基于压缩传感的超分辨率红外成像研究

Research on Super Resolution Infrared Imaging by Compressive Sensing

为解决红外图像系统复杂度与成像分辨率之间的矛盾,采用压缩传感(compressive sensing,CS)理论对红外成像系统进行研究。通过对原始红外图像进行稀疏化,构造基于高斯随机噪声的测量矩阵,实现对目标的压缩感知,以较少数目的测量信号表示目标,获取目标的稀疏表达,基于对目标的稀疏表达,构造基于正交匹配追踪的重构算法对目标信号进行重构,实现以较少的测量信号构造较高分辨率的图像。在几种典型红外目标图像上的分析表明,压缩传感理论可实现对目标的超分辨率成像,以较低分辨率的传感器获得较高分辨率的目标信息,重构出的目标红外图像与相应高分辨率传感器所获得的图像之间误差较低。

To solve the conflict between complexity and image quality of infrared imaging systems,the theory of compressive sensing was applied to research the infrared imaging systems.The sparse representation was obtained by projecting the original infrared image to a special space.To position the sparse of the original image,a measure matrix was constructed based on Gaussian noise distribution,then a measurement with less elements of the target can be obtained.A reconstruction method based on orthogonal matching pursuit was used to reconstruct the original infrared image from the measurement.So far super-resolution can be implemented by the sensor with low resolution.The results on several typical infrared images show that super-resolution imaging can be realized by compressive sensing theory,high resolution information can be sampled by low resolution sensors,and the mean square error was low between the reconstructed image and original image.