关键词: 关键词: 超分辨率, 图像重建, 深度卷积神经网络, 级联, 深度边缘滤波器

Abstract: Abstract: Super-resolution technology has always been of concern. In order to further improve the resolution of existing image super-resolution methods, a high-performance deep convolution neural network (HDCN) was proposed to reconstruct a high resolution image in a fixed scale factor. By cascading several HDCN models, the problem that many traditional models cannot upscale images in alternative scale factors was solved. Deep edge-aware filter was implemented between every adjacent HDCN model. The implementation of edge-aware filter can limit error accumulation during repetitive upscaling and highlight the texture information. The super-resolution image reconstruction experiment was carried out on high-performance cascade deep convolution neural network (HCDCN) model with Set5 and Set14 database. It proves the effectiveness of introducing the deep edge-aware filter. The performance evaluation of HCDCN method and other image super-resolution method demonstrates the superior performance of HCDCN method.

Key words: super-resolution, image restruction, deep convolutional network, cascade, deep edge-aware filter