A Qt and OPENCV based application for camera calibration and projector mapping

Hello,

We are developers of gaming solutions that forms a basis of training for homeland security personnel. In this software, we will have computer generated imagery of virtual targets that trainees will be made to fire upon.

The main hardware components of our system are as follows:-

1. One PC running the gaming software and other UI related items, databases and report generation modules.

2. One or more pairs of camera-projector

(NOTE: The projector-camera and the screen is STATIONARY and not moving)

3. Pulse coded laser modules.

Project Description and Background Information.

1. The projector and Camera will be placed side-by-side at a distance ranging between 10 ft to 18 feet from a projection surface usually a wall.

2. The projector will project an image that has 1920x1080 resolution to the wall, which will eventually measure (at 18 feet as mentioned in above point) an image size that will be 18ft high and 7 ft wide.

3. The camera that we are using is an area scan camera having pixel resolution 1440x1080 pixels having a varifocal lens to ensure coverage of the projection surface.

6. Projector-camera Pairs. On a 18ft x 10ft wall, we have about 6 recruits using the system, keeping a one meter shoulder-to-shoulder distance in mind. When more than 6 recruits are to be on a system, we add an additional pair of camera-projector to increase the number of recruits.

PROBLEM DEFINITION AND REQUIREMENT

A. However, when we initially start out, we notice that we cannot ever perfectly align the camera field of view and projector field of view. Additionally, we need a way of defining or mapping the co-ordinates of the camera and the projector, so that when a trainee fires upon a target that is projected on the wall, the system then is able to calculate where the virtual bullet will hit or miss the target in the software, thereby awarding him points as necessary.

For this, three things are required.

A. One image, to allow the user to align the camera so that it “sees” the complete projection. This can be a checkerboard.

B. One image, so that any perspective correction and unwrapping is required.

C. Finally, correction of lens distortion. (Optical Flow + Calculation of Essential Matrix).

D. All of the above parameters, when applied to the final image and to subsequent incoming frames, should lead eventually to a system that is highly accurate in determining the location of the hit or miss.

Looking forward to receiving your proposals and taking this project forward at the earliest.

Thanks and Regards

Roahan

Pengaturcaraan C++ Kejuruteraan Perisian OpenCV Qt

ID Projek: #22037607

Tentang projek