Overview
Image viewer is a typical application for large display. It makes use of the high-resolution large display and lead to more efficient way for viewing images. Image viewer utilizes the output of the calibration phase (geometric result file and alpha masks), makes the necessary geometric and photometric correction and shows the resulted image on the multi-projector display. With the benefit of the much larger display area and higher resolution, uses can see both the overall picture and details at the same time.
Algorithms
In order to display the image on the multi-projector based large display, the input image texture shown on every projector need to be warped according to the calibration result. This includes geometric correction and photometric correction. The details of the algorithms are introduced below.
For geometric correction, we use what we called Piecewise-Linear Methods. The basic idea is to apply a piecewise-linear mapping of the projector coordinates to camera coordinates. The tessellation is implemented using standard Delaunay triangulation. The geometric correction is implemented using bilinear interpolation of these mappings at the vertices of the triangles. This method needs minimum calculation and only need to apply interpolation triangle-by-triangle to the corresponding tessellated mesh in the projectors coordinate space. The final result is a geometrically correct image in the camera's view (or in the view of the person who stand at the spot of the camera). The algorithm is also very robust and can apply to curved surface or even corner of the wall. For a more specific explanation, please refer to Chapter 3.3.4 of Reference [1].
For photometric correction, we use the blending techniques to address overlapped regions and try to smooth color transitions across different projectors. The smooth transitions can be achieved by using a linear or cosine ramp, which attenuates pixel intensities in the overlapped region. For one pixel in the overlapped area, the more it near the boundary of the projector, the less the intensity of this pixel in the projector will be. In order to control the intensity of each pixel, we generate an alpha map for each projector in the calibration stage. The generation of alpha map follows these steps: first, we identify the display area each projector occupied in the camera image; then we calculate the distance transform image for each projector; next, we use the value of the distance transform image as weight and normalize it to 0~255. We save this normalized image as alpha image. During the photometric correction, each of the pixels to be shown will use this alpha image as an alpha mask. A gamma corrrection may also apply. For a more specific explanation, please refer to Chapter 4.4.3 of Reference [1].
Download
Please go to the download page.
How to use the code
Before you start, you should first choose the configuration you would like to adopt according to your intention and available hardware. We provided codes for multi-projector multi-PC configuration and multi-projector single-PC configuration. Below are the step-by-step instructions about how to use the code.
Multi-Projector Multi-PC Configuration:
- Download the corresponding binary package named ImageClient (control client running on master PC) and ImageServer (render server running on render PC) from the above "Download" Section;
- Connect the PCs and projectors as showed in Figure 3 except not necessary to connect the camera;
- Copy the calibration result file (result2x2.txt), alpha image (Px.bmp) and the demo images to the working directory of all the render servers;
- Run the following program on all the rendering PCs from command line with:
ImageServer.exe - Modify the configuration file named Client4.cfg according to your actual configuration (IP address, port, gamma, etc);
- Run the control client program on the master PC from command line with:
ImageClient.exe Client4.cfg demo1.bmp - If everything goes well, you will see the image demo1.bmp showing on the large display correctly;
- You can use mouse and keyboard to zoom in/zoom out, change the position and so on, please see the command line help of the program.
Multi-Projector Single-PC Configuration:
- Download the corresponding binary package named MMonImageViewer from the above "Download" Section;
- Connect the PCs and projectors as showed in Figure 4 except not necessary to connect the camera;
- Make sure that you have multiple graphics cards installed properly;
- From the desktop of your Windows XP, go to "Right-Click->Properties->Settings", enable all the monitors connected by select the "Extend my Windows desktop onto this monitor" check box; Make sure the top left projector is set as primary display and drag the monitor icons to match the physical arrangement;
- Copy the calibration result file (result2x2.txt), alpha image (Px.bmp) and the demo images to the working directory of the PC;
- Run the following program on the PC from command line with:
MMonImageViewer.exe Client4.cfg demo1.bmp - If everything goes well, you will see the image named demo1.bmp showing on the large display correctly;
- You can use mouse and keyboard to zoom in/zoom out, change the position and so on, please see the command line help of the program.