In this project we studied the field of stereoscopic imaging from different points of view.
INTRODUCTION
In this project we studied the field of stereoscopic imaging from different points of view, built our own stereoscopic images using a single camera and standard PC equipment and finally built an application which allows the user to “walk” within a stereoscopic scene.
Future students developing other applications connected to stereoscopic vision shall be able to use this project as a basis.
ABSTRACT
Stereoscopic viewing is complicated:
In nature, our brain receives two pictures from both our eyes and processes that information to a 3D picture.
The human cortex is built from pairs of columns received from each eye.
Along the rows of each column different shapes, sizes and motion are interpolated. Depth is calculated using the difference in pixel shift from the two pictures. Larger difference means more depth in the observed image.
We have seen that the best way to create a 3D image is by mimicking the way our eyes function. That is, providing each eye with the image it supposes to capture in terms of angles and picture quality and by tacking in consideration the way our brain interpolates that information.
There are some mathematical ground rules for receiving a good stereo image.
We learned and used those rules when we merged our own ‘right eye’ and ‘left eye’ images taken by a shifting a standard single.
CCD camera and combining the images into a stereoscopic image.
From the different methods for merging pictures, we chose color separation due to simplicity and easy to supply red-blue glasses.
Combining the pictures was accomplished using commercial ‘Depth Charge’ software, which enabled full control over the processing stage up to a pixel shift.
The practical stage of our project was building a computer interface using Visual Basic in order to present the combined picture on a standard PC enabling basic actions such as zoom in, zoom out, scrolling and more.
METHODS OF PRESENTING A 3D PICTURE
There are different methods of seeing 3D pictures but they all share the same basic principle: exposing each eye to a different view of the observed object. This project focused on two major ways for presenting 3D pictures.
1. Color separation
In this method each view of the object is painted in a different color (usually red/blue). By using special glasses with compatible color filters 3D picture is accepted.
2. Time separation
Instead of using colors, a shutter is used on special glasses that expose each eye to its correct image while blocking the others site. This calls for a synchronization circuit between the projector and the glasses used. In this way high quality 3D pictures are accepted.
STEREO RULES
For creating 3D images there are a few ground rules one must comply with.
1. Every picture from the two combined pictures must be sharp and without any distortions.
2. Identical points from both single views must lie on the same horizontal line. There shall be no vertical parallax between them.
3. Calculation of the stereo base is depended on the distance from the nearest object, the focal length and enlargement factor.
There are several graphs that helps in calculating the stereo base or an approximation can be used of about one thirtieth of the distance from the camera to the near point object.
4. Using one camera one can take 3D pictures of none moving objects only.
5. No roll of the camera is allowed between the two pictures.
6. The two pictures taken should not be replaced. Meaning right image for right eye and left image for left eye.
7. There shall be no use of flash with one mono-camera. Using flash is allowed only with two parallel cameras which there flash was properly synchronized.
THE APPLICATION
As the final stage of our project we built an application that enables viewing 3D pictures of the VISL laboratory.
Using the mouse or keyboard arrows one can “scroll” around the laboratory using red/blue eyeglasses, also zooming
in and out from the window observed.
A reset button sets the window back to its original settings and places the current picture at the left upper corner of the screen.
The application window:
The images which the application is based on:
ACKNOWLEDGMENTS
We would like to that the Ollendorff Minerva Center Fund for supporting this project.