:: System Design :: Experiments :: Publications :: Acknowledgements :.

Reactive Illumination and Imaging of High-Speed Events

We consider Reactive Visual Systems, which are a class of projector-camera systems that adaptively image and illuminate a dynamic environment. Examples include adaptive front lighting in vehicles, dynamic stage performance lighting, adaptive dynamic range imaging and volumetric displays. Based on the lessons learned from computer simulations, a low latency and low jitter, tight closed-loop reactive visual system is built, which is demonstrated by imaging fast moving scenes. End-to-end latency, jitter analysis, and prediction algorithms are also measured and investigated.

System Design

arch_diag_th.png Goal: Assume that the scene (physical environment) consists of multiple objects that are moving independently at differing speeds and directions. The goal of a Binary Reactive Visual System would be to sense (image) the scene, detect objects and predict their future locations, and illuminate or dis-illuminate the objects.
system_th.png Overall Approach: Our co-located imaging and illumination system consists of a camera, processor, and spatial light modulator (SLM). The camera captures images of the scene. The processor analyzes the images and computes an illumination pattern. The SLM modulates light with high resolution over space and time. Illumination patterns can be sent from the processor at 4,000 Hz via a custom HDMI interface board.


In these experiments, we demonstrate our low-latency system (1-1.5ms reaction time) by illuminating explosion events. Objects are placed in a dark environment and illuminated with near infrared (NIR) light sources. The NIR-sensitive system camera captures images of the scene, an algorithm detects objects, and then the system illuminates the objects for image/video capture. Focusing light on the event allows for higher contrast against the scene background.


"Performance Characterization of Reactive Visual Systems"
Subhagato Dutta, Abhishek Chugh, Robert Tamburo, Anthony Rowe, Takeo Kanade and Srinivasa G. Narasimhan. IEEE International Conference on Computational Photography (ICCP), 2015 . Houston, TX. [pdf]


This research was funded in parts by a grant from the Intel Science and Technology Center for Embedded Computing, a grant from the U.S. Department of Transportation (Carnegie Mellon University Transportation Center (T-SET)), a gift from Ford Motor Company, a grant from the Office of Naval Research (N00014-11-1-0295), and an NSF CAREER Award (IIS- 0643628).

back to projects   back to top