红外图像的目标检测、识别与跟踪技术研究

【作者】 姜锦锋；

【导师】 郭雷；

【作者基本信息】 西北工业大学 ， 模式识别与智能控制， 2004， 硕士

【摘要】 随着精确制导武器的发展，对精确制导信息处理技术的要求越来越高，它已成为制约武器发展的关键因素之一。由于红外成像制导具有诸多优点，已成为当今世界精确制导技术发展和研究的主流和方向。本文根据红外成像制导的基本特征、制导信息处理方法研究的发展现状、存在的主要问题以及工程实际应用的基本要求，在红外图像预处理、图像分割、目标识别和跟踪等方面，进行了系统地研究。本文的主要内容如下： (1)结合热成像技术原理对红外图像的特征进行了分析，重点研究了红外图像的灰度直方图，提出了隔行／列直方图统计的思想。以灰度直方图作为统计手段和依据，提出了自适应灰度分段线性变换的图像增强算法。 (2)研究了一种用于红外成像制导信息处理的自适应、多尺度、抗噪声小波边缘检测方法。该方法利用小波变换的自动变焦的多分辨分析思想，构造了适合的噪声中边缘检测的多分辨率高斯梯度算子。实验结果表明，该方法是一种快速、有效的自动化目标检测方法。 (3)在模式识别理论的基础上，探讨了图像识别系统，研究了不变量(不变矩、圆度)在目标识别中的应用，并针对圆度法用于直升机识别给出了应用实例，提出拓扑思想在识别中应用的方案，并针对坦克提出了一种命中点选择方案和算法。 (4)介绍了成像跟踪常用的两大方法，波门法、匹配法，并在波门法中提出了基于命中点选择的命中点跟踪法，研究了跟踪预测算法的应用，针对红外图像目标提出了一种卡尔曼滤波改进算法。更多还原

【Abstract】 With the development of precision guided weapons, the requirements on information processing techniques for precision guidance is becoming higher and higher. This has become a key factor, which restricts the development of the weapon. Since IR imaging guidance possesses many advantages, it has become the trend and direction for current development and study of precision guidance technologies worldwide. On the basis of the fundamental characteristics of IR imaging guidance, the current development status of guidance information processing methods, the main problems existed as well as the basic requirements on its practical engineering applications, this paper has made systematic in depth studies on various aspects as preprocessing of IR images, target image segmentation , target recognition and target tracking. The main achievements of this paper are as fellows:(1)The thermal imaging theory and the characteristics of infrared image are analyzed in this dissertation. Image histogram is mainly discussed and an interlace row/column histogram statistic method is proposed. Based on histogram research, two novel image enhancement algorithms are put forward. It is adaptive piecewise linear transform.(2)A Wavelet edge detection method which is self-adaptive, multi -scale and anti-jamming has been studies for use in formation processing for IR imaging terminal homing guidance. By means of multi-resolution analysis concept for auto-zooming of wavelet transformation, a multi-resolution Gaussian gradient operator which is suitable for noisy edge detection is configured .The experiment results show that this method is an automatic rapid and effective target detection method.(3)ATR is studied in this paper both invariant moments to round degree. And , against the helicopter the result of test is given. The idea of topological is introduced in the fried of ATR, the hitting-point choice on the basis of characteristic points is studied and aiming at the tanker a applied algorithm is proposed..(4)It studies the target tracking, the hitting-point tracking is proposed .What’s more, based on the comparison and analysis of the various algorithms of tracking andtarget prediction, a new Kalman filter algorithm is presented.更多还原