Fully Automatic Projector Calibration on Planar Surface with a Sensor Based Approach

Abstract

Abstract Portable projectors are used to create large scale displays in occasions such as classroom events, entertainment events, and in mobile augmented reality applications. Usually, the default settings of such projectors are suitable to project on orthogonal planes where, the projector’s optical axis is perpendicular to both the horizontal and the vertical sides of the displaying surface. However, due to its portability, projectors are being increasingly used for non-orthogonal projections as well. In such situations, due to the absence of fully automatic calibration procedures, the users have to go through a tedious manual calibration process to correct its quadrilateral shape projections and obtain distortion-free rectified visualizations. Being dependent on the visual inputs to identify the distortions of the displayed image is a common limitation of the existing calibration techniques. A projector is a dual of a camera. Therefore, the visual input-based calibration systems use a few orientation camera-inputs and camera-projector pair calibrations to compute the distortion on the display. Different orientations of the cameras cover the entire display and compute the specific calibration parameters of the relative arrangement between the projector and the camera. Scale-Invariant Feature Transform or Speeded Up Robust Features algorithm is used for the pre-calibration phase and the continuous re-calibration with loop-feed mechanisms. In the correction mechanism, fiducial and screen marking boards affect the user’s immersive feeling and the scenery. The outcome of the research presented in this thesis is a projector display distortion correction method based on a projector arrangement relative to the projector’s display using a non-linear distortion correction algorithm. A novel hardware platform based on a small-scale hardware apparatus called Beam Adjustment Technology (BAT) is used in this solution, to do calibrations. It is a relatively small apparatus and thus, does not hinder the portability of the projector. The BAT uses, an efficient software framework for continuous automatic projector calibration based on inputs from Inertial Measurement Unit (IMU) instead of visual inputs. The proposed equations for estimating distortion corrections are based on perspective transformations using IMU and the distance sensor outputs. Furthermore, the system can estimate the arrangements of the projector and its light beam propagation in 3D world relative to the planar display to generate a distortion-free display.The solution has been tested with 30 different sizes of images and more than 1000 poses of the projector. Its projective display deviations have been tested with standard horizontal and vertical coordinates. The mean value of this deviation is 2° and 3° with respect to the horizontal and vertical coordinates, respectively. Information on the users’ experience was also collected to evaluate the system qualitatively. According to the user response, the system outputs (projected displays) were rectified and were free from distortions. The area occupied by the corrected display is decreased with the deviation of the projector from its orthogonal position. However, continuous distortion correction is important to preserve user’s immersive feelings and convey the real interpretations about the images.