【作者】 颜益；

【导师】 洪景新；

【作者基本信息】 厦门大学 ， 信号与信息处理， 2009， 硕士

【摘要】 红外技术是20世纪初新出的一种不可见光技术,目前已被广泛应用于军事和民事领域,如红外探测,红外监视等。而红外探测焦平面作为红外技术的核心部分,起着关键的作用。近年来,由于非制冷微测辐射热计型红外焦平面阵列的制造技术有了重大突破,使得红外系统的体积变小、重量变轻、价格不断降低,极大地推动了红外系统在民事用途的各个领域中的推广。我国的红外技术研究起步比较晚,与先进的国家还有较大的差距,因此研究红外技术显得紧迫且有意义。红外焦平面上存在的非均匀性是目前大家广泛讨论的问题,本文先简单介绍目前红外热成像技术的发展状况,及一些红外探测器分类。分析红外探测器非均匀产生的原因,对目前国内广泛的应用两类非均匀校正算法做了深入的研究,即基于定标的非均匀校正算法和基于场景的自适应算法。在此基础上提出了一种改进的神经网络自适应算法,并将随机信号处理中收敛因子的估算应用到该算法。其次,为了下一步在硬件上实现神经网络自适应算法,本文在FPGA开发板上设计了一套能存储两帧红外图像的存储系统。其主要存储器是该开发板自带的DDR SDRAM。另外,对红外探测器的特性做了详细的介绍,在常用图像增强算法的基础上,结合实验室实际的需要,选择了一种线性增强算法和自适应计算均值相结合的算法,并在此算法上做了一定的改进,对探测器输出的数据做准适应的图像增强处理,并在FPGA上将该算法实现。实验结果表明,本文的设计都实现了预期的目标。更多还原

【Abstract】 Infrared technology is a new invisible ray technology in the beginning of the 20th century and currently it is widely applied in the military and civilian field,such as infrared detecting,infrared stakeout and so on.Infrared focal plane plays a key role as the core of the infrared technology.Recently,as the manufacture technology of uncooled microbolometer infrared focal plane array got a great breakthrough,it made infrared system’s weight more lighter,its volume more smaller and its price more cheaper.It does greatly accelerate the promotion of the infrared system to be applied in every aspect of the civilian field.The study of the infrared technology began relatively late in our country,so there is still large gap when compared with the developed countries.Therefore,it is urgent and meaningful to do the study.We all know that the non-uniformity in IRFPA(Infrared focal plane array) is widely discussed for the moment.Firstly,the paper plans to briefly introduce the development status of the infrared technology and some sorts of infrared imaging system in modern times.By the way,the paper also analyses the causes of non-uniformity in IRFPA,and studies the two sorts of the non-uniformity correction algorithms in IRFPA in the country.For example:the non-uniformity algorithm on the base of scaling and scene.On this basis,the paper brings forward an improved adaptive algorithm of NN(Neural Network).Secondly,the paper devises a suit of infrared image memory system that could put in two frame images on FPGA in order to run the adaptive algorithm of NN,and its main memory is DDR SDRAM which is owned by the development board.Finally the paper does detailed presentation of the characteristic of the infrared thermal imager that is often used by the lab.On the basis of the common image enhancing algorithm,when considering the requirements of our lab,the paper selects an algorithm that combines a linear enhancement algorithm and an adaptive calculation of the mean value algorithm,carries out the image enhancement processing with the output data from the thermal imager,and runs the algorithm on FPGA.The experiment indicates that all the schemes in the paper achieve the desired objectives.更多还原