Our approach uses a stereo camera to capture a pair of images of the same object staggered a few inches apart. Pixels are used to identify common reference points across a pair of images and this allows the software to estimate the distance from the camera to each point within those images. This is very similar to how our eyes actually work. VanGogh’s current implementation works well with Point Grey’s camera, but will work with any type of photo-based solution.
Our auto-registration technology uses a fixed point outside the object being scanned in order to register various scanning points on the surface of the object. Our algorithm registers all the scanned points using the outside fixed point as the common reference point. Then the location of every point in every scan of the object is identified. Once that is accomplished, all the points are combined to form a complete 3D model. Of course, there’s a complex algorithm that’s also filtering out any scanned-base noise and filling in the holes of the 3D scan using all the points from the multiple scans. The end result is a very accurate model since the process uses the “best” 3D scan points available from each individual scan.
No matter the scanning approach, any single scan may have some distortion known as noise or holes. Our auto-registration technology easily eliminates scanner-related noise and fills in any holes. It’s like taking multiple pictures of the same object and then taking just the best pixels from all the images to create a perfect picture. This eliminates any need to manually clean up or align the individually scanned images.
Whether you’re looking for a complete 3D model or you simply want to enhance your current scanning solution to get a more accurate scan from a single angle, our auto-registration technology is the perfect solution for any application.