Clutter-aware adaptive projection inside a dynamic environment

This paper presents a framework for a computationally adaptive projection metaphor using a handheld projector inside a dynamic cluttered environment. In addition to conventional self-correcting projection features, the framework uses multiple clutter tracking and adaptive target generation to define the clutter-aware target area for projection in a reliable manner. Using a paired projector-camera system, the framework first builds high spatial frequency feature maps using a Laplacian pyramid approach. The feature maps are then passed to a rejection step to eliminate spurious features caused by contents of the projected image. After the resulting features representing clutters are processed by the appropriated design tracker, the target area for projection is generated. Finally, the desired information for projection is rendered and sent back to the projector. The framework can be used effectively for a clutter-aware handheld projector-based system without the need for a complex hardware setup or with any prior need to clean up the environment.