【作者】 高洪；

【导师】 顾红；

【作者基本信息】 南京理工大学 ， 通信与信息系统， 2007， 硕士

【摘要】 在很多领域中，人们往往只对视频图像中的运动目标感兴趣。在这些应用领域中(例如：视频监控、基于视频的控制、人机界面和机器人技术等)，如何把运动目标从有干扰的背景中检测出来是需要研究的关键技术。本文研究的是基于BLACKFIN视频动目标检测系统的设计与实现。本视频动目标检测系统是利用ADV7181采集摄像头输入的标准视频信号(PAL制式)，通过ADSP-BF531的PPI口传输到SDRAM存储：由ADSP-BF531作为核心处理器完成视频动目标检测，产生视频字符叠加数据存储在DPRAM中；由FPGA完成视频字符叠加。本文首先给出了系统方案和硬件电路设计，讨论了各模块电路的测试，获得了操作DMA、PPI和存储单元等模块合适的配置参数。然后，详细描述了背景差分法的视频动目标检测的DSP和相关FPGA软件设计。具体方法是先计算背景图像与当前图像的灰度差，并对差分图像进行二值化，然后通过连通性分析检测动目标，并提取目标信息，生成视频字符叠加数据。通过系统联调表明这种背景差分法的视频动目标检测是有效可行的。更多还原

【Abstract】 People are usually interested in the moving objects in the video images. In these application fields such as video surveillance, video-based control, human-machine interface and robot technology, etc., how to detect moving objects from the background with interference, is the key technique which needs to find out. This dissertation discusses the design and the realization of video moving objects detection system based on BLACKFIN.In video moving target detection system, ADV7181 is used to sample standard video signals (PAL standard).The PPI port of ADSP-BF531 is used to transmit video image to SDRAM memory. Video moving target detection is processed by the ADSP-BF531 processor. The data for video character superposed are created and stored in the DPRAM. The superposition of video characters is completed by the FPGA. Systematic scheme and hardware design are discussed, modules of the system are tested, and suitable configuration parameters of DMA, PPI and Memory are obtained. Then, a detailed description of DSP and FPGA software design based on the background difference method is completed. The luminance difference between background and current images is calculated. Differential images are used to dectct the target and then connectivity analysis is performed. Target information is extracted to generate data for video superposition. The system debugging shows that the difference arithmetic of the video moving target detection is feasible and effective.更多还原