Research

In this page, we show short descriptions of our works related to the ipProjector project. The very first systems of us involve obstacle-aware (a.k.a. environment-aware, clutter-aware) projection concept. This concept is suit for primary tests and can show whether a portable projector can actually sense an environment.

This project is now in progress. More information about new architectures, frameworks and novel interactive applications will be added.

Clutter-aware adaptive projection inside a dynamic environment

This work appeared in Proc. of ACM VRST 2008, Bordeaux, France.

Devices: a mobile projector attached to a web camera

Keywords: projector-camera system, clutter-aware projection, particle filters, multiple target tracking, laplacian pyramids

Continuing from the previous system appeared in ICARCV 2008, in this system, our mobile projector can actually perform an adaptive obstacle-aware projection on a surface. We do not project any light pattern, because it causes interruption in real-time environment analysis. Instead, laplacian pyramid technique is applied to detect strong edges representing obstacles on a planar surface.

Clutter-aware dynamic projection system using a handheld projector

This work appeared in Proc. of IEEE ICARCV 2008, Hanoi, Vietnam.

Devices: a mobile projector attached to a web camera

Keywords: projector-camera system, clutter-aware projection, particle filters, multiple target tracking, structure light pattern

In real life, part of a nearby surface is often hidden by some unknown objects. Projecting image onto those unknown objects cannot gaurantee correct appearances of the original image. In this system, we develop a program that simulates a dynamic obstacle-aware projection on a real surface occupied by some objects. The challenge is that how can we recognize those objects whose appearance, disappearance and movements are unknown?

In this system, objects are detected by looking at reflection of the projected stripe pattern. Multiple objects are then tracked by particle filters. Finally, a dynamic projection area is calculated at the biggest object-free location relating to previous movements.

From the simulation, the system is able to locate an appropriate obstacle-free area for projection. Even though a projector is moved or objects are entered or removed from an environment unpredictably, the system can recognize them all and generate a new projection area that best corresponds to current conditions of an environment.

> Learn more

Self-correcting handheld projectors inside cluttered environment

This work appeared in Proc. of IWAIT 2008, Hsinchu, Taiwan.

Devices: a mobile projector attached to a web camera

Keywords: projector-camera system, self-correcting projector, image stabilization, scale-rotation invariant features, cluttered environment

In mobile projector researches, keystone distortion and image stabilization are usually performed by referring to some pre-defined markers attached on a planar empty surface. This is not realistic for a mobile interactive system. Why don't we take advantage from those obstacles and transform them to paper-pins to pin down our projected images on an actual surface.

In this system, we observe possibility of creating a self-correcting mobile projector that utilizes information extracted from objects found inside an environment. The main concept is to extract scale-rotation invariant feature points from surrounding obstacles and then using them to create a frame-of-reference. We successfully achieve a simple interactive system that can prove the proposed concept. Our mobile projector (attched to a web camera) can project a stabilized undistorted image at an area free from obstacles.