Camera Merges Disciplines to Produce 3-D Images

Laurel M. Sheppard

By combining technologies from medical imaging (computed tomography) and from radio astronomy (interferometry), researchers at the University of Illinois' Beckman Institute for Advanced Science and Technology have developed a lensless camera that can capture three-dimensional images on a computer. Such a camera has an infinite depth of field, which means that it never requires focusing.

Researchers combined computed tomography and interferometry techniques to capture this 3-D image of two toy soldiers. The lensless camera does not require focusing, simplifying data processing. Courtesy of the University of Illinois.

The optical system is made up of a rotational shear interferometer with a 5-cm-aperture beamsplitter and 5-cm folding mirrors; a mechanical shutter; a 3-nm bandpass spectral filter centered on a 633-nm wavelength; and a 512 x 512-pixel charge-coupled device (CCD) detector array. Data analysis is done on a personal computer with 512 MB of memory.

The object to be imaged rotates on a stage in front of the interferometer, which collects light bouncing off the object in many directions, filters it and allows the light to interfere. The CCD array captures the interference pattern, which then goes through processing by several mathematical algorithms. A Fourier transform algorithm transfers the interference data into a two-dimensional projection that can be analyzed by tomographic algorithms, which transfer the data into a set of image slices representing the 3-D object.

The entire imaging system is easy to build and relatively inexpensive (several thousand dollars), with the major cost being the computer. Faster imaging times would lead to higher costs. The current system produces an image in about 10 minutes, according to David J. Brady, team leader and a professor of electrical and computer engineering. "However, we are currently in the process of constructing a system consisting of an array of 30 computers," he said, "so that images can be produced in real time."

The research was funded by the Defense Advanced Research Projects Agency in Arlington, Va., which is interested in developing sensor arrays for target recognition. Because the surface of an object can be mapped with a resolution better than 1 mm, microscopy and machine vision tasks also could benefit. Brady believes that the system could be used for 3-D video applications, in which a person could "walk" around a recorded scene, similar to virtual reality.