研究目的
To contribute to easing the projector calibration task by bringing a photogrammetry-based method that does not require the use of any camera, called the Interactive Cameraless Projector Calibration (ICPC) method.
研究成果
The ICPC method provides a simple and effective way to calibrate projectors without cameras, achieving accuracies of 1-2 pixels in object space with 12 control points. It is suitable for multiple projectors and reduces calibration time, making it practical for applications like projected augmented reality.
研究不足
The method requires the user to see the points in real space with enough resolution, which can be cumbersome at distances. It may not be fully automated and relies on user interaction. Greater accuracies could be achieved with more precise measurement tools like a topographic total station.
1:Experimental Design and Method Selection:
The method uses the Direct Linear Transformation (DLT) equations for geometric calibration, combined with an interactive approach using projected augmented reality. It involves establishing 2D/3D correspondences interactively without cameras.
2:Sample Selection and Data Sources:
Three real boxes with known dimensions are used as the calibration bench. Control points (CPs) are selected from these boxes, with 3D coordinates measured manually (e.g., with a measuring tape) or from a virtual model.
3:List of Experimental Equipment and Materials:
A Vivitek DLP projector (1280x800 pixel resolution), boxes for calibration, a measuring tape, and a computer with Unity 3D and OpenCV library for implementation.
4:Experimental Procedures and Operational Workflow:
The user interactively moves projected points to match real CPs on the boxes. The calibration is computed in real-time or off-line using DLT equations. Two setups are tested: one with the projector tilted and not centered, and another with the projector parallel and centered.
5:Data Analysis Methods:
Accuracy is assessed using variance of least-square fitting, residuals in pixels, and root-mean-square error (RMSE). Errors in object space are measured and visualized.
独家科研数据包,助您复现前沿成果,加速创新突破
获取完整内容