Super Resolution Reconstruction of Images Based on Interpolation and Full Convolutional Neural Network and Application in Medical Fields
Authors:
Open Access
- 23 December 2019
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 186470-186479
Abstract
The traditional image to enlarge algorithms include nearest neighbor interpolation, bilinear interpolation and high-order interpolation. In order to achieve super-resolution reconstruction of images, a new algorithm combining traditional algorithms and deep learning is proposed. The framework is divided into two parts. Firstly, the deep reconstruction of the low-resolution data is performed by the ability of deep learning to extract features automatically. Then, combining with the traditional interpolation reconstruction results, the deep learning algorithm is used again for training and learning, and finally the high-resolution reconstructed data is obtained. The algorithm is validated using an online public test data set. The results show that the algorithm has a significant effect on the MSE (mean squared error) and PSNR (Peak Signal to Noise Ratio). Compared with the traditional interpolation algorithm and the single deep learning algorithm, the proposed algorithm has higher performance. Moreover, the proposed algorithm is perfect for the reconstruction of the details, the outline is clear, and the high-quality image is obtained.Keywords
This publication has 24 references indexed in Scilit:
- Image Super-Resolution Using Deep Convolutional NetworksIEEE Transactions on Pattern Analysis and Machine Intelligence, 2015
- Spatial Pyramid Pooling in Deep Convolutional Networks for Visual RecognitionIEEE Transactions on Pattern Analysis and Machine Intelligence, 2015
- Behaviour Analysis of Multilayer Perceptronswith Multiple Hidden Neurons and Hidden LayersInternational Journal of Computer Theory and Engineering, 2011
- Edge-driven Image Interpolation using Adaptive Anisotropic Radial Basis FunctionsJournal of Mathematical Imaging and Vision, 2009
- Nonlocal Edge-Directed Interpolation2009
- Super-resolution through neighbor embedding2004
- Adaptively Quadratic (AQua) Image InterpolationIEEE Transactions on Image Processing, 2004
- New edge-directed interpolationIEEE Transactions on Image Processing, 2001
- Subpixel edge localization and the interpolation of still imagesIEEE Transactions on Image Processing, 1995
- Backpropagation Applied to Handwritten Zip Code RecognitionNeural Computation, 1989