Continuous-time Intensity Estimation Using Event Cameras

We present a method to reconstruct image intensity using an event camera. Our method runs in real-time, and outputs a continuous-time image intensity state with microsecond temporal resolution, and high dynamic range. We use a per-pixel complementary filter to fuse image frames with events. If no frames are available, our method also performs well using only events. Reference: Cedric Scheerlinck, Nick Barnes, Robert Mahony, "Continuous-time Intensity Estimation Using Event Cameras", Asian Conference on Computer Vision (ACCV), Perth, 2018. PDF: https://cedricscheerlinck.com/files/2018_scheerlinck_continuous-time_intensity_estimation.pdf Website: https://cedricscheerlinck.com/continuous-time-intensity-estimation Code: https://github.com/cedric-scheerlinck/dvs_image_reconstruction Dataset: https://drive.google.com/open?id=1Jv73p1-Hi56HXyal4SHQbzs2zywISOvc