摘要
针对传统的三帧差分法运动目标检测算法容易出现“空洞”现象,进而导致检测物体不完整的问题,提出一种改进的三帧差分法算法检测运动目标。改进算法先对连续的三帧图像进行预处理得到灰度图像,进而对其两两差分,获得两幅差分图像;利用阈值分割得到二值图像,经形态学的膨胀处理得到新的二值图像,通过逻辑与运算后,最后通过形态学操作的腐蚀处理,得到最终的运动目标图像。改进算法在真实场景下,通过对高速运动的乒乓球进行实验测试,并与传统的三帧差分法进行结果对比。实验结果表明:改进算法可以明显解决“空洞”现象导致的运动物体不全的问题,并对运动目标检测的更加准确和完整;整体计算较简单,可以实现实时性处理,且具有很好的鲁棒性。
The traditional three-frame difference method of moving target detection algorithm is prone to"holes",which leads to the problem of incomplete detection of objects.An improved three-frame difference method algorithm is proposed to detect moving targets.The improved algorithm preprocessed three consecutive frames of images to obtain gray images,and then differentiated them to obtain two difference images.The binary image is obtained by threshold segmentation,and the new binary image is obtained by morphological swelling processing.After logic and operation,the final moving target image is obtained by morphological corrosion processing.The improved algorithm is tested on high-speed ping-pong in real scenes and compared with the traditional three-frame difference method.Experimental results show that the improved algorithm can obviously solve the problem of incomplete moving objects caused by"void"phenomenon,and detect moving objects more accurately and completely.The overall calculation is simple,can realize real-time processing,and has good robustness.
作者
丁文龙
尹朵
邱崧
DING Wenlong;YIN Duo;QIU Song(Institute of Communication and Electronic Engineering,East China Normal University,Shanghai 200241,China)
出处
《智能计算机与应用》
2022年第3期180-182,共3页
Intelligent Computer and Applications
基金
上海市科委人工智能专项资助(19511120800)
关键词
运动目标检测
三帧差分法
空洞现象
图像处理
阈值分割
moving target detection
three frame difference method
void phenomenon
image processing
threshold segmentation
