基于部分特征卷积聚合的图像超分辨率重建方法OA

Objective With the development of deep learning,the performance of single-image super-resolution reconstruction networks has been significantly improved.However,these networks are often accompanied by high computational complexity and memory consumption,which hinders their deployment on edge devices or in real-time application scenarios.To address this issue,an image super-resolution reconstruction network based on partial feature convolutional aggregation is designed.Methods First,a convolution was used to extract features from low-resolution images.Then,the extracted features were processed by partial feature convolutional aggregation blocks.Finally,an upsampling block was used to reconstruct the high-resolution image.The partial feature convolutional aggregation block consists of two modules:the partial feature convolution module,which further reduces the number of parameters and the computational complexity of the network,and the feature aggregation module,which fully mixes convolved features with unconvolved features to improve network performance.Results Experimental results showed that,compared with advanced lightweight networks such as ECBSR and RepSR,the proposed network achieved improvements in both peak signal-to-noise ratio(PSNR)and structural similarity index measure(SSIM)on four standard benchmark datasets.Conclusion The proposed method achieves a better balance between the number of parameters,computational complexity,and performance.It is suitable for super-resolution reconstruction tasks on resource-constrained edge devices such as mobile phones,VR,and AR.