红外成像系统目标获取性能预测方法研究

发布时间：2018-03-28 12:34

本文选题：目标获取性能预测　切入点：人眼视觉特性　出处：《中国科学院长春光学精密机械与物理研究所》2017年博士论文

【摘要】：红外成像系统性能评价技术对于红外成像技术的发展和应用具有重要意义。红外系统的评价方法主要分为外场测试法和预测法。相对于外场测试法,预测法在实验室条件下即可完成,其预测过程简便且成本低廉,成为当今最重要的性能评价方法。然而,现有的预测方法存在一定的缺陷:成像仿真方法依赖数字建模技术,且无法对输出的仿真图像给出准确的定量评价;性能理论模型的实验室测试过程无法准确反映外场的实际情况且严重依赖主观测试;已有的杂波度量方法对人眼视觉特性的重视不够,无法准确的反映背景杂波对目标获取性能的影响。基于上述背景,本文从三个方面开展了研究:改进现有的成像仿真和性能理论模型方法;提出带有人眼视觉特性的背景杂波量化方法;将杂波和目标区域特征相融合,提出基于图像的目标获取性能预测方法。其具体工作主要包括:(1)基于评价仿真模型的目标获取性能预测方法。分析比较了成像仿真方法和性能理论模型,指出它们的缺陷和不足,并采用模块化的思想将二者结合,建立评价仿真模型。利用所建的评价仿真模型,预测了两个典型红外成像系统的目标获取性能,比较二者性能的差异。为建立更加准确的模型,建立了基于方向性的混叠效应评价模型,并据此定量评价混叠效应对红外焦平面阵列成像质量的影响。(2)基于背景杂波的目标获取性能预测方法。分别提出了基于最佳阈值的背景杂波量化尺度(ORT杂波尺度)和基于选择性视觉注意的背景杂波量化尺度(SVA杂波尺度),这两种方法将人眼的适应性和选择性引入到背景杂波量化中,并通过均方根误差曲面寻找最低点的方法模拟人眼对杂波的响应,据此给出量化结果。将杂波量化结果作为预测因子带入到性能预测模型中,实现基于背景杂波的目标获取性能预测,利用Search_2数据库和相关性分析方法评价预测结果的准确性,结果表明基于该杂波尺度的性能预测结果准确度较高。设计三个实验,用以分析目标区域特征(对比度、大小和位置)对背景杂波性能预测准确度的影响,探讨预测方法的适用范围。(3)基于图像的目标获取性能预测方法。基于what通路和where通路理论,提出了基于目标区域特征的获取性能预测方法,并与背景杂波预测方法相结合,得到基于图像的目标获取性能预测方法。借助Search_2数据库,评价了该预测方法的准确度,并与其它预测方法进行了比较,表明该方法的有效性。
[Abstract]:The performance evaluation technology of infrared imaging system is of great significance to the development and application of infrared imaging technology. The evaluation methods of infrared imaging system are mainly divided into external field test method and prediction method. The prediction method can be completed under the condition of laboratory, and its prediction process is simple and low cost, so it has become the most important performance evaluation method nowadays. However, the existing prediction methods have some defects: the imaging simulation method depends on digital modeling technology. And the simulation image can not be evaluated accurately and quantitatively, and the laboratory test process of the performance theory model can not accurately reflect the actual situation of the field and rely heavily on subjective test. The existing clutter measurement methods pay less attention to the visual characteristics of human eyes and can not accurately reflect the impact of background clutter on the performance of target acquisition. In this paper, three aspects of research are carried out: improving the existing imaging simulation and performance theoretical models; proposing a quantization method of background clutter with human visual characteristics; combining clutter with target region features; The performance prediction method of target acquisition based on image is proposed. The main work includes: 1) the performance prediction method of target acquisition based on evaluation simulation model. The imaging simulation method and performance theory model are analyzed and compared. Their defects and shortcomings are pointed out, and the evaluation simulation model is established by combining the two models with the idea of modularization. The target acquisition performance of two typical infrared imaging systems is predicted by using the established evaluation simulation model. In order to establish a more accurate model, a direction-based aliasing effect evaluation model is established. The effect of aliasing effect on the imaging quality of infrared focal plane array (IRFPA) is evaluated quantitatively. (2) the target acquisition performance prediction method based on background clutter is presented. The ORT clutter scale based on the optimal threshold is proposed respectively. ) and SVA clutter scale based on selective visual attention. These two methods introduce the adaptability and selectivity of human eyes into background clutter quantization. The response of human eyes to clutter is simulated by the method of finding the lowest point on the root mean square error surface, and the quantization results are given. The clutter quantization result is introduced into the performance prediction model as a prediction factor. The performance prediction of target acquisition based on background clutter is realized, and the accuracy of prediction results is evaluated by using Search_2 database and correlation analysis method. The results show that the accuracy of performance prediction based on the clutter scale is high. Three experiments are designed. It is used to analyze the influence of target region feature (contrast, size and position) on the accuracy of background clutter performance prediction, and to discuss the applicable range of prediction method. The method of target acquisition performance prediction based on image is discussed. Based on the theory of what path and where path, This paper presents a performance prediction method based on target region feature, and combines it with background clutter prediction method to obtain performance prediction method based on image. The accuracy of this method is evaluated by Search_2 database. Compared with other prediction methods, this method is effective.
【学位授予单位】：中国科学院长春光学精密机械与物理研究所
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP391.41

【相似文献】