【作者】 朱洪波；

【导师】 王乘；

【作者基本信息】 华中科技大学 ， 系统分析与集成， 2011， 博士

【摘要】 在当今高度发达的信息时代,信息处理已成为一项关系到国计民生的重要任务。信息处理是一个对信息进行识别和分类的过程,它能有效提高信息的有效性和抗干扰能力,并改善信息的主观感官效果。经过处理的信息可以更好为人们的生产生活提供帮助和依据,帮助人们做出正确的分析和决策。多源图像融合就是这样一类综合多传感器的图像信息融合处理过程,它能去除干扰信息,增强图像的分辨和视觉效果。多源图像融合能对不同传感器所生成的图像进行更好的结合,得到目标场景更加优质、可靠的图像。多源图像融合已经成为当前图像处理领域研究的一个热点,并广泛应用于国防、医疗、航天、工农业生产、安全监测、抢险救灾、人民生活等国民经济建设的众多领域。抗干扰的多源图像融合利用了多传感器信息互补的优势,对多图像进行综合处理,根据融合要求的不同,凸显目标信息,生成一幅可用性更高的融合图像。本文分析了国内外图像融合相关领域的发展现状,研究了相关的理论和技术,对比了多种主流多尺度分析图像融合分解工具和融合算法,并探索了云雾成像的特性,从而总结出一套可以抗云雾干扰的多源图像融合模型,及其融合计算的公式,并仿真验证。论文提出了基于亮度变化测度的云雾检测算子。研究了云雾在可见光和红外光图像中成像的特性,总结出云雾成像规律,并采用局部梯度和局部方差联合的方法提出亮度变化测度的概念,描述云雾成像的区域特性。论文提出了基于亮度变化测度的可见光和红外光抗云雾融合方法。采用多尺度分析的方法,提出针对可见光和红外光图像的基于亮度变化测度的抗云雾干扰图像融合方法,并给出计算公式,并通过两组实验来比较和验证该方法的有效性、正确性和优越性。论文提出了基于亮度变化测度的多源融合模型。将可见光和红外光的两源亮度变化测度的图像融合方法扩展到三源图像融合,并用仿真模拟实验加以验证,再将亮度变化测度的图像融合模型推广到多源,在不限定图像源数量的情况下,给出了推广模型的融合计算公式,并加以分析讨论,与两源融合、三源融合的公式统一起来,形成一套完整、通用的计算公式。更多还原

【Abstract】 In today’s highly advanced information age, information processing has become a task of vital importance related to the national economy and the people’s livelihood. Information processing is a process of identifying and classifying information, and it can enhance the effectiveness of information and anti-interference capabilities, which can also improve the subjective senses of the information. And the processed information will be more helpful to the life and the production of people, and help them to make the right analysis and decision. Multi-source image fusion is an information fusion process, which can integrate multi-sensor images’ information, eliminate the interference signal, and enhance the resolution and the visual effects of fused images. Via multi-source image fusion, better images with high-quality and reliability target scenes can be generated from images producted by different sensors. Multi-source image fusion has become a hot research field of image processing, and widely used in many areas of national economic construction, such as national defense, medical, aerospace, industrial and agricultural production, security monitoring, disaster relief, people’s livelihood, and etc.The anti-interference multi-source image fusion takes the advantage of the complementary information from different sensors, integrates the images, highlights the target information, and generates a fusion image with higher availability according to a certain integration requirement. This paper analyzed the current research status, researched to the related theories and technologies, contrasted a variety of major multi-scale decomposition analysis tools and integration algorithms of image fusion, and explored the imaging properties of clouds, proposed a set of multi-source image fusion model which is anti-cloud interference, put forward the fusion calculation formula with verification. A cloud detector based on brightness variation metric was proposed. The paper explored the characteristics of clouds imaging in the visible and infrared images, summarized the law of cloud imaging and proposed the concept of the brightness variation metric combining the local variance with the local gradient, and the metric can represent the local characteristics of cloud imaging.The anti-cloud interference image fusion model for visible and infrared images based on brightness variation metric was proposed. By multiscale analysis the visible and infrared images were decomposed, and the decomposed componets were fused based on brightness variation metric respectively. Then through two sets of experiments we compared and verified the validity, accuracy and superiority of the model.The anti-cloud interference image fusion model for multi-sensor images based on brightness variation metric was proposed. The model of image fusion based on brightness variation metric for two sources is extended to three sources, and the extended model was verified through simulation experiments. And thenthe model of image fusion based on brightness variation metric was extended to multiple sources, that is, not limited to the number of image sources. The fusion formulas of the proposed new model were given, analyzed,and discussed, as a result, the new model unified with the fusion formulas of two sources and three sources and formed a set of complete, general computational formulas.更多还原