Deconvolution Inverse Problem

Deconvolution is the process of recovering a high-resolution image from a blurred or degraded version, a common inverse problem across diverse fields. Current research focuses on improving deconvolution accuracy and efficiency using deep learning architectures, such as neural networks and diffusion models, often integrated with traditional methods like Plug-and-Play ADMM or incorporating physics-based models of the blurring process (e.g., point spread functions). These advancements are significantly impacting various applications, from enhancing medical ultrasound images and astronomical observations to improving hyperspectral image analysis, by enabling more precise and detailed reconstructions. The development of robust metrics for assessing the suitability of point spread functions for deconvolution is also a key area of investigation.