摘要
由于外界环境的复杂性,视频监控中检测出的运动目标往往轮廓缺损,并伴随阴影和噪声等问题。提出了一种方法用以改善轮廓、消除阴影和噪声。该方法首先利用五帧梯度图像的差分确定运动目标的变化区域,再与背景差分法的结果相或,获取了运动目标的完整轮廓。分析了阴影在HSV颜色空间的特性,用以消除阴影,并根据噪声的分布特性去除噪声。处理结果表明,处理速度为20 ms/帧,达到实时性要求,运动目标消除了阴影和噪声,具有较强的鲁棒性。
Due to the complexity of the external environment, moving target detected in video surveillance always possessed contour defects, as well as shadow and noise problems. This paper proposed a method to improve the contour, eliminate the shadow and noise. Firstly, five frame gradient image was used to determine the changing region of the moving object. Then it performed a logic aperation or with the results of background difference method to obtain a complete outline of the moving target. It analysed the characteristics of shadow in HSV color space , so as to eliminate the shadow,and to remove the noise in accordance with the distribution characteristics of noise. The larocessing results show that in the processing speed of 20ms per frame, it achieves real-time requirements .The moving target eliminates the shadow and noise, and has strong robustness.
出处
《电子设计工程》
2010年第8期130-133,共4页
Electronic Design Engineering
基金
国家自然科学基金资助项目(60772102)
关键词
运动目标检测
梯度图像
五帧差分法
背景差分法
阴影消除
去噪
moving target detection
gradient image
five-frame difference method
background difference method
shadow
