Tianyi Mao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Tianyi Mao has a total of 50 co-authors. Here are the top co-authors by number of shared publications:
Shujin Zhu
4 shared publications- Xu Ma
Beijing Institute of Technology
4 shared publications - Gonzalo R. Arce
University of Delaware
3 shared publications - Qian Chen
School of Electronic and Optical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
3 shared publications Weiji He
Jiangsu Key Laboratory of Spectral Imaging and Intelligence Sense, Nanjing University of Science and Technology, Nan Jing, China
3 shared publicationsXiubin Dai
3 shared publications
Latest Publications
- 5 November 2026
- Published by Optica Publishing Group in Applied Optics
- Vol. 65 (32) , 11502
Three-dimensional residual projection completion for coded-aperture cone-beam computed tomography
Coded aperture cone-beam computed tomography reduces X-ray exposure by blocking a subset of detector-view rays, but the resulting interleaved missing samples make direct volumetric reconstruction computationally demanding and prone to artifacts. This paper presents a projection-domain reconstruction framework in which a 3D U-Net learns residual corrections to complete the coded projection tensor before Feldkamp–Davis–Kress reconstruction. The network takes the measured coded projection, an interpolation-based base estimate, and the binary aperture mask as input, and it predicts a bounded residual correction only for unmeasured samples while preserving measured data exactly. A training objective combining missing-region fidelity, seam intensity consistency, and seam gradient consistency is used to suppress discontinuities around coded-aperture boundaries. Experiments on 188 projection volumes show that the selected loss composition improves projection completion and yields reconstruction results close to the reference FDK images. Over 38 testing cases, the mean Ref-vs-Pred reconstruction PSNR is 44.6 dB and the mean SSIM value is 0.98.
- 5 November 2026
- Published by Optica Publishing Group in Applied Optics
- Vol. 65 (32) , 11502
Three-dimensional residual projection completion for coded-aperture cone-beam computed tomography
Coded aperture cone-beam computed tomography reduces X-ray exposure by blocking a subset of detector-view rays, but the resulting interleaved missing samples make direct volumetric reconstruction computationally demanding and prone to artifacts. This paper presents a projection-domain reconstruction framework in which a 3D U-Net learns residual corrections to complete the coded projection tensor before Feldkamp–Davis–Kress reconstruction. The network takes the measured coded projection, an interpolation-based base estimate, and the binary aperture mask as input, and it predicts a bounded residual correction only for unmeasured samples while preserving measured data exactly. A training objective combining missing-region fidelity, seam intensity consistency, and seam gradient consistency is used to suppress discontinuities around coded-aperture boundaries. Experiments on 188 projection volumes show that the selected loss composition improves projection completion and yields reconstruction results close to the reference FDK images. Over 38 testing cases, the mean Ref-vs-Pred reconstruction PSNR is 44.6 dB and the mean SSIM value is 0.98.
- Open Access
Coded Aperture Optimization in X-Ray Computed Tomography via Sparse Covariance Matrix Estimation
Coded aperture X-ray computed tomography (CAXCT) measures coded X-ray projections to reconstruct the inner structure of an object. Coded apertures, which determine the point spread function, can be designed to improve the reconstruction quality, but most approaches are computationally expensive, leading to very small images. In this paper, a sparse covariance matrix estimation approach is introduced to minimize the information loss sensed by projections corresponding to large tomographic images. The covariance matrix representing the map of the overlapping information of the projections is obtained by using block matrix multiplication and sparse estimation. A heuristic variant algorithm with a noise factor is presented to search the combinations of D projections leading to maximum non-overlapping information acquisition, where D is the number of unblocking elements on the coded apertures. Numerical experiments with simulated datasets show that the optimization performance of the proposed method is comparable to that of state-of-the-art methods with small images. Further, for the analyzed cases, coded aperture optimization was performed with images by analyzing coefficients smaller than in the covariance matrix.Open AccessCoded Aperture Optimization in X-Ray Computed Tomography via Sparse Covariance Matrix Estimation
Coded aperture X-ray computed tomography (CAXCT) measures coded X-ray projections to reconstruct the inner structure of an object. Coded apertures, which determine the point spread function, can be designed to improve the reconstruction quality, but most approaches are computationally expensive, leading to very small images. In this paper, a sparse covariance matrix estimation approach is introduced to minimize the information loss sensed by projections corresponding to large tomographic images. The covariance matrix representing the map of the overlapping information of the projections is obtained by using block matrix multiplication and sparse estimation. A heuristic variant algorithm with a noise factor is presented to search the combinations of D projections leading to maximum non-overlapping information acquisition, where D is the number of unblocking elements on the coded apertures. Numerical experiments with simulated datasets show that the optimization performance of the proposed method is comparable to that of state-of-the-art methods with small images. Further, for the analyzed cases, coded aperture optimization was performed with images by analyzing coefficients smaller than in the covariance matrix. - 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 23 (1545598X) , 1-5
Multispectral Time Delay and Integration Imaging With One Coded Measurement
In this letter, we present a design that enables multispectral time delay and integration (TDI) imaging to be performed with a single coded measurement. A pixel-level spectral coded aperture, which resembles a spectral Bayer pattern, is integrated onto a TDI-CCD sensor, and thus, the spectral cube is coded and aliased during the TDI scanning. Subsequently, reconstruction can be performed within the framework of snapshot compressive imaging from the resulting aliased measurement. This method bypasses the conventional power scaling issue, a critical limitation of current TDI-CCD sensors, ensuring that an increase in spectral bands does not lead to a corresponding increase in readout power dissipation. Numerical experiments performed on ten distinct scenes demonstrate the method’s capacity for robust reconstruction, achieving average scores of Q $14 = 0.903$ , $\text {SAM = 1.729}$ , $\text {ERGAS = 1.774}$ , $\text {SSIM = 0.96}$ , and $\text {PSNR = 34.7}$ dB.
- 20 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 23 (1545598X) , 1-5
Multispectral Time Delay and Integration Imaging With One Coded Measurement
In this letter, we present a design that enables multispectral time delay and integration (TDI) imaging to be performed with a single coded measurement. A pixel-level spectral coded aperture, which resembles a spectral Bayer pattern, is integrated onto a TDI-CCD sensor, and thus, the spectral cube is coded and aliased during the TDI scanning. Subsequently, reconstruction can be performed within the framework of snapshot compressive imaging from the resulting aliased measurement. This method bypasses the conventional power scaling issue, a critical limitation of current TDI-CCD sensors, ensuring that an increase in spectral bands does not lead to a corresponding increase in readout power dissipation. Numerical experiments performed on ten distinct scenes demonstrate the method’s capacity for robust reconstruction, achieving average scores of Q $14 = 0.903$ , $\text {SAM = 1.729}$ , $\text {ERGAS = 1.774}$ , $\text {SSIM = 0.96}$ , and $\text {PSNR = 34.7}$ dB.
- 13 February 2025
- Published by Optica Publishing Group in Applied Optics
- Vol. 64 (6) , 1471-1478
Design of on-chip coded X-ray sensors for high-confidence super-resolution radiological imaging
Robustness and confidence are crucial in super-resolution tasks for X-ray radiological imaging, both in terms of X-ray sensors and image processing techniques. Due to the extremely short wavelengths of X-rays, optical modulation in X-ray imaging systems is highly challenging. Consequently, single-image super-resolution techniques in computer vision have become a popular approach in medical imaging. However, these methods can lead to artifacts and the plasticity phenomenon, potentially compromising image quality and diagnostic accuracy. In this manuscript, we introduce an on-chip coded aperture design that is both cost-effective and helps overcome the additional constraints and limitations commonly found in traditional computational-imaging-based super-resolution X-ray systems. This design generates sub-pixel physical coding on X-ray sensors, and thus a three-dimensional (3D) computational decoding approach is presented that transforms the two-dimensional (2D) super-resolution reconstruction into a 3D compressed sensing problem. Unlike super-resolution restoration in computer vision, this problem can be mathematically solved to yield interpretable reconstructions under the constraints of the restricted isometry property. Our experimental results, complemented by qualitative analysis, demonstrate the superiority of our design in X-ray radiological imaging, effectively mitigating artifacts and the “plastic-like” appearance frequently associated with conventional super-resolution techniques.
- 13 February 2025
- Published by Optica Publishing Group in Applied Optics
- Vol. 64 (6) , 1471-1478
Design of on-chip coded X-ray sensors for high-confidence super-resolution radiological imaging
Robustness and confidence are crucial in super-resolution tasks for X-ray radiological imaging, both in terms of X-ray sensors and image processing techniques. Due to the extremely short wavelengths of X-rays, optical modulation in X-ray imaging systems is highly challenging. Consequently, single-image super-resolution techniques in computer vision have become a popular approach in medical imaging. However, these methods can lead to artifacts and the plasticity phenomenon, potentially compromising image quality and diagnostic accuracy. In this manuscript, we introduce an on-chip coded aperture design that is both cost-effective and helps overcome the additional constraints and limitations commonly found in traditional computational-imaging-based super-resolution X-ray systems. This design generates sub-pixel physical coding on X-ray sensors, and thus a three-dimensional (3D) computational decoding approach is presented that transforms the two-dimensional (2D) super-resolution reconstruction into a 3D compressed sensing problem. Unlike super-resolution restoration in computer vision, this problem can be mathematically solved to yield interpretable reconstructions under the constraints of the restricted isometry property. Our experimental results, complemented by qualitative analysis, demonstrate the superiority of our design in X-ray radiological imaging, effectively mitigating artifacts and the “plastic-like” appearance frequently associated with conventional super-resolution techniques.
- 12 September 2024
- Published by IOP Publishing in Physica Scripta
- Vol. 99 (10) , 105543
Effective and efficient coded aperture cone-beam computed tomography via generative adversarial U-Net
Coded aperture cone-beam computed tomography (CBCT) represents a crucial method for acquiring high-fidelity three-dimensional (3D) tomographic images while reducing radiation exposure. However, projections are non-uniformly and discontinuously sampled with the coded apertures placed in front of the x-ray source, leading to very small reconstruction scale and time-intensive iterations. In this study, an alternative approach to reconstruct coded aperture CBCT based on generative adversarial U-net is proposed to effectively and efficiently reconstruct large scale 3D CBCT images. Our method entails predicting complete and uniform projections from incomplete and non-uniform coded projections, enabling the requirement of continuity for the use of analytical algorithms in 3D image reconstruction. This novel technique effectively mitigates the traditional trade-off between image fidelity and computational complexity inherent in conventional coded aperture CBCT reconstruction methods. Our experimental results, conducted using clinical datasets comprising CBCT images from 102 patients at Nanjing Medical University, demonstrate that high-quality CBCT images with voxel dimensions of 400 × 400 × 400 can be reconstructed within 35 s, even when 95% of projections are blocked, yielding images with PSNR values exceeding 25dB and SSIM values surpassing 0.85.
- 12 September 2024
- Published by IOP Publishing in Physica Scripta
- Vol. 99 (10) , 105543
Effective and efficient coded aperture cone-beam computed tomography via generative adversarial U-Net
Coded aperture cone-beam computed tomography (CBCT) represents a crucial method for acquiring high-fidelity three-dimensional (3D) tomographic images while reducing radiation exposure. However, projections are non-uniformly and discontinuously sampled with the coded apertures placed in front of the x-ray source, leading to very small reconstruction scale and time-intensive iterations. In this study, an alternative approach to reconstruct coded aperture CBCT based on generative adversarial U-net is proposed to effectively and efficiently reconstruct large scale 3D CBCT images. Our method entails predicting complete and uniform projections from incomplete and non-uniform coded projections, enabling the requirement of continuity for the use of analytical algorithms in 3D image reconstruction. This novel technique effectively mitigates the traditional trade-off between image fidelity and computational complexity inherent in conventional coded aperture CBCT reconstruction methods. Our experimental results, conducted using clinical datasets comprising CBCT images from 102 patients at Nanjing Medical University, demonstrate that high-quality CBCT images with voxel dimensions of 400 × 400 × 400 can be reconstructed within 35 s, even when 95% of projections are blocked, yielding images with PSNR values exceeding 25dB and SSIM values surpassing 0.85.
- Open Access
- 28 October 2022
- Published by MDPI AG in Electronics
- Vol. 11 (21) , 3497
Fast Self-Attention Deep Detection Network Based on Weakly Differentiated Plant Nematodess
High-precision, high-speed detection and classification of weakly differentiated targets has always been a difficult problem in the field of image vision. In this paper, the detection of phytopathogenic Bursaphelenchus xylophilus with small size and very weak inter-species differences is taken as an example. Our work is aimed at the current problem of weakly differentiated target detection: We propose a lightweight self attention network. Experiments show that the key feature recognition areas of plant nematodes found by our Self Attention network are in good agreement with the experience and knowledge of customs experts, and the feature areas found by this method can obtain higher detection accuracy than expert knowledge; In order to optimize the computing power brought by the whole image input, we use low resolution images to quickly obtain the location coordinates of key features, and then obtain the information of high resolution feature regions based on the coordinates; The adaptive weighted multi feature joint detection method based on heat map brightness is adopted to further improve the detection accuracy; We have constructed a more complete high-resolution training data set, involving 24 species of Equisetum and other common hybrids, with a total data volume of more than 10,000. The algorithm proposed in this paper replaces the tedious extensive manual labelling in the training process, improves the average training time of the model by more than 50%, reduces the testing time of a single sample by about 27%, optimizes the model storage size by 65%, improves the detection accuracy of the ImageNet pre-trained model by 12.6%, and improves the detection accuracy of the no-ImageNet pre-trained model by more than 48%.
- 28 October 2022
- Published by MDPI AG in Electronics
- Vol. 11 (21) , 3497
Open AccessFast Self-Attention Deep Detection Network Based on Weakly Differentiated Plant Nematodess
High-precision, high-speed detection and classification of weakly differentiated targets has always been a difficult problem in the field of image vision. In this paper, the detection of phytopathogenic Bursaphelenchus xylophilus with small size and very weak inter-species differences is taken as an example. Our work is aimed at the current problem of weakly differentiated target detection: We propose a lightweight self attention network. Experiments show that the key feature recognition areas of plant nematodes found by our Self Attention network are in good agreement with the experience and knowledge of customs experts, and the feature areas found by this method can obtain higher detection accuracy than expert knowledge; In order to optimize the computing power brought by the whole image input, we use low resolution images to quickly obtain the location coordinates of key features, and then obtain the information of high resolution feature regions based on the coordinates; The adaptive weighted multi feature joint detection method based on heat map brightness is adopted to further improve the detection accuracy; We have constructed a more complete high-resolution training data set, involving 24 species of Equisetum and other common hybrids, with a total data volume of more than 10,000. The algorithm proposed in this paper replaces the tedious extensive manual labelling in the training process, improves the average training time of the model by more than 50%, reduces the testing time of a single sample by about 27%, optimizes the model storage size by 65%, improves the detection accuracy of the ImageNet pre-trained model by 12.6%, and improves the detection accuracy of the no-ImageNet pre-trained model by more than 48%.
- Open Access
- 10 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Journal
- Vol. 14 (3) , 1-9
Pseudo-Random Spread Spectrum Technique Based Single-Pixel Imaging Method
With the increasing interest in autonomous transport, machine vision and remote sensing, several three-dimensional time-of-flight (ToF) imaging technologies that can recover objects fine details have been presented. Single-pixel imaging provides an alternative to ToF photon-counting imaging with a scanned laser spot. In this work, we present a pseudo-random spread spectrum technique based single-pixel ToF imaging method, which can avoid the range ambiguity, broaden the range of signal frequencies. In addition, the Quadratic Correlation Reconstruction algorithm is applied to further increase the accuracy. With environmental light and system noise, compared to the conventional ToF single-pixel lidar we demonstrate that this approach enhances scene reconstruction quality with depth accuracy improvements of 10.5. This method may open a new gate for improved non-scanning lidar systems.
- 10 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Photonics Journal
- Vol. 14 (3) , 1-9
Open AccessPseudo-Random Spread Spectrum Technique Based Single-Pixel Imaging Method
With the increasing interest in autonomous transport, machine vision and remote sensing, several three-dimensional time-of-flight (ToF) imaging technologies that can recover objects fine details have been presented. Single-pixel imaging provides an alternative to ToF photon-counting imaging with a scanned laser spot. In this work, we present a pseudo-random spread spectrum technique based single-pixel ToF imaging method, which can avoid the range ambiguity, broaden the range of signal frequencies. In addition, the Quadratic Correlation Reconstruction algorithm is applied to further increase the accuracy. With environmental light and system noise, compared to the conventional ToF single-pixel lidar we demonstrate that this approach enhances scene reconstruction quality with depth accuracy improvements of 10.5. This method may open a new gate for improved non-scanning lidar systems.
- 22 February 2022
- Published by Optica Publishing Group in Optics Express
- Vol. 30 (5) , 7677-7693
Static coded aperture in robotic X-ray tomography systems
Coded aperture X-ray computed tomography is a computational imaging technique capable of reconstructing inner structures of an object from a reduced set of X-ray projection measurements. Coded apertures are placed in front of the X-ray sources from different views and thus significantly reduce the radiation dose. This paper introduces coded aperture X-ray computed tomography for robotic X-ray systems which offer positioning flexibility. While single coded-aperture 3D tomography was recently introduced for standard trajectory CT scanning, it is shown that significant gains in imaging performance can be attained by simple modifications in the CT scanning trajectories enabled by emerging dual robotic CT systems. In particular, the subject is fixed on a plane and the CT system uniformly rotates around the r −axis which is misaligned with the coordinate axes. A single stationary coded aperture is placed on front of the robotic X-ray source above the plane and the corresponding X-ray projections are measured by a two-dimensional detector on the second arm of the robotic system. The compressive measurements with misalignment enable the reconstruction of high-resolution three-dimensional volumetric images from the low-resolution coded projections on the detector at a sub-sampling rate. An efficient algorithm is proposed to generate the rotation matrix with two basic sub-matrices and thus the forward model is formulated. The stationary coded aperture is designed based on the Pearson product-moment correlation coefficient analysis and the direct binary search algorithm is used to obtain the optimized coded aperture. Simulations using simulated datasets show significant gains in reconstruction performance compared to conventional coded aperture CT systems.
- 22 February 2022
- Published by Optica Publishing Group in Optics Express
- Vol. 30 (5) , 7677-7693
Static coded aperture in robotic X-ray tomography systems
Coded aperture X-ray computed tomography is a computational imaging technique capable of reconstructing inner structures of an object from a reduced set of X-ray projection measurements. Coded apertures are placed in front of the X-ray sources from different views and thus significantly reduce the radiation dose. This paper introduces coded aperture X-ray computed tomography for robotic X-ray systems which offer positioning flexibility. While single coded-aperture 3D tomography was recently introduced for standard trajectory CT scanning, it is shown that significant gains in imaging performance can be attained by simple modifications in the CT scanning trajectories enabled by emerging dual robotic CT systems. In particular, the subject is fixed on a plane and the CT system uniformly rotates around the r −axis which is misaligned with the coordinate axes. A single stationary coded aperture is placed on front of the robotic X-ray source above the plane and the corresponding X-ray projections are measured by a two-dimensional detector on the second arm of the robotic system. The compressive measurements with misalignment enable the reconstruction of high-resolution three-dimensional volumetric images from the low-resolution coded projections on the detector at a sub-sampling rate. An efficient algorithm is proposed to generate the rotation matrix with two basic sub-matrices and thus the forward model is formulated. The stationary coded aperture is designed based on the Pearson product-moment correlation coefficient analysis and the direct binary search algorithm is used to obtain the optimized coded aperture. Simulations using simulated datasets show significant gains in reconstruction performance compared to conventional coded aperture CT systems.
- Open Access
- 10 February 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (4) , 1817
Point Cloud Segmentation from iPhone-Based LiDAR Sensors Using the Tensor Feature
With widely used LiDAR sensors included in consumer electronic devices, it is increasingly convenient to acquire point cloud data, but it is also difficult to segment the point cloud data obtained from these unprofessional LiDAR devices, due to their low accuracy and high noise. To address the issue, a point cloud segmentation method using the tensor feature is proposed. The normal vectors of the point cloud are computed based on initial tensor encoding, which are further encoded into the tensor of each point. Using the tensor from a nearby point, the tensor of the center point is aggregated in all dimensions from its neighborhood. Then, the tensor feature in the point is decomposed and different dimensional shape features are detected, and the point cloud dataset is segmented based on the clustering of the tensor feature. Using the point cloud dataset acquired from the iPhone-based LiDAR sensor, experiments were conducted, and results show that both normal vectors and tensors are computed, then the dataset is successfully segmented.
- 10 February 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (4) , 1817
Open AccessPoint Cloud Segmentation from iPhone-Based LiDAR Sensors Using the Tensor Feature
With widely used LiDAR sensors included in consumer electronic devices, it is increasingly convenient to acquire point cloud data, but it is also difficult to segment the point cloud data obtained from these unprofessional LiDAR devices, due to their low accuracy and high noise. To address the issue, a point cloud segmentation method using the tensor feature is proposed. The normal vectors of the point cloud are computed based on initial tensor encoding, which are further encoded into the tensor of each point. Using the tensor from a nearby point, the tensor of the center point is aggregated in all dimensions from its neighborhood. Then, the tensor feature in the point is decomposed and different dimensional shape features are detected, and the point cloud dataset is segmented based on the clustering of the tensor feature. Using the point cloud dataset acquired from the iPhone-based LiDAR sensor, experiments were conducted, and results show that both normal vectors and tensors are computed, then the dataset is successfully segmented.
- 27 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 6 (25730436) , 73-86
Coded Aperture Optimization in X-Ray Tomography via Sparse Principal Component Analysis
Coded aperture X-ray computed tomography (CAXCT) systems reconstruct high quality images of the inner structure of an object from a few coded illumination measurements. Since the CT system matrix is highly structured, random coded apertures lead to lower quality image reconstructions. In this paper, the noisy forward models of CAXCT in both Gaussian noise and Poisson noise are formulated and analyzed. In addition, a coded aperture optimization approach based on sparse principal component analysis is proposed to maximize the information sensed by a set of fan-beam projections. The complexity of the proposed optimization method is on the same order of magnitude as that of state-of-the-art methods but provide superior image quality. Computational experiments using simulated datasets and real datasets show gains up to $\sim$ 4.3dB with SNR = 25dB in the reconstruction image quality compared with that attained by random coded apertures.- 27 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 6 (25730436) , 73-86
Coded Aperture Optimization in X-Ray Tomography via Sparse Principal Component Analysis
Coded aperture X-ray computed tomography (CAXCT) systems reconstruct high quality images of the inner structure of an object from a few coded illumination measurements. Since the CT system matrix is highly structured, random coded apertures lead to lower quality image reconstructions. In this paper, the noisy forward models of CAXCT in both Gaussian noise and Poisson noise are formulated and analyzed. In addition, a coded aperture optimization approach based on sparse principal component analysis is proposed to maximize the information sensed by a set of fan-beam projections. The complexity of the proposed optimization method is on the same order of magnitude as that of state-of-the-art methods but provide superior image quality. Computational experiments using simulated datasets and real datasets show gains up to $\sim$ 4.3dB with SNR = 25dB in the reconstruction image quality compared with that attained by random coded apertures.