摘要
由于复杂场景中具有动态背景,里面存在着大量的伪目标,加上小运动目标具有距离远、面积小、纹理颜色特征不明显、细节特征部分丧失等特点,因此针对该问题,提出了一种改进算法,首先将第一帧图像作为初始化的背景图像;然后,用当前帧与背景帧图像做差分,得到差值图像;最后对差值图像基于OTSU的自适应二值化处理,实现背景更新。实验证明,与传统背景差分算法相比,改进的算法具有明显的优势,能够快速、准确、完整地从复杂场景中完成小运动目标的提取。
Because of the dynamic background in the complex scene,there are many false targets in it,and the small moving target has small area. The texture color feature is not obvious,and the detail feature is partially lost. In order to solve these problems,an improved algorithm is proposed. Firstly,the first frame image is used as the initialization background image.Then the difference image between the current frame and the background frame image is made,and the difference image is obtained. Finally,the difference image is based on the adaptive two-value process based on OTSU to realize the back scene updating. The experimental results show that compared with the traditional background difference algorithm,the improved algorithm has obvious advantages,and it can quickly,accurately and completely conduct the extraction of small moving targets from complex scenes.
作者
孙文华
SUN Wenhua(Center of Network Information,Nanehang Institute of Teehnology,Nanehang 330099,China)
出处
《南昌工程学院学报》
CAS
2018年第3期60-64,共5页
Journal of Nanchang Institute of Technology
基金
江西省教育厅科技项目(GJJ171560)
南昌工程学院大学生创新创业训练计划项目(2017054)
关键词
复杂场景
运动目标
背景图像
背景更新
complex scenes
moving targets
background images
background updating
