Some content of this application is unavailable at the moment.
If this situation persist, please contact us atFeedback&Contact
Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters


Field of the Invention
The present invention relates to a method for mapping video and images created from wide angle or non-standard lens systems. Specifically, the present invention aims to alter the curvature and warp of a wide angle image or a 360 degree panoramic image and then enable these images to be steered around or through via an external input. The contemplated images may be still frame, live video, DVD, broadcast cable or from any other source.
Background of the Invention
Lenses with a viewing angle of between 180° and 360° have been used to provide image information, which is mapped. The image is divided into sections to provide multiple images to a viewer as though the viewer were inside the environment in which the image was taken. Relative position information may be supplied through sensors in a head mounted display or through a j oystick.
Numerous applications exist for wide-angle images, including those provided by fisheye lenses, doughnut shaped lenses, and spherical lenses. These include providing virtual tours of real estate, in which a user can select particular areas of a house or other property to view more closely, security applications, theme park rides, surgical procedures, virtual travel tours and interactive television. Prior art devices that have endeavored to display a view of a 180° plus video image have presented views marked by warping and curvature.
In the past this has been accomplished externally by providing a screen that has the same curvature as the lens used to record the image.
To present a realistic view, it is important that deformation and flicker be minimized, if not eliminated.
Prior devices have not achieved greater than 20 frames per second display rates. A minimum of 30 frames per second display rate is needed to provide a non-jumpy continuous display of images, which lends realism to the viewer.
There is a need for a lens viewing system, especially a wide-angle lens viewing system, which runs in real time.
Summary of the Invention
The present invention relates to a method of using software to map wide-angle-view video images onto a virtual figure that matches the shape of the lens used to record the image or corrects for distortion in the lens. A designer can create the data points of the figure using any mathematical software. In this case a CAD-like program was used. The mapping would be done internally using software or hardware techniques. A graphics accelerator is currently used to accomplish this mapping. The lens matching shape causes the projected image to appear undistorted in a viewing space larger than the viewer's field of vision.
The camera "lens" may consist of a single lens or several lenses. For the sake of simplicity, the recording lens will be referred to as an effective lens.
It is important to remember that there is no real physical shape created in this process.

This method is more equivalent to the application of an inverse transfer function to the recorded coordinates. The effective lens causes the recorded image to be distorted. The effective lens may be thought of as a transfer function that maps the actual spatial coordinates of objects to recorded image points. If the effective lens was a uniformly flat single lens, the recorded image would be undistorted. Distortion occurs because the effective lens is not a uniformly flat single lens. The curvature in a lens or system of lenses is directly related to the transfer function. By mapping the recorded image data points to data points
corresponding to the lens shape we are effectively applying the inverse of that transfer function to the recorded image elements.
The present invention offers the ability to view an environment in real time or non-real time. In non-real time a viewer could stop and "look around" at any given point and time, observing different sections of the mapped image. For real time viewing, a hardware accelerator provides what appears to the viewer to be a natural continuous image by using a sufficiently high frame rate.
The present invention may also be applied to lenses that have a field of view of less than 180°, such as a 120° lens or a 140° lens.
The present invention further relates to a method for viewing wide-angle lens photographs without distortion. It comprises the steps of converting a photographic image into a bitmap; loading the bitmap into memory; creating a shape (corresponding to the original lens shape) onto which the image bitmap will be mapped; loading the shape into memory; mapping the image bitmap onto the shape; receiving position information; calling software routines to provide for changing views using orientation functions, in which the frame per second rate is up to 60 frames per second.
In the method of the present invention, the step of mapping the image bitmap entails using a portion of the image. The size of the portion is dependent upon the video source size. The portion size may encompass an entire wide-angle lens view. The portion is rectangular in shape and is sometimes square.
In the method of the present invention, the step of interpolating the image bitmap.
In the method of the present invention, virtually no degradation in the quality of the original image occurs.

In the method of the present invention, the preferred initial shape is called from a library. The shape is then trimmed to match the shape of a viewing lens.
The video data may be live, from broadcast cable, or from DVD.
In the method of the present invention, the step of receiving position information involves polling a head-mounted display' s position sensors.
In the method of the present invention, the position information includes roll, pitch and yaw data.
Brief Description of the Drawings
Figure 1 depicts a functional block diagram of the present invention.
Figure 2 depicts a top view of hemispheric wire frame on which a wide-angle lens view is mapped.
Figure 3 depicts a front view of hemispheric wire frame on which a wide-angle lens view is mapped.
Figure 4 depicts a perspective view of hemispheric wire frame on which a wide-angle lens view is mapped.
Figure 5 depicts a perspective view of an image mapped onto the hemispheric wire frame on which a wide-angle lens view is mapped.
Figure 6 shows a flowchart for the present invention using a DVD.
Figure 7 shows a wire frame for a spherical lens matching shape.
Figure 8 shows a bottom up view for a spherical lens matching shape.
Figure 9 shows a top down view for a spherical lens matching shape.
Figure 10 shows two lenses that form a 360° doughnut shape.
Figure 11 shows a wire frame matching the shape of the lenses in Figure 10.
Figure 12 shows a cross section of a rectangular solid having a V-shaped inside surface upon which the video data is mapped.
Figure 13 shows a cross section of a hemispheric shape for mapping video data.
Figure 14 shows the relative image density along a slice of a picture taken by a wide-angle lens.
Figure 15 is a graphical representation of the image data points being mapped onto data points corresponding to the shape required for correction.
Figure 16 shows a virtual camera that moves with respect to the hemispheric surface. Detailed Description of the Preferred Embodiment
The method of the present invention maps a wide-angle viewing lens video image onto a shape corresponding to that lens, and displays sections of the mapped shape in a full screen view.

Typical wide-angle lenses include " fisheye" lenses and doughnut or torus shaped lenses.
The image to be mapped may be " live" and altered as it is filmed, it may come from storage, it may be cable broadcast, or it may come from a DVD (digital versatile disk) (in which wide-angle view video data is stored on one of two bands of information).
A functional block diagram of an embodiment of the present system is shown in figure 1. The image 2 to be processed passes through a lens 4. A camera 6 captures the frame image. The frame image may be optionally stored 8 in memory. Whether the image represents a stored frame or a live frame, the frame is converted to an image bitmap 10.
Information relating to the lens used for filming 4 is used to retrieve a matching lens shape from memory 22. Alternatively, if the lens shape is particularly uncommon, it may be used to generate a new shape in memory. The lens matching shape is loaded 24. The image bitmap is dithered 26. The dithered image bitmap is mapped onto the lens matching shape 28. Viewing position information is updated 30. The portion of the image bitmap to be viewed is scaled according to the zoom information 32 presented on an image update sequence 34. A query line 36 is checked. If the user enters a command to terminate the session, video image updating stops 38. If the user does not enter a command to stop the session, position information and zoom command input are updated.
A steerable real time image is displayed on a CRT, LCD, or other video viewer. A computer receives image data from the video viewer or transmits image data to the video viewer.
A more detailed description of the process in capturing an image from a lens to display follows.
The basic process of the invention involves several steps.
Using a commercial CAD-like software application, shapes are created and then stored at a particular location in memory that is retrievable through a program such as Microsoft Graphics Direct X Version 6.0. (It should be noted that this method is not dependent upon the Graphics software used. Other graphics accelerators may be used including more recent versions of Direct X.) A standard shape 14 is called from a 3D Studio file. In the preferred embodiment, a solid sphere was used. For a wide-angle lens, the sphere was cut in half; one half being deleted. The shape, in this case a hemisphere, was then copied and scaled to a slightly smaller size 16. A Boolean subtraction of the copy from the original created a hollow hemisphere 18. Similarly, for a doughnut or other nonstandard lens shape, the shape is copied, the copy is scaled, and a subtraction is performed of the shape with respect to the copy to create a shape, as in Figure 13. Alternatively, the shape may include surfaces that match the shape of lenses that have surfaces having a variety of nonstandard shapes, as shown in Figure 12.
The shape is stored in memory as vertices in an ASCII file. Adjacent vertices of three or more points establish a polygon that encloses at least one pixel. The number of pixels enclosed is dependent upon the zoom factor input to the system.
When the shape is first created, it is malleable as if it were a wire frame. Figures 2-4 show detailed views of the hollow hemisphere from various angles. Figure 2 shows a top-down view of the hollow hemisphere 50. Figure 3 shows a frontal view of the hollow hemisphere 50. Figure 4 shows a perspective view of the hollow hemisphere 50. A designer is able to alter the form of the shape by dragging on the wire frame. The designer transforms the shape into a shape matching a wide-angle lens 20. The resultant shape may be hemispheric, oval, or of some other form. Matching the shape of the lens is critical to reducing distortion to a viewer.
The present shape forming method is from the perspective of the user a purely geometrical exercise. Any arithmetic processes are built into the shape. There are neither trigonometric calls nor complex function calls as in the prior art. Difficulties in the prior art have been encountered in the mathematical modeling of certain lenses. The graphics accelerator card that is used permits this mechanical process.
Figures 7-9 show a wire frame 50 for a spherical lens matching shape. Two spherical shapes slightly differing in size are subtracted from one another to produce a hollow sphere. The video image lensing feed 90 is located toward the center of the lens.
Figures 10-11 show two lenses 100 and 102 from Cyclo Vision Technologies, Inc., which form a 360° toroidal shape. A video image lensing feed 90 is located at the center of the lens. A wire frame 50 follows the contours of the shape.
The video data is mapped onto the inner surface of the shape, forming an inversion of the video data image. Therefore, the outer surface may take on any form. For purposes of viewing standard video, a 360 degree hollow sphere, a 180 degree hemisphere, or a cylindrical shape are used. The hollow lens matching shape may even be a square brick shape upon which video is played.
In Figure 14, a strip of image data points along a lens image is taken. The data points are shown to be denser toward the outside of the lens image. In Figure 15, the data points along the strip in Figure 14 are mapped onto a curved surface.
DVD, broadcast cable, and EVIAX video may also be presented by the method of the present invention. The video is always mapped to a transformed shape representing the shape of the lens through which the image is captured. This lens matching shape is key to the process in reducing distortion.

Scaling and manipulative operations were performed 20. Scaling is performed by calling a Direct X library function set_scale. Manipulation is performed by calling a Direct X library function set_position.
A copy of the entire wide-angle lens image (i.e., a scene) is made. A copy of a frame (i.e., the portion of the scene that is to be viewed) is made. By making copies before performing enhancement operations, such as blending for crispness and clarity, problems with distortion or tearing of the image on the screen are minimized.
The hollow lens matching shape is saved in a file format known as an X file 22, which is a three dimensional image format which enables the Direct X Version 6.0 libraries to recognize it.
Once a digital photograph is loaded into memory, the image is mapped directly to the inside of the hollow lens matching shape. Figure 5 shows an image mapped to the hollow lens matching shape 50. Specific software has been developed to retrieve an image that has been mapped to the previously created shape. This software associates the streaming data onto the shape in the universe. The Direct X Version 6.0 software loads the shape into memory. The software of the present invention creates a wrap to a shape and centers a picture on the shape. The software is preferably written in C++.
Following the transferring of the digital image to the PC, the software (interpolation engine) will take the image and perform some enliancements on the video before utilizing it such as adjustments for background lighting, pixellation, contrast, and balance. The interpolation engine uses the zoom factor and position coordinates to compute gray scale values of the RGB values during zoom in and zoom out operations. This is done on the fly. In a particular set up, a Pentium 350 MHz computer was used.

Illustrative software code for performing some of the enhancement operations

(filtering) is as follows:
D3DUtil_InitMaterial ( mtrl, l.Of, l.Of, l.Of ) ;
mtrl.power = 40.0f ;
g_pmtrlObjectMtrl -> SetMaterial ( &mtrl ) ;
g_pmtrlObjectMtrl -> GetHandle ( pd3dDevice, &hmtrl ) ;
pd3dDevice -> SetLightState ( D3DLIGHTSTATE_MATERIAL, hmtrl );
pd3dDevice -> SetLightState ( D3DLIGHTSTATE_AMBIENT, 0x404040 );
D3DTextr_RestoreAllTextures (pd3dDevice) ;
pd3dDevice -> SetTexture ( 0, D3DTextr_GetTexture ("picl.bmp") )
pd3dDevice -> SetRenderState (D3DRENDERSTATE_SHADEMODE,

//pd3dDevice -> SetLightState (D3DLIGHTSTATE_MATERIAL, hMat);
pd3dDevice -> SetRenderState (D3DRENDERSTATE_CULLMODE, CullMode);

pd3dDevice -> SetRenderState (D3DRENDERSTATE_DITHERENABLE,
pd3dDevice -> SetRenderState (D3DRENDERSTATE_TEXTUREPERSPECTIVE, Texture_Perspective); ρd3dDevice -> SetRenderState (D3DRENDERSTATE_
pd3dDevice -> SetRenderState (D3DRENDERSTATE_TEXTUREHANDLE, TextureHandle);
pd3dDevice -> SetRenderState (D3DRENDERSTATE_TEXTUREMAG,
pd3dDevice -> SetRenderState (D3DRENDERSTATE_TEXTUREMlN,
pd3dDevice -> SetRenderState (D3DRENDERSTATE_SRCBLEND, srcBlend);
pd3dDevice -> SetRenderState (D3DRENDERSTATE_DESTBLEND, dstBlend);

pd3dDevice -> SetRenderState (D3DRENDERSTATE_FILLMODE, FillMode);
pd3dDevice -> SetTexture (0, TextureHandle);

Once a digital image is mapped to the inside of the hollow lens matching shape, the software creates a "virtual" camera in the center of the scene, as shown in Figure 16. Calls to the VFX software libraries link the orientation and the rotation elements of the camera to an external head mounted display. In this case, the device is the VFX 3D HMD, but could be a different head mounted display. Software has been developed to convert the position information such as yaw, pitch and roll from the VFX format to values usable for the Direct X Version 6.0 software.

The software code for loading a hemisphere for camera surfaces (a generic procedure) is as follows:
// Load a DirectX ,X file
/*while ( FALSE = g_bNewFileObject )
//CHAR* strFileName = PromptUserForFileToLoad ( hWnd ) ;
CHAR* strFileName = " c:\\windows\\desktop\\ball2.x" ;
if ( NULL = strFileName ) SendMessage ( hWnd, WM_CLOSE, 0, 0 )
return 0 ;

g_pFileObject -> Load ( strFileName )
// If the file was loaded, exit the loop. Else, display an error,
if ( g_pFileObject )
g_ bNewFileObj ect = TRUE;
MessageBox ( NULL, TEXT ("Error loading specified X file") , TEXT (" Xfile" ), MB_OK|MB_ICONERROR );

A hand held cyberpuck or a similar mouse-type tool allows zooming in and out of the section of the image being displayed. The software also allows a viewer to view different portions of the wide-angle lens image as though the viewer was actually in an environment where the image was taken.
As part of the initialization procedure, the puck view is set to the view that is presented to the viewer. Only portions 55 of a wide angle lens view are presented to a viewer at a given time. In other words, there is no mapping out of the entire wide-angle lens image onto a two dimensional screen.
The VFX head mounted display software developed as part of the present invention polls the VFX head mounted display for position data gotten through magnetic and liquid sensors within the head mounted display. The VFX head mounted display software is preferably written in C++.
In the head mounted display software, the software checks to see if the puck is attached. If a puck is attached, the software queries to find out if any buttons on the puck have been pushed.
The head mounted display software, of the present invention, checks to see if the system is in automode or keymode. If not, then the head mounted display is properly attached and is transmitting data. The software sets minimum and maximum limits on the values of the yaw, roll and pitch so as to keep the view within the constraints of the image. The software also converts the head mounted display yaw, pitch and roll to a format usable by the Direct X library functions.

In the software of the present invention, after the various positional calculations are made, a virtual camera is set to a new point in the image. Repositioning through the software is done before showing a new frame. The present invention includes a wrapping function and a VFX head mounted display pulling algorithm.
Following the mapping procedure, the Direct X Version 6.0 library allows the implementation of acceleration with a 3D graphics accelerator to pennit frame rates which have been previously unachievable by other distortion correcting techniques. Using the hardware accelerator, frame rates of 60 frames per second are possible. This has immediate application for IMAX screen type viewing.
The software automatically polls the computer for an acceptable hardware accelerator and selects one. Hardware acceleration is critical because the polygon count of the hollow lens matching shape would greatly reduce the performance of the software.
Figure 6 shows a flowchart for the steps involved in showing images stored on DVD.

In the first step, in an operation block 70, the DirectDraw and Direct3D objects are created. This includes the designation of surface pointers and construction of Meshbuilder objects. The Meshes correspond to the playback shape. They are the DirectX files created in 3D Studio.
In the second step, in operation block 72, the hardware accelerators are enumerated. The equipment is checked for the proper hardware acceleration support. Hardware acceleration provides the requisite frame rate for video viewing.
In the third step, in operation block 74, the Direct3D viewport is created. All images to be displayed are drawn in the viewport.
In the fourth step, in operation block 76, the virtual figure file is loaded into the Direct3D Meshbuilder object and is stored for use.
In the fifth step, in operation block 78, the surface of the mesh is prepared.
DirectDraw surfaces are attached to the mesh surfaces and are prepared for accepting video data.
In the sixth step, in operation block 80, video card register calls are made. Calls are made to the video card registers to tell it where to put the video data when it is streamed.
Finally, the program enters a display update loop in operation block 82. In the display update loop, the mesh is loaded into the viewport and positioned appropriately, video data is then directed onto the mesh surface (blitting), an entire scene is drawn in the viewport, and tracker information from a head mounted display or mouse is checked to position the image.
While the invention has been described with reference to preferred embodiments, those familiar with the art will understand that various changes may be made without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation to the teachings of the invention without departing from the scope of the invention. Therefore, it is intended that the invention not be limited to the particular embodiments disclosed as the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope and spirit of the appending claims.