摘要
如何快速准确地检测出复杂背景中的运动物体,是图像处理中的一个重要问题。昆虫经长期对自然的适应,能够对视野内的细微运动做出迅速反应,并将目光锁定在运动物体上。初级运动检测模型模拟了蝇的视觉神经处理系统的功能,本文基于该初级运动检测模型,选取一阶无限脉冲响应数字滤波器作为延迟器,增加了高斯差分单元,在提取更多目标细节的同时,去除了信号的随机噪声,为相关模型的计算提供了更多有用的信息。本文基于Matlab软件平台,检测视频中的运动车辆。仿真实验结果表明,该方法不受复杂背景的干扰,精确地提取出运动目标。
How to detect a moving object fast and correctly in a complicated surrounding is one of the most important issues to be solved. Insects could catch a shift in their view field and make an action to prey or flee, which indicates they have peculiar capability to acclimatization. Fly visual system was modeled with an Elementary Motion Detector (EMD). We chosen first-order IIR digital filter as the time delay unit and made the Difference of Gaussians method to extract more details and eliminate random noise. We carried out simulation experiment based on software Matlab to detect a moving auto. From the result, this method can detect the moving object without the distraction by complicated environment.
出处
《光电工程》
CAS
CSCD
北大核心
2013年第8期13-18,共6页
Opto-Electronic Engineering
基金
国防科技重点实验室基金项目资助