基于仿生系统的双视角运动目标检测方法  

Moving target detection method from double angle based on bionic system

作　　者：丁晓娜[1] 孟祥艳[1] 

机构地区：[1]西安工业大学电子信息工程学院,陕西西安710021

出　　处：《电子设计工程》2016年第11期18-20,25,共4页Electronic Design Engineering

基　　金：陕西省教育厅专项科研项目(14JK1341)

摘　　要：针对如何快速而精准地检测出运动目标问题,借鉴昆虫复眼在视觉信息处理上的独特技巧和原理,提出一种基于仿生视觉系统的运动目标快速检测方法。为确保检测效果,利用序列图像中目标和背景信息在空间和时间域上的灰度梯度变化差异及运动连续性的特点,首先借助复眼对背景侧抑制处理机制,采用DOG滤波和区域重构相结合的方法,减少噪声,增强目标信息;其次依据EMD结构和原理,设计出针对序列图像的相关型EMDs目标检测模型,检测出运动目标。仿真结果表明,该方法在滤除背景干扰同时,能快速而准确地检测复杂环境中的运动目标。How to detect moving targets accurately and quickly, which proposes a fast moving object detection method, based on the receiving and processing principles of visual information from insects＇ compound eyes. To get the reliability of target detection in the sequence of the target image, owing to the features that the gray gradients between object、background and noise is different obviously in space and time domain, and motion continuity, the moving target detection method based on background suppression later, the EMDs model is adopted to detect moving targets information accurately, and exclude the interference of background motion. Simulation results show that the method can detect moving targets accurately while filter out background noise reliably.

关 键 词：图像序列 侧抑制 EMDS 高斯差分滤波 目标检测 

分 类 号：TN919.8[电子电信—通信与信息系统]