基于GAN和DWT的船舶红外偏振图像超分辨方法

Super resolution of ship infrared polarized image based on GAN and DWT

针对海面船舶红外偏振成像分辨率低、细节不清的问题,提出一种小波变换与生成对抗网络结合的方法提高图像分辨率。使用纯卷积神经网络模型(ConvNeXt)改进超分辨网络(SRGAN),采用非局部均值对原始低分辨船舶红外偏振图像进行去噪,用改进的SRGAN对低分辨率图像进行初始超分辨率,使用二维离散小波变换提取初始超分图像的细节信息,最后将细节信息通过小波逆变换与原始低分辨率的船舶红外偏振图像融合。与传统超分辨方法相比,本文方法得到的超分辨图像的峰值信噪比和结构相似性有明显提升。本文实现了红外偏振图像超分辨率与细节信息融合的同时进行,得到的超分辨率图像既保留了原始图像的红外偏振信息,又融合了高分辨率的细节信息。

Aiming at the problem of low resolution and unclear details of infrared polarization imaging of sea ships,a method combining wavelet transform and generative adversarial network is proposed to improve image resolution.Firstly,the pure convolutional neural network model(ConvNeXt)is used to improve the super resolution network(SRGAN),and the original low resolution ship infrared polarimetric image is denoised by using non local mean.Then,the low resolution image is initially super resolved with the improved SRGAN,and the detail information of the initial super resolved image is extracted using a two dimensional discrete wavelet transform.Finally,the detail information is fused with the original low resolution ship infrared polarization image through the inverse wavelet transform.Compared with the traditional super resolution method,the peak signal to noise ratio and structural similarity of the super resolution image obtained by the proposed method are significantly improved.In this paper,the infrared polarization image super resolution and detail information fusion isachieved at the same time and the obtained super resolution image not only retains the infrared polarization information of the original image,but also fuses the high resolution detail information.