Professor Tian’s Paper on Adaptive 3D Descattering is the Cover Feature in Nature’s Light: Science & Applications
CISE faculty affiliate Lei Tian (ECE, BME) has published a paper entitled Adaptive 3D descattering with a dynamic synthesis network that was featured on the cover of Nature’s Light: Science & Applications.
Tian’s paper focused on training a descattering network for image recovery in scattering media using an adaptive learning framework, termed dynamic synthesis network (DSN). The framework is able to adjust the model weights and adapt to different scattering conditions. The adaptability is achieved by a novel “mixture of experts” architecture that enables dynamically synthesizing a network by blending multiple experts using a gating network. Tian’s group found that training the DSN on simulated data generated by a physically accurate multiple scattering model enables generalization to real experiments and achieves robust 3D descattering.
The implications of this research are vast, and this concept may be applied to image denoising and imaging in scattering media, computational fluorescence microscopy, and imaging and light control in complex media, such as biological tissues.
