Detecting Roads in Stabilized Video with the Spatio-Temporal Structure Tensor

Video provides strong cues for automatic road extraction that are not available in static aerial images. In video from a static camera, or stabilized (or geo-referenced) aerial video data, motion patterns within a scene enable function attribution of scene regions. A “road”, for example, may be defined as a path of consistent motion — a definition which is valid in a large and diverse set of environments. The spatio-temporal structure tensor field is an ideal representation of the image derivative distribution at each pixel because it can be updated in real time as video is acquired. An eigen-decomposition of the structure tensor encodes both the local scene motion and the variability in the motion. Additionally, the structure tensor field can be factored into motion components, allowing explicit determination of traffic patterns in intersections. Example results of a real time system are shown for an urban scene with both well-traveled and infrequently traveled roads, indicating that both can be discovered simultaneously. The method is ideal in urban traffic scenes, which are the most difficult to analyze using static imagery.     

基于压缩域的视频检索系统模型

文章提出了一个基于内容的视频检索系统模型,该模型直接对压缩视频数据进行处理,可节约解压缩的时间,减少运算量,提高系统处理效率。文章对该模型的系统结构以及系统实现所需的一些关键技术进行了介绍。     

视频阵列处理器多层次分布式存储结构设计

随着视频编解码标准的不断演进,算法处理的数据量也随之剧增。多核结构并行化处理技术在提升算法计算速度的同时,使得存储结构成为了整个编解码系统性能的瓶颈。针对视频编解码算法访存的局部性、各算法之间数据交互频繁性、算法内部大量临时数据不交互性的特点,设计并实现了由私有存储层和共享存储层构成的多层次分布式存储结构。通过Xilinx公司的Virtex-6系列xc6vlx550T开发板对设计进行测试,实验结果表明,该结构在保持简洁性和可扩展性的同时,最高可提供9.73 GB/s的访存带宽,能够满足视频编解码算法数据访存的需求。     

采用八叉树和OSG分页结点的海量点云三维可视化

针对现有海量点云可视化方法存在索引构建时间长、内存占用大等问题，研究一种八叉树索引结合OSG分页结点的快速可视化方法，可在占用较小内存的基础上快速建立点云索引并实时调度。采用八叉树索引结构对海量点云进行数据组织，建立各层级的八叉树结点并以文件映射的方式分块保存，对结点文件重组织转换为支持OSG渲染引擎的多分辨率点云数据。采用基于OSG分页结点的实时调度技术，对海量点云进行高质量可视化。与目前两款主流的点云数据处理商业软件进行实验对比分析，结果表明所提方法具有索引建立速度快、内存占用小等优点，同时可视化交互更加流畅，适用于各种配置计算机下海量点云数据的调度管理与实时可视化。     

Parallel display of objects represented by linear octrees

The storage, display, and manipulation of three dimensional volumetric information requires large amounts of computing resources, both in terms of memory, and processing power. Most existing serial algorithms that display 3-D objects on a 2-D screen are found to be too slow to process the large amounts of volume data in a reasonable time. Hence, one way to increase the performance of the display algorithm is to process individual volume elements (voxels) in parallel. The first part of this paper presents a brief over view of the linear octree data structure which represents 3-D objects by an eight-way branching tree, while the second part focusses on the parallel display of such objects. We have shown that, for an object represented by a linear octree and enclosed in a 2ⁿ×2ⁿ×2ⁿ universe, the maximum number of voxels that can be processed in parallel is 3ⁿ, and the maximum number of time steps required to display such an object is 4ⁿ. This paper presents a set of formulae which identify the processing element (PE) as well as the time step in which a given linear octree node is processed. Similarly, a procedure which determines the locational code of a linear octree node which must be processed by a given PE, at some specific time step, is presented, along with a strategy for determining whether a PE is active or idle     

基于特征空间的人脸超分辨率重构

超分辨率图像重构是利用关于同一场景的多帧低分辨率图像重构出一幅具有更高分辨率图像的过程.传统的超分辨率图像重构算法是基于像素空间,通过利用高、低分辨率像素空间之间的映射关系来求解,具有计算复杂性高等缺点. 针对低分辨率人脸放大问题,提出了一个基于特征空间的人脸超分辨率图像重构算法.与传统算法相比,该算法不仅降低了计算复杂性,还具有更好的鲁棒性.     

Incremental Volume Rendering Using Hierarchical Compression

We present a new algorithm here for efficient incremental rendering of volumetric datasets. The primary goal of this algorithm is to give average workstations the ability to efficiently render volume data received over relatively low bandwidth network links in such a way that rapid user feedback is maintained. Common limitations of workstation rendering of volume data include: large memory overheads, the requirement of expensive rendering hardware, and high speed processing ability. The rendering algorithm presented here overcomes these problems by making use of the efficient Shear-Warp Factorisation method which does not require specialised graphics hardware. However the original Shear-Warp algorithm suffers from a high memory overhead and does not provide for incremental rendering which is required should rapid user feedback be maintained. Our algorithm represents the volumetric data using a hierarchical data structure which provides for the incremental classification and rendering of volume data. This exploits the multiscale nature of the octree data structure. The algorithm reduces the memory footprint of the original Shear-Warp Factorisation algorithm by a factor of more than two, while maintaining good rendering performance. These factors make our octree algorithm more suitable for implementation on average desktop workstations for the purposes of interactive exploration of volume models over a network. Results from tests using typical volume datasets will be presented which demonstrate the ability of the algorithm to achieve high rendering rates for both incremental rendering and standard rendering while reducing the runtime memory requirements.     

基于FPGA的实时视频处理平台设计

为了能够实时地采集、处理、显示视频,设计并实现了一种基于双PowerPC硬核架构的实时视频处理平台;用硬件实现视频的预处理算法,并以用户IP核的形式添加到硬件系统中,上层的视频处理软件程序则直接从存储器中调用预处理后的图像数据;重点介绍了在FPGA上构建双PowerPC硬核架构的硬件系统;采用乒乓控制算法缓存一行图像数据;用DMA的方式将图像数据保存在存储器中;以边缘检测作为视频预处理算法的一个实例,在平台上实现,实验结果表明,用本平台实现仅需40ms;本平台能够实时处理视频,具有较高的实用价值。     

ZLBM: zero level binary mapping technique for video security

In recent days, providing security to data is a crucial and critical task in many image processing applications. Specifically, video security is an important and demanding concept. For this purpose, some of the embedding, encoding and decoding techniques are mentioned in existing works, but it has some drawbacks such as increased time complexity, computational complexity and memory consumption. Moreover, it does not provide high security during video transmission. To overcome all these issues, a new technique, namely, Zero Level Binary Mapping (ZLBM) is proposed in this paper for video embedding scheme. The motivation of this paper is to provide high security during video transformation by using the video steganography technique. At first, the cover and stego videos are given as the inputs and it will be converted into the video frames for further processing. Here, the Fuzzy Adaptive Median Filtering (FAMF) technique is employed to remove the impulse noise in the video frames. Then, the pixels in the filtered frames are grouped by using the block wise pixel grouping technique. After that, the frames are embedded with the help of ZLBM technique and encoded based on the patch wise code formation technique. On the receiver side, the inverse ZLBM and block wise pixel regrouping techniques are applied to get the original cover and stego videos. The novel concept of this paper is the use of ZLBM and patch wise code formation techniques for video embedding and compression. The main advantages of the proposed system are high security, good quality and reduced complexity. The experimental results evaluate the performance of the proposed video embedding technique in terms of Peak Signal-to-Noise Ratio (PSNR), Mean Squared Error (MSE), Compression Ratio (CR), Bits Per Pixel (BPP) and Signal-to-Noise Ratio (SNR).     

An efficient memory control method for video and image processing in digital TV

High definition (HD) and ultra-high definition (UHD) digital TV require high-resolution images and lots of data transfers between processors and memory devices often become the bottleneck of the system. Video and image signal processing usually require blocks of square or rectangular shaped pixel data for signal processing. It requires frequent precharging and activating new rows, and results in extra latencies for reading and writing pixel data in memory devices. This paper proposes an efficient memory controller for video and image processing to reduce the latencies for reading and writing blocks of pixel data. The controller stores a frame of pixel data by distributing contiguous lines of pixel data to multiple banks in sequence. Its efficiency is enhanced more with an interface protocol such as AMBA AXI in which outstanding transactions are allowed. Memory controllers according to the proposed scheme are designed and the performance and the efficiency are compared with the previous works.     

基于自适应八叉树划分的高精度四面体可视化

利用八叉树结构将四面体数据转化为规则网格数据,能有效提高系统的交互性能.八叉树的划分层次越高,绘制效果越好,但数据的存储空间以及处理时间也将大幅增多.提出自适应的规则化表示方法来构建八叉树结构,改进原有的单一采样策略,并结合深度信息将采样结果转换成适用于GPU的八叉树纹理结构.然后采用光线投射算法来对体数据进行绘制,根据各区域深度不一的特点,提出了变步长的采样绘制策略.实验结果表明,本文方法降低了数据的空间存储量和处理时间,同时在绘制质量、绘制效率方面都得到了较大提高.     

基于FPGA的运动目标检测系统

视频图像处理要求高速运算能力,在处理技术不断提高和算法复杂度不断提升的情况下,并行处理的可编程逻辑器件的高速运算能力和可重复执行多任务的特性在视频图像处理领域得到了极大的发挥。与传统的串行处理DSP芯片为核心器件的视频图像处理方案相比,单片FPGA芯片和嵌入其内部酌NiosII软核处理器不仅能够达到运算速度的要求,而且成本更低、设计更简单。系统由I2C模块、视频译码模块、存储模块和检测模块组成,模块之间由Avalon总线链接。系统基于QuartusⅡ、MATLAB和ModelSim软件工具设计与仿真．实验结果表明能够达到预期的要求。     

基于VFW的远程视频捕获技术研究

多媒体技术在人们日常生活中的应用越来越广泛,图像、视频、音频等多媒体数据逐渐成为信息处理领域中主要的信息媒体形式。视频捕获技术是信息处理中的重要环节,研究该项技术具有重要的实用价值。文章提出一种基于VFW的远程视频捕获方法。该方法利用VFW捕获视频数据,采用H.263编码标准压缩视频数据,利用面向连接协议的流式套接字实现实时视频流的传输,结合多线程技术实现视频文件播放。然后,基于Windows操作系统设计实现了远程视频捕获系统。实验结果表明,该方法CPU占用率低、内存占用小,可靠性强,具有较好的应用价值。     

数字视频跟踪技术的理论前提研究

图像的颜色空间、AVI视频文件的结构和读取、BMP文件的结构和特点等,这些都是进行视频跟踪的前提。该文基于视频序列进行目标跟踪,选用Windows API的方法对AVI视频进行分解;在保存分解出来的图像帧时使用了便于处理的BMP格式;在对数字图像处理所基于的颜色空间的选择上,选用了既比较接近人眼处理速度又较快的HSI颜色空间。     

用于建立三维GIS的八叉树编码压缩算法

复杂的空间数据结构在三维GIS领域中占有突出的地位,它直接关系到GIS的功能和效率,为了有效地进行三维GIS大量数据的存储和管理,重点讨论了三维GIS栅格数据结构中的八叉树编码压缩技术,由于Morton码值的排序是实现八叉树编码压缩的基础,为此,根据Morton码排序的特殊性,提出了采用时间复杂度为O（n）的计数排序算法,使排序速度大为撇提高,在此基础上进行压缩处理,并对算法的时间及空间复杂度进行了分析,在PC机上进行的模拟实验结果表明,在目标复杂度一定的前提下,八叉树存储数据占用空间小（当分割阶次为9阶时,八叉树存储量只占栅格存储量的4.32%）,是一种较为理想的描述复杂海量地理空间数据的压缩结构。     

一种智能视频监控系统中的行人检测方法

在智能视频监控系统的行人检测中,目前使用的目标检测算法R-CNN和YOLO系列算法或速度较慢,无法满足实时性要求;或需要较大的GPU显存空间,难以部署。YOLOv3-tiny算法作为YOLO系列的精简版本,对设备要求较低、速度快,但该算法精度较低。本文通过调整YOLOv3-tiny算法的grid cell横纵方向数量、优化YOLOv3-tiny算法网络结构、聚类确定anchor的数量及尺寸,得到改进的YOLO-Y算法,并通过数据增强方法对训练数据集进行扩充。改进的YOLO-Y算法将mAP从90%提升到92%,Recall从95%提升到97%,检测速度达到26帧/s,占用约1 GB显存空间。实验结果表明改进的YOLO-Y算法显著提高了算法检测精度,具有实时性,且不需要太大的显存空间,满足大部分智能视频监控系统的要求。     

SurvSurf: human retrieval on large surveillance video data

The volume of surveillance videos is increasing rapidly, where humans are the major objects of interest. Rapid human retrieval in surveillance videos is therefore desirable and applicable to a broad spectrum of applications. Existing big data processing tools that mainly target textual data cannot be applied directly for timely processing of large video data due to three main challenges: videos are more data-intensive than textual data; visual operations have higher computational complexity than textual operations; and traditional segmentation may damage video data’s continuous semantics. In this paper, we design SurvSurf, a human retrieval system on large surveillance video data that exploits characteristics of these data and big data processing tools. We propose using motion information contained in videos for video data segmentation. The basic data unit after segmentation is called M-clip. M-clips help remove redundant video contents and reduce data volumes. We use the MapReduce framework to process M-clips in parallel for human detection and appearance/motion feature extraction. We further accelerate vision algorithms by processing only sub-areas with significant motion vectors rather than entire frames. In addition, we design a distributed data store called V-BigTable to structuralize M-clips’ semantic information. V-BigTable enables efficient retrieval on a huge amount of M-clips. We implement the system on Hadoop and HBase. Experimental results show that our system outperforms basic solutions by one order of magnitude in computational time with satisfactory human retrieval accuracy.     

MPEG-4系统中基于FPGA实现数据采集及预处理

介绍了一种基于DSP的MPEG-4视频压缩系统中,利用FPGA控制视频数据实时采集并对原始视频数据进行预处理的设计方案及实现,解决了原始视频数据格式与MPEG-4压缩算法不兼容的问题,从而提高视频压缩系统的整体性能。此设计充分利用FPGA的高速并行处理能力,通过合理的调度,以较少的存储资源实现数据处理与传输。     

Using difference intervals for time-varying isosurface visualization

We present a novel approach to out-of-core time-varying isosurface visualization. We attempt to interactively visualize time-varying datasets which are too large to fit into main memory using a technique which is dramatically different from existing algorithms. Inspired by video encoding techniques, we examine the data differences between time steps to extract isosurface information. We exploit span space extraction techniques to retrieve operations necessary to update isosurface geometry from neighboring time steps. Because only the changes between time steps need to be retrieved from disk, I/O bandwidth requirements are minimized. We apply temporal compression to further reduce disk access and employ a point-based previewing technique that is refined in idle interaction cycles. Our experiments on computational simulation data indicate that this method is an extremely viable solution to large time-varying isosurface visualization. Our work advances the state-of-the-art by enabling all isosurfaces to be represented by a compact set of operations.     

无线多媒体传感器网络中基于多类传感器信息的分布式视频处理

无线多媒体传感器网络中的多媒体传感器节点的功率、计算和存储能力及带宽资源受限,迫切需要低复杂度的视频处理技术对大数据量的视频进行压缩传输。基于来自网络中监控同一目标场景的多种类型传感器的信息,给出了一个新的无线多媒体传感器网络分布式视频处理系统框架;提出了一种基于多类型传感数据融合和多视角的GOP（group of picture）划分方法;在解码端,考虑单一视角视频序列之间较强的时间相关性,产生时间相关的边信息;利用来自多个视频传感节点的视频序列间的空间多视角相关性,产生多视角相关的边信息,并提供两种边信息的融合和选择机制,提高边信息的准确度和可靠性。最后仿真实验结果表明该方法的有效性和优越性。