Some research groups use stochastic differential equations for mathematical image processing. Which research groups do use stochastic processes in general and/or stochastic differential equations in the field of computer graphics/vision?
Here are some examples I like, not of groups but of a few specific works in some areas. For most cases, it's a lot more common to see non-SDE stochastics (e.g. machine learning using stochastics) than, say, explicit simulation of continuous spacetime processes.
Image Super-resolution In Random Walk Models from Geometry-driven Image Super-Resolution, Fablet et al model the image's level lines using 2D Orstein-Uhlenbeck processes and simulate them to get the high-res image. Also see works like Stochastic super-resolution image reconstruction, Tiana and Mab.
Non-Linear Filtering Check out the works on shape analysis and geometric flows by Unal et al (Feature-Preserving Flows: a Stochastic Differential Equation's View and Stochastic Differential Equations and Geometric flows). Pretty interesting from a theory point of view.
Graphics Using stochastics for noise in graphics is pretty common (e.g. flow noise), but for SDEs specifically there are papers like Langevin Particle: A Self-Adaptive Lagrangian Primitive for Flow Simulation Enhancement by Chen et al and Stochastic Modeling of Immersed Rigid-body Dynamics by Xie and Miyata.
Stochastic Motion Textures See Animating Pictures with Stochastic Motion Textures by Chuang et al. (Check out their videos of animated still images.)
Image Segmentation In Stochastic Mean Curvature Motion in Computer Vision: Stochastic Active Contours, Juan et al stochastically evolve contours to segment images.
Tracking On the more vision side, see Learning to track the visual motion of contours by Blake et al, which uses SDEs, and its kin.