摘要
针对运动目标检测中噪声大、实时性差、提取目标不完整的问题,设计一种以TMS320DM642为核心的嵌入式实时运动目标检测系统。系统通过解码芯片TVP5150将相机采集到的图像数据转换到YUV颜色空间,利用占用内存小、运行速度快的视觉背景提取算法(ViBe)来提取运动目标信息,并辅以二值形态学运算来消除误检和忽略微小物体的运动,然后将图像数据传递给芯片SAA7121H编码输出。实验结果验证了系统设计的有效性。
Aiming at the problem of large noise, poor real-time performance and incomplete extraction of object in moving object detection, a method of designing an embedded real-time moving object detection system was discussed based on TMS320DM642.The camera collected image data were converted into YUV color space by decoding chip TVP5150, making use of the visual background extraction algoriihm （ViBe） to extract the information of moving object which occupied small memory and had fast running speed, and two valued morphological operations were used to eliminate the false detection and the motion of small objects,then the image data was passed to chip SAAT121H for coding output.The experimental results verify the effectiveness of the system design.
出处
《仪表技术与传感器》
CSCD
北大核心
2016年第11期54-57,共4页
Instrument Technique and Sensor
