Image Compressed Sensing Network Based on Texture-Structure Attention Mechanism

As widely acknowledged, compressed sensing plays a crucial role in image compression and reconstruction, with significant advancements in the field. However, challenges persist, particularly in recovering intricate texture details at low sampling rates. This paper introduces a texture-structure attention mechanism-based image compressed sensing network. This network features a dual-path architecture: the texture path focuses on restoring image texture details, while the structure path aims at reconstructing the image's main structural components. The texture-structure attention mechanism integrates structural path features into the texture path, enhancing the network's focus on regions with dense texture information to improve overall texture detail reconstruction. Experimental results on the SET11 dataset show that the proposed method has higher reconstruction accuracy than several advanced deep learning-based and traditional iterative methods.