区域,利用间隔投影和互相关运算获取局部运动分量,通过基于平均误差门限的迭代步骤筛选后剩余的局部运动分量求解全局运动矢量;若判定存在低频扫描分量,还需对多帧图像的全局运动矢量作均值滤波处理。实验结果表明：基于分区灰度投影的稳像算法相比传统灰度投影法,在低对比度的自然场景图像和模拟装配图像中的稳像精确度分别提升119.1%和55.8%;同时执行时间只有块匹配算法的0.5%。能够有效消除随机抖动,快速输出稳定连贯的视频画面,保证机器人系统顺利完成卫星的地面装配工作。     

A method for measuring the dynamics of velocity vector fields in a turbulent flow using smoke image-visualization videos

A new optical method for determining flow-velocity vector fields that allows investigations of unsteady and fast processes is described. The method is based on measuring the displacements of turbulent structures, which are visualized in a light sheet, within a fixed time interval between consecutive video frames. The method was tested using measurements of pulsating-flow velocity. It was shown that the difference between the mean flow velocity in the investigated flow measured using this method and a hot-wire anemometer does not exceed 1%, while the deviation of the rms velocity pulsations is within 4.5%.     

单对立色流运动下的抠图跟踪

当目标与周围背景相似性较高,且目标运动形式复杂时,很难精确跟踪目标。针对传统目标跟踪方法因目标外观的不精确建模致使模型退化而产生漂移的问题,提出了一种单对立色流特征下的抠图跟踪方法。首先,根据彩色视频帧包含的丰富颜色信息,将单对立色颜色编码方式与Lo G兴趣点检测器结合,获取颜色兴趣点作为目标表示集合;其次,在目标运动预测阶段,通过估计相邻帧兴趣点的流匹配关系,获得预测帧的前景目标兴趣点估计;最后,利用获得的前景兴趣点估计进行抠图,重新划分当前帧的前背景标记,并对模型进行更新。该算法在Segtrack视频跟踪数据集上进行验证,定性定量分析了跟踪目标快速运动,目标形变和光照干扰下的跟踪效果。实验结果表明该算法可有效提高形变目标跟踪的准确性,优于当前目标跟踪的先进算法。     

Fast and Simple Motion Tracking Unit with Motion Estimation

Surveillance system using active tracking camera has no distance limitation of surveillance range compared to supersonic or sound sensors. However, complex motion tracking algorithm requires huge amount of computation, and it often requires expensive DSPs or embedded processors. This paper proposes a novel motion tracking unit based on different image for fast and simple motion tracking. It uses configuration factor to avoid noise and inaccuracy. It reduces the required computation significantly, so as to be implemented on Field Programmable Gate Array(FPGAs) instead of expensive Digital Signal Processing(DSPs). It also performs calculation for motion estimation in video compression, so it can be easily combined with surveillance system with video recording functionality based on video compression. The proposed motion tracking system implemented on Xilinx Vertex-4 FPGA can process 48 frames per second, and operating frequency of motion tracking unit is 100 MHz.     

结合优化梯度滤波与投影不变的航拍视频配准

考虑航拍视频配准中Harris检测的角点位置易发生偏移和产生伪角点以及匹配点精度不一致等问题,提出一种基于优化梯度滤波和投影不变的航拍视频配准算法.首先,提出一种基于优化梯度滤波的Harris检测器来实现角点的精确定位,并选择在尺度变化下局部区域最稳定的角点作为待配准点.然后,利用Delaunay三角网对待配准点进行初始匹配.最后,提出利用交比不变性方法提取精确度最高的少数匹配点,并对模型参数进行估计,完成视频帧间的配准.实验结果表明,利用提出的方法可实现帧间的精确配准,对分辨率为320 pixel×240 pixel的8组航拍序列的平均几何保真度误差仅为0.869,并能有效提取地面运动目标.     

结合多特征的视频镜头检测方法

针对视频镜头边缘检测准确率低的问题,提出了一种新的基于多特征的视频镜头检测算法。首先按时序读取多帧图像,并转换为灰度图;进一步将帧图像均匀分块,计算每个图像块的平均梯度,构造视频动态纹理;比较相邻帧视频动态纹理的相关性及两帧SIFT特征的匹配程度,根据匹配结果得出预检测结果;接下来与步长低于人眼刷新频率的下一帧动态纹理及SIFT特征相比较,得到最终的结果。通过对多组不同类型的视频数据进行实验,均能取得较高的召回率和准确率。该文算法对结构较复杂的渐变镜头进行检测,也能取得较高的检测准确率和召回率。     

DSP／BIOS环境下视频图像火焰检测算法实现

针对DSP／BIOS环境下在线视频图像火焰检测算法实现中存在的巨量数据运算和提高CPU利用率之间的矛盾,提出了以TMS320DM642为处理器的视频图像处理硬件系统构建,并在DSP／BIOS下设计了基于任务通信的实时视频图像火焰检测系统。此外,给出了基于颜色和火焰蔓延动态特性的检测规则,利用Ping-pong缓存技术实现了EDMA数据传输和CPU处理的并行操作,大大提高了CPU的利用率。试验结果表明,该算法不仅能有效地提取火焰区域,而且能使单帧图像的处理速度较无缓存技术和缓存技术分别提高60倍和1．55倍。     

A robust digital watermarking scheme for video copyright protection in the wavelet domain

This paper proposes a novel digital watermarking method for video based on a multi-resolution wavelet decomposition. The watermark used is a binary image. This watermark is embedded in the wavelet coefficients of the LH, HL and HH sub-bands of the second wavelet decomposition level by quantization. Every bit of the watermark is spread over a number of wavelet coefficients with the use of a key. The resulting watermarking scheme can be used for public watermarking applications, where the original image is not available for watermark extraction. We have tested the resilience of the watermarking algorithm against a series nine different attacks for different videos and improved the decoding BER by redundant embedding of the same watermark in different frames and by using an error correction code. Experimental results show that the embedded watermark is invisible and robust to attacks.     

Automated and semi-automated cell tracking: addressing portability challenges

Cell tracking is a key task in the high-throughput quantitative study of important biological processes, such as immune system regulation and neurogenesis. Variability in cell density and dynamics in different videos, hampers portability of existing trackers across videos. We address these potability challenges in order to develop a portable cell tracking algorithm. Our algorithm can handle noise in cell segmentation as well as divisions and deaths of cells. We also propose a parameter-free variation of our tracker. In the tracker, we employ a novel method for recovering the distribution of cell displacements. Further, we present a mathematically justified procedure for determining the gating distance in relation to tracking performance. For the range of real videos tested, our tracker correctly recovers on average 96% of cell moves, and outperforms an advanced probabilistic tracker when the cell detection quality is high. The scalability of our tracker was tested on synthetic videos with up to 200 cells per frame. For more challenging tracking conditions, we propose a novel semi-automated framework that can increase the ratio of correctly recovered tracks by 12%, through selective manual inspection of only 10% of all frames in a video.     

基于FMM方法的发动机失谐叶盘瞬态响应分析

基本失谐模型FMM(Fundamental Mistuning Model)是以一种基于失谐叶盘结构的单模态族子集为模态变换阵的减缩方法,采用该方法建立分析模型,可以容易识别出控制失谐响应的基本参数,基于该理论在不同描述框架和假设前提的若干典型简化形式,给出了若干分析应用实例,以及这些方法的瞬态响应分析。     

基于视频行场消隐期的大容量FLASH存储控制器

为克服传统大容量FLASH视频存储控制器时序设计复杂、缓存资源要求较高的缺点,设计了一种利用视频行场信号消隐期进行时序控制的FLASH视频存储控制器。该控制器基于FPGA时序设计,利用视频行场同步信号消隐期时间写入FLASH的读出和写入控制命令。由于无需缓存资源即可实现多级流水线的设计,提高了时序控制效率,简化了时序设计过程。基于Verilog硬件描述语言,设计了3级流水线和并行控制时序,数据达120 MB/s,实现了对2 048pixel×1 752pixel/15frames高速视频数据的实时存储与回放。仿真与实验结果均表明,系统时序设计正确,大容量FLASH阵列读写操作正常,可实现视频数据的采集、存储、回放等多种功能。     

一种基于TFRC的MPEG-4视频传输和评估系统

TFRC协议非常适合作为视频传输的协议,质量评估对于视频传输系统至关重要。本文提出了一种基于TFRC协议的实时视频传输系统及其在Evalvid改进框架下的评估模型。针对实际的视频文件来评估传输的质量和效率,分析了不同类型的视频帧在传输过程中的丢失情况,以及接收端的视频画面质量。NS2仿真实验结果表明该系统在复杂的网络环境中传输实时视频时,TFRC协议的友好性和平稳的发送速率使得在接收端能获得满意的视频质量。     

Omnifocus video camera

The omnifocus video camera takes videos, in which objects at different distances are all in focus in a single video display. The omnifocus video camera consists of an array of color video cameras combined with a unique distance mapping camera called the Divcam. The color video cameras are all aimed at the same scene, but each is focused at a different distance. The Divcam provides real-time distance information for every pixel in the scene. A pixel selection utility uses the distance information to select individual pixels from the multiple video outputs focused at different distances, in order to generate the final single video display that is everywhere in focus. This paper presents principle of operation, design consideration, detailed construction, and over all performance of the omnifocus video camera. The major emphasis of the paper is the proof of concept, but the prototype has been developed enough to demonstrate the superiority of this video camera over a conventional video camera. The resolution of the prototype is high, capturing even fine details such as fingerprints in the image. Just as the movie camera was a significant advance over the still camera, the omnifocus video camera represents a significant advance over all-focus cameras for still images.     

编码质量与容错性自适应的跳帧视频转码

为了使跳帧视频转码在有较好编码质量的同时兼顾转码后新码流的容错性,研究了一种编码质量与容错性自适应的跳帧视频转码方法。提出了基于最优编码质量的跳帧转码算法,依据质量影响系数,在所设计的滑动窗结构中进行了最优跳帧方案的选择,即计算各种跳帧可能性对编码质量的影响,最大程度地保留并编码运动强度较大的帧,提升编码质量。基于此算法,估计跳帧后的错误敏感度和错误影响系数,进一步提出了编码质量与容错性自适应的跳帧方法,使压缩效率和容错能力之间达到动态平衡。实验结果表明:丢包传输时,本方法可使再次解码视频的PSNR值提高0.05~1.2 dB,从而满足跳帧视频转码对帧率准、质量好、容错性强等方面的要求。