低照度CMOS与非制冷红外图像融合技术研究

发布时间：2018-11-15 12:24

【摘要】：在视频图像探测领域,普通CMOS相机体积小、质量轻,适合白天工作;ICCD内部存在微通道板结构,较适合在夜间工作;低照度CMOS能在昼夜工作且便于携带,在实际应用中具有独特的优势。本文针对CIS0431低照度CMOS相机与PIC0384型非制冷红外焦平面研制了图像融合系统电子组件部分,并对基于FPGA的低照度CMOS与非制冷红外实时图像融合算法进行研究。本文首先在课题组已有的红外图像处理系统的基础上,以FPGA作为为核心处理芯片,改进并设计低照度CMOS与非制冷红外图像融合硬件平台,并对硬件系统进行调试;其次,研究并实现低照度CMOS与非制冷红外图像预处理及双源图像配准,低照度CMOS图像预处理主要是在10~(-3)lux照度下消除高斯噪声,而红外图像预处理过程包括非均匀校正与灰度直方图增强图像对比度;再次,研究拉普拉斯金字塔图像融合算法,并在系统中实现改进的拉普拉斯金字塔分解图像融合算法;最后,进行成像实验,对系统的成像效果进行评估,在白天系统以低照度CMOS图像为主,当照度在10~(-3)lux至1lux时对低照度CMOS与红外图像进行融合,当照度低于10~(-3)lux时低照度CMOS相机已无法清晰成像,此时以非制冷红外作为系统的图像输入源,同时对本文提出的算法与加权平均融合算法进行比较并进行客观评价,实验结果表明,本文设计的低照度CMOS与非制冷红外图像融合系统能对图像进行实时融合且效果比较理想。
[Abstract]:In the field of video image detection, the ordinary CMOS camera is small in size, light in mass, suitable for daytime operation, and has a microchannel board structure in ICCD, which is more suitable for working at night. Low illuminance CMOS can work day and night and is easy to carry, so it has unique advantages in practical application. In this paper, the electronic component of image fusion system is developed for CIS0431 low illuminance CMOS camera and PIC0384 uncooled infrared focal plane, and the algorithm of low illuminance CMOS and uncooled infrared real-time image fusion based on FPGA is studied. Firstly, on the basis of the infrared image processing system of our research group, the hardware platform of low illuminance CMOS and uncooled infrared image fusion is improved and designed with FPGA as the core processing chip, and the hardware system is debugged. Secondly, we study and realize low illuminance CMOS and uncooled infrared image preprocessing and dual source image registration. Low illuminance CMOS image preprocessing is mainly to eliminate Gao Si noise under 10 ~ (-3) lux illumination. The infrared image preprocessing process includes non-uniform correction and gray histogram enhancement image contrast; Thirdly, the image fusion algorithm of Laplacian pyramid is studied, and the improved image fusion algorithm of Laplacian pyramid decomposition is implemented in the system. Finally, the imaging experiment was carried out to evaluate the imaging effect of the system. In the daytime, the low illuminance CMOS image was the main one. When the illumination was from 10 ~ (-3) lux to 1lux, the low illuminance CMOS was fused with the infrared image. When the illumination is below 10 ~ (-3) lux, the low illuminance CMOS camera can no longer be clearly imaged, and the uncooled infrared is used as the input source of the system. At the same time, the algorithm proposed in this paper is compared with the weighted average fusion algorithm and evaluated objectively. The experimental results show that the low illuminance CMOS and uncooled infrared image fusion system designed in this paper can fuse the images in real time and the effect is satisfactory.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】