The recent developments in the field of computational imaging and visualization have been marked by significant advancements in algorithms and methodologies aimed at enhancing image reconstruction, noise reduction, and real-time processing capabilities. A notable trend is the integration of machine learning techniques with traditional regularization methods to solve under-determined linear inverse problems, enabling high-quality reconstructions from limited data. Additionally, there's a growing emphasis on the theoretical underpinnings of discretization schemes in tomographic imaging, with novel interpretations and convergence analyses providing deeper insights into their performance. The field is also witnessing the emergence of efficient algorithms for constrained sampling and dynamic image reconstruction, leveraging non-reversible dynamics and motion estimation to improve convergence rates and computational efficiency. Furthermore, the development of real-time, high-performance visualization software and discrete integral transforms for 3D volumes underscores the ongoing efforts to enhance the usability and performance of tools for research and educational purposes.

Noteworthy Papers

• Adaptive Weighted Total Variation boosted by learning techniques in few-view tomographic imaging: Introduces a neural network-based approach for spatially adaptive weighting in Total Variation regularization, enabling high-quality reconstructions without prior noise knowledge.
• Non-Reversible Langevin Algorithms for Constrained Sampling: Proposes skew-reflected non-reversible Langevin dynamics for faster convergence in constrained sampling problems, outperforming traditional reversible dynamics.
• Bayesian Despeckling of Structured Sources: Establishes a theoretically grounded despeckling method for structured stationary stochastic sources, achieving superior reconstruction performance.
• Efficient Dynamic Image Reconstruction with motion estimation: Introduces a new regularization method incorporating motion estimation for dynamic inverse problems, enhancing computational efficiency and reconstruction quality.
• VTX: Real-time high-performance molecular structure and dynamics visualization software: Presents an open-source molecular visualization software optimized for real-time performance and usability in research and education.