Embedded Files
We are actively recruiting PhD students, see details in the Join Us Tab.
Automobile Camera Array
Automobile Camera Array
This project fuses data from multiple sensors to improve the quality of the camera array system, leading to scalable, low-cost solution for pavement maintenance.
Camera Array
Camera Array
Using the principles from Structure from Motion (SfM), we can construct "camera arrays" in a series of camera positions. These camera arrays allow for instantaneous 3D reconstructions. This improvement in procedure allows us to capture dynamic events. This process also allows us to make the cameras mobile while scanning for 3D reconstructions.
GPS + IMU Fusion & Kalman Filter
GPS + IMU Fusion & Kalman Filter
To aid in the 3D Reconstructions, we provide GPS location data in images' metadata. We do this via sensor fusion between GPS & IMU data, using a Kalman Filter for location estimation.
IMU-Assisted Deblurring
IMU-Assisted Deblurring
By synchronizing high-frequency IMU data with the camera shutter, we record the exact rotational trajectory of the device during exposure. This motion data allows us to calculate a precise Point Spread Function (PSF)—a map of exactly how pixels smeared across the sensor. Unlike traditional "blind" deblurring that guesses motion, we use this known trajectory to mathematically reverse the smear (non-blind deconvolution), effectively restoring the sharp image hidden within the blur.
Normal SfM Result
Normal SfM Result
Camera Array SfM Result
Camera Array SfM Result
3D depth map (scalar field)
3D depth map (scalar field)
Page updated
Report abuse