High-resolution enhanced cross-subspace fusion network for light field image superresolution

Shixu Ying, Shubo Zhou, Xueqin Jiang, Yongbin Gao, Feng Pan, Zhijun Fang

Abstract: Highlights•LF subspace features are progressively integrated with our cross-attention mechanism.•HR textures and edges are restored with the adaptive weight attention mechanism.•Our method can produce HR images with sharper edges and more faithful textures.