South-Central MinZu University,a. College of Electronic and Information Engineering; b. Hubei Key Lab of Intelligent Wireless Communtication; c. College of Computer Science,Wuhan 430074,China
The cross-fusion of multi-resolution features of images has good potential to improve the reconstruction quality of compressed sensing images. An image compressive sensing reconstruction method based on multi-resolution feature fusion with Transformer has been studied. The measurement values of the input image are initially reconstructed to obtain a set of low-resolution images with reduced dimension. Then two channels are used to extract the features cross-fused of the different resolution images in parallel. Finally, the features output by the two channels are used to reconstruct the original image and reconstruct the downsampled image. Transformer is used to cross-fuse the features of multi-resolution images to better utilize the long-range correlation of the images. Extensive experimental results are compared to verify the effectiveness of the proposed method in balancing the complication of the network and improving image reconstruction quality.
DONOHOD L. Compressed sensing[J]. IEEE Transactions on Information Theory, 2006, 52(4): 1289-1306.
[2]
DUARTEM F, DAVENPORTM A, TAKHARD, et al. Single-pixel imaging via compressive sampling[J]. IEEE Signal Processing Magazine, 2008, 25(2): 83-91.
[3]
LIY, DAIW, ZOUJ, et al. Structured sparse representation with union of data-driven linear and multilinear subspaces model for compressive video sampling[J]. IEEE Transactions on Signal Processing, 2017, 65(19): 5062-5077.
[4]
ZHANGJ, ZHAOD, ZHAOC, et al. Image compressive sensing recovery via collaborative sparsity[J]. IEEE Journal on Emerging and Selected Topics in Circuits and Systems, 2012, 2(3): 380-391.
[5]
ZHANGJ, ZHAOC, ZHAOD, et al. Image compressive sensing recovery using adaptively learned sparsifying basis via L0 minimization[J]. Signal Processing, 2014, 103: 114-126.
[6]
AHSENM, VIDYASAGARM. Error bounds for compressed sensing algorithms with group sparsity: A unified approach[J]. Applied and Computational Harmonic Analysis, 2017, 43(2): 212-232.
[7]
ZHANNGJ, GHANEMB. ISTA-Net: Interpretable optimization-inspired deep network for image compressive sensing[C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 1828-1837.
ZHANGZ, LIUY, LIUJ, et al. AMP-Net: Denoising-based deep unfolding for compressive image sensing[J]. IEEE Transactions on Image Processing, 2020, 30: 1487-1500.
[10]
ZHANGJ, ZHAOC, GAOW. Optimization-inspired compact deep compressive sensing[J]. IEEE Journal of Selected Topics in Signal Processing, 2020, 14(4): 765-774.
[11]
SHIW, JIANGF, LIUS, et al. Image compressed sensing using convolutional neural network[J]. IEEE Transactions on Image Processing, 2019, 29: 375-388.
[12]
YAOH, DAIF, ZHANGS, et al. DR2-Net: Deep Residual Reconstruction Network for image compressive sensing[J]. Neurocomputing, 2019, 359: 483-493.
[13]
KULKARNIK, LOHITS, TURAGAP, et al. ReconNet: Non-iterative reconstruction of images from compressively sensed measurements[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE, 2016: 449-458.