光学学报, 2009, 29 (12): 3404, 网络出版: 2009-12-23
基于小波多尺度和最大似然法的红外目标探测
Infrared Target Detection Based on Wavelet Multi Scale and Maximum Likelihood
探测概率 小波多尺度边缘检测 最大似然法 信杂比 detection probability wavelet multi-scale edge detection maximum likelihood signal-to-clutter ratio (SCR)
摘要
由美国夜视和电子传感器管理局(NVESD)研究提出的经典视场目标探测模型,未能反映出信杂比(SCR)在探测过程中的影响程度。采用小波多尺度边缘检测法来仿效视觉系统,将边缘概率(POE)与经小波多分辨分析法改造后的雷诺一致性方程的均方根(RMS)结合,得到基于小波多尺度POE算法。再利用最大似然法对目标和背景分类探测进行处理,从而获得目标模式识别的最大似然概率;将信杂比计算引入红外目标探测中,推导出了高、中、低信杂比下的目标探测、分类和识别预测方法,解决了红外目标探测过程中与信杂比的关系。
Abstract
The classical NVESD view-field searching for target detection probability couldn′t reflect signal-to-clutter ratio (SCR) influence on detection. The wavelet multi-scale edge detection is adopted to emulate vision system,confining probability of edge (POE) with the root meam square (RMS) of Reynold identical equations which is revised by wavelet multi-resolution analysis,to derive a new algorithm based on wavelet scale. Then the maximum probability was utilized to process a target and its backgrounds detection,so as to derive the maximum probability for target pattern identification. The SCR calculation was introduced into the infrared target detection algorithm to derive the prediction for target detection,classification and identification at high,mediate and low SCR conditions,so that the relation between infrared target detection and SCR is revealed.
李朝晖, 王冰, 陈明. 基于小波多尺度和最大似然法的红外目标探测[J]. 光学学报, 2009, 29(12): 3404. Li Zhaohui, Wang Bing, Chen Ming. Infrared Target Detection Based on Wavelet Multi Scale and Maximum Likelihood[J]. Acta Optica Sinica, 2009, 29(12): 3404.