Three-dimensional Scanning System Which Combines a Photometric Stereo Method With a Structured Light Method to Obtain the Shape of an Object

Reference#: P02256

Existing art scanning technologies have several differences. For example, laser scanners are not as high-resolution. Scanning electron microscopes are higher in resolution but far more time-consuming and noisy. They also do not provide color information.

Researchers at The Johns Hopkins Applied Physics Laboratory have developed a device which combines structured light scanning and photometric stereo. The result is a 3D scanner that is capable of extremely high resolution scanning (limited by the pixel size of the camera) in relatively small amounts of time while also providing color information on the object being scanned. The final scanned image is free of laser speckle and other noise characteristics that are generally encountered with 3D laser scanning devices.

An object is placed on a platform at the center of a rotation stage. A camera with a telecentric lens is suspended above the object. Attached to the rotation stage is a digital projector. The projector is rotated above the object and projects a series of illumination patterns onto the object. These patterns consist of uniform white, red, green and blue illumination and structured light patterns of arbitrary color. Images of the object under each illumination and projector position are acquired. The uniform white projected images are used to obtain estimates of the surface normal of the object using a photometric stereo techique. The uniform color projected images are used to obtain a color map of the object. Structured light patterns are used to measure the height of the object with respect to a reference plane.

Normal data from photometric stereo analysis is accurate locally, but does not form a consistent surface and cannot be integrated to obtain a globally accurate object shape. Height data from structured light analysis, on the other hand, is accurate globally but noisy and inaccurate on small local scales. The two data sets are combined to determine the true object shape using the minimization algorithm developed.

Mr. K. Chao
Phone: (443) 778-7927