【作者】 常洪花；

【导师】 张建奇；

【作者基本信息】 西安电子科技大学 ， 物理电子学， 2005， 硕士

【摘要】 本文以背景杂波为研究对象，论述了背景杂波的量化描述方式，分别讨论了背景杂波对人眼视觉和机器视觉成像系统目标获取性能的修正。在人眼视觉成像系统中，基于人眼集中观察图像边缘区域的基本特性，提出了三个图像表征尺度：整体背景杂波尺度、局部背景杂波尺度和图像特性尺度。并在此基础上，提出一种对NVESD经典人眼目标获取性能模型进行修正的新方法。所建立的修正模型即保留了原模型的简易性，又充分体现了背景杂波的干扰效应。在机器视觉成像系统方面，以系统的信号干扰比(SIR)为理论基础，引入巴特沃斯模型量化描述背景杂波的功率谱密度，并定量描述几种经典的图像增强算法对杂波的抑制作用，建立了充分考虑背景杂波效应的机器视觉目标探测性能。更多还原

【Abstract】 This paper discusses how the background clutter in the scene affects the target acquisition performance of human vision and machine vision based IR imaging systems with typical methods for quantifying background clutter as the research object respectively. To model the target acquisition capability of an IR imaging system operated by a human, one should take into account how the clutter in the scene affects the human performance. Based on the human’s propensity to focus on edges and its adaptability to scale, we present a global clutter metric, a local clutter metric and a target metric. Using them we propose a new method for modifying classical models of human target acquisition. The modified model takes account of the background clutter effects without markedly increasing the complexity of the classic model. In terms of machine vision based systems, background clutter greatly reduces the effective SIR, thereby increasing the false alarm rate or decreasing the probability of detection. At the same time, more and more complex algorithms are adopted in the detection procedure to improve SIR and increase probability of detection. The detection performance in a sense of probability of detection is highly related to the complexity of the background clutter and the applied algorithm. Based on the Butterworth model for quantifying power spectral density (PSD), we present a robust performance evaluation technique with emphasis on the quantitative description of the background clutter and the detection algorithms.更多还原