IOE OpenIR  > 光电技术研究所博硕士论文
Alternative TitleA Automatic Detection Method for Moving Target in Complex Environment
Thesis Advisor蔡敬菊
Degree Grantor中国科学院光电技术研究所
Place of Conferral光电技术研究所
Degree Discipline信号与信息处理
Keyword目标检测 特征提取 运动估计 块匹配 Fpga
Abstract在成像跟踪系统中,目标检测处于最底层,它是各种高级处理如目标识别、目标跟踪的基础。在复杂背景下,目标和背景灰度交错,具有双重复杂性,使得传统的基于灰度特征、区域特征、形状及纹理特征的检测方法难以对目标进行有效地检测,而运动特征是可利用的有效特征。 本文分析了目前比较典型的运动目标检测算法:光流法、背景减法和帧间差分法。在此基础上,提出了一种基于运动估计的运动目标自动检测方法,利用相邻两帧图像的运动信息,更新特征块提取阈值,实时地提取图像中最能体现背景运动特征的特征块,采用块匹配的运动估计方法获取各特征块的运动矢量,再依据所有特征块运动矢量的统计特性提取背景运动矢量,配准差分实现运动目标的检测。实验结果表明,此方法检出率高、通用性好,而且运算简单,易于实时实现。 本文将算法在现有的图像处理平台上成功实现,所有操作集成在Xilinx XC3S1000芯片上。对基于FPGA的目标检测系统的结构和软件框架进行了深入分析,结合算法的特点,提出了该系统的具体实现方案。在FPGA软件设计上,采用了模块化编程的思想,在模块中采用了乒乓操作、串并转换、流水线等思想来提高处理速度。系统仿真、调试通过,调试结果表明,该系统能够有效地检测到复杂背景下的运动目标,具有较好的场景适应性。
Other AbstractTarget detection is on the bottom layer in the imaging tracking system, which is the basis of all kinds of senior processing including target recognition and target tracking. In visually complex environment, those features, such as gray scale, texture and shape, can hardly be used to detect the moving object efficiently because of the dual complexity of target and background. Fortunately the motion characteristic is available and useful for detecting moving target. Based on the analysis and comparison to some classical target detection algorithms, such as optical flow, background subtraction and frame difference, an automatic detection method for moving target in complex environment was proposed. In this method, an adaptive threshold was adopted to extract the blocks which can stand for the motion characteristics of the background using the motion information of the two sequential frames of the image. Block matching was applied to each block to estimate its motion vector. And then the motion vector of the background can be achieved based on the statistic characteristic of the motion vectors of the blocks. At last, the moving target was gained after matching and difference operation. Experiments turned out that the algorithm works efficiently in detecting the moving target in complex environment. The algorithm was implemented successfully on the existing image processing platform. All operations were integrated in Xilinx XC3S1000. The system structure and software structure of the target detection system were analyzed deeply. And then the implementation scheme of the system was proposed in view of the characteristic of the algorithm. Modularization programming was used in FPGA software design and some strategies, such as ping-pong operation, serial to parallel conversion, pipelining, were adopted to quicken the processing. The system got through the debugging. Debugging results turned out that the system can successfully detect the moving target in complex environment and has good adaptability to the environment.
Document Type学位论文
Recommended Citation
GB/T 7714
徐功益. 复杂背景下运动目标的自动检测[D]. 光电技术研究所. 中国科学院光电技术研究所,2007.
Files in This Item:
File Name/Size DocType Version Access License
10001_20042801512705(5934KB) 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[徐功益]'s Articles
Baidu academic
Similar articles in Baidu academic
[徐功益]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[徐功益]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.