Research on U-Net Based Imaging Algorithm for Compton Camera
Aiming at the possible artifacts of Compton camera in the case of sparse projection data imaging,the article designs an improved U-net convolutional neural network,which consists of two modules:the joint reconstruction module of the decoding layer and the high and low fre-quency display enhancement module.Firstly,the high and low frequency display enhancement module is introduced in each layer of the decoding layer,which introduces the attention into the frequency domain of the image to effectively enhance the expression ability of the feature infor-mation;and then the joint reconstruction module of the decoding layer is used to fuse the feature information of different decoding layers to reduce the feature dilution in the process of image reconstruction.The experiments show that the improved model achieves a maximum improve-ment of 0.05 and 2.651 dB over the base model in the SSIM and PSNR metrics,respectively.
NETL
NSTL
万方数据
