How it Works
The following are thumbnails of 16 images shot with a hand-held, consumer-level digital camera (Canon PowerShot A20). The camera was set in the Auto mode, which led to dramatic differences in illumination and exposure between the images. Notice the movement of the people in the foreground of these photos.
Step 1: Organize Images - No correction
When i2k Quickage™ generates a montage (or panorama) it does so in a series of steps not seen by the end user. In the first step it automatically recognizes which images belong together, places the images relative to each other, and merges the images. This step, called "image registration", uses DualAlign's highly-advanced, proprietary i2align™ algorithm, which is also being used in medical and government applications. The figure on the right shows what the montage looks at the end of the registration process.
Step 2: Distortion Correction
Next, i2k Quickage™ corrects the distortions that result from the image registration process, straightening the montage and making it flatter. While the wide combined field of view of these photos does not allow for complete straightening, the result, shown on the right, is much better than what would be obtaining by using an extremely wide-angle lens to capture the same field of view.
Step 3: Smart Blend - Seam Selection
The ghost images of moving people in the foreground and the small misalignments caused by using a hand-held camera are removed next by i2k Quickage™. Another of i2k's proprietary algorithms is applied to choose seams that avoid the misalignments and steer clear of movement. The blending regions between the images are also established to allow smooth transitions between images while preserving the finest details.
Step 4: Camera Correction
No matter how good the seam selection and blending are, there still can be brightness differences between photos, especially when taking pictures using automatic exposure or using a cell-phone camera. Also, the lenses used in most cameras cause at least some vignetting - a drop-off in brightness on the image periphery. i2k's camera and lens analysis software corrects for these differences and drop-offs, creating final brightness and colors that optimally combine the images while simultaneously preserving the values of the original photos as much as possible. The result is a seamless, visually-pleasing montage.
Final Step: The Final, Auto-Cropped Result
Finally, i2k Quickage™ must still create what appears to be a single image.
While users are able to save the montage and crop it later themselves, i2k Quickage™ can automatically crop the montage by finding the largest rectangle that fits inside the uncropped montage. The result is the beautiful composite photo shown on the right. Can you find any boundaries of the original 16 photos? Will your friends believe that this was created automatically from these photos without any effort on your part beyond selecting the images?