摘要
提出用模式和模式分布理论分析活动图像边界的特征以及硬件检测、提取动目标轮廓的方案．在５点处理窗口中，分别检测出当前帧内的水平轮廓和垂直轮廓，再和前一帧的图像轮廓进行异或操作，检测出活动图像轮廓．给出了软件模拟结果，详细论述了硬件设计原理和方法．该检测方法具有一定的抗噪能力、运算量小、速度快、可实时工作、成本低等优点．
The edge signature of moving image is analyzed by means of the pattern and pattern distribution. The scheme for the detection and extraction of moving target outline is presented. In 5 point processing window, both the horizontal edge and vertical edge of present frame are detected and extracted. The edge signal of a moving picture is made by an exclusive OR operation between edge data and frame delayed edge data. Simulation results are given. The principle and method of hardware design are presented in detail. The scheme is of noise immunity, less operand, high speed, low cost and good real time performance.
出处
《华中理工大学学报》
CSCD
北大核心
1997年第11期54-56,共3页
Journal of Huazhong University of Science and Technology
关键词
边缘检测
目标识别
图像处理
噪声
交通管理
edge detection
feature extraction
target recognition
real time image processing .
