NU 2020-053
INVENTORS Oliver S. Cossairt* Boxin Shi Zihao Wang Peiqi Duan Aggelos K. Katsaggelos* Tiejun Huang
ABSTRACT Event cameras are novel bio-inspired sensors that have advantages over traditional frame cameras such as high speed, high dynamic range and low power consumption. However, current commercial and research prototypes of event cameras bear low spatial resolution and significant noise. Northwestern researchers have developed an imaging system and signal processing method that achieves 8x spatial super resolution for an existing event camera prototype. This system simultaneously offers a unifying framework that enables low noise imaging and filtering. The output can be widely applied to existing event-based algorithms that are highly dependent on spatial resolution and noise robustness. Their work reflects a systematic performance improvement in applications such as high frame-rate video synthesis, feature/corner detection and tracking, as well as high dynamic range image reconstruction. For imaging and sensing applications, it offers high resolution, high speed and high dynamic range video acquisition. For computer vision application, it offers high (variable) frame-rate video frame synthesis, motion-blurred image deblur, optical flow estimation, feature detection and tracking, and depth estimation. For robotics applications, it provides visual inertial odometry, simultaneous localization and mapping.
APPLICATIONS
ADVANTAGES
PUBLICATION Wang Z, Duan P, Cossairt O, Katsaggelos A, Huang T and Shi B (2020) Joint Filtering of Intensity Images and Neuromorphic Events forHigh-Resolution Noise-Robust Imaging. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
IP STATUS A US Patent application has been filed