低信噪比图像目标检测技术研究——Ⅰ.数据流程与背景感知  

Approach to Target Detection Technique in Low Signal to Noise Ratio ImageⅠ: Data Flow Chart and Background Estimating

作　　者：胡增成 庾农 

机构地区：[1]驻上海航空电子公司军事代表室,上海200233

出　　处：《航空电子技术》2007年第4期27-30,51,共5页Avionics Technology

基　　金：国家自然科学基金(10477019);上海市自然科学基金(05ZR14134)

摘　　要：提出了一种将背景区域感知与目标特征分析相结合的自动提取图像目标的层次化数据处理流程;探讨了线性预测估计和非线性均值滤波两种常用背景估计技术和自适应空间滤波的形态学背景感知方法。实验测评表明,形态滤波算法对背景的自适应感知能力强,能真实贴切地反映图像背景的起伏变化规律。This paper presents a multi-stage data processing procedure that can be used for automatic image target detecting through combining background perception with target character analysis. In this paper we approach two kinds of common background estimate techniques, including linear predictive estimating and nonlinear mean filtering, and morphological estimating algorithm for adaptive space filtering. Experimental results show that the adaptive awareness of the morphological estimating algorithm is most effective. Its output information can be consistent with the fluctuant change status of image background.

关 键 词：计算机视觉 数据流程 参数估计 噪声抑制 目标检测 

分 类 号：TN957.52[电子电信—信号与信息处理]