刘晓宇 于洵 丁良华 韩峰 龚昌妹

摘 要：在红外成像系统中，红外探测器的阵列大小和像元尺寸决定了红外图像的分辨率。由于红外探测器的制作工艺问题，很难通过增大阵列大小或减小像元尺寸的方式来提升红外图像分辨率。为了解决上述问题，将分块压缩感知理论应用到红外成像系统之中，编码模板位于前置红外镜头的像面，并对目标场景所成的像进行多次编码。编码后的像经过中继红外镜头被红外探测器所采集，采用OMP算法对采集到的信号以分块的方式进行重构，最终得到目标场景图像。对不同分块大小重构原始图像进行仿真，并将仿真的重构图像与降采样图像进行对比分析。仿真结果表明，分块大小越大，重构图像的峰值信噪比越高，重构时间越长。重构图像的质量明显优于降采样图像，可以实现低分辨率红外探测器重构高分辨率图像，为提高红外图像分辨率提供了新的方向。

关键词：压缩感知;分块重构;红外成像;像平面编码

中图分类号：TP391.41 文献标识码：A 文章编号：1001-5922（2021）04-0077-05

Abstract：In the infrared imaging system， the array size and pixel size of the infrared detector determine the resolution of the infrared image. Due to the manufacturing process of the infrared detector， it is difficult to increase the resolution of the infrared image by increasing the size of the array or reducing the size of the pixel. In order to solve the above problems， the block-compressed sensing theory is applied to the infrared imaging system. The coding template is located on the image plane of the front infrared lens， and the image formed by the target scene is coded many times. The encoded image is collected by an infrared detector through a relay infrared lens， and the collected signal is reconstructed in blocks using the OMP algorithm， and finally the target scene image is obtained. The original image reconstructed with different block sizes is simulated， and the simulated reconstructed image is compared with the down-sampled image. The simulation results show that the larger the block size， the higher the peak signal-to-noise ratio of the reconstructed image and the longer the reconstruction time. The quality of the reconstructed image is obviously better than that of the down-sam

0 引言

红外探测器能在低照度环境下探测到物体发出的红外辐射，具有抗干扰能力强、全天候工作的优点[1]。在高分辨率红外探测器制作工艺方面，西方发达国家较为成熟，并对该领域技术进行封锁。我国在高分辨率红外探测器技术方面起步较晚，并存在一定的技术瓶颈[2]，因此需要一种提升红外成像分辨率的方法。

2004年，Candes等人提出了压缩感知理论。……