Diffraction Rangefinding in Java.

Abstract

We describe a prototype of a diffraction range finding camera implemented in, and simulated by, Java. This sensor is able to digitize objects in 3D using only a single video camera, a diffraction grating and a means to position the target. The system is novel because it uses diffraction range finding to digitize objects and because it uses only the Java language for image processing, simulation and control. Embedded machine vision applications of Java are still rare. Java's support for video digitization and serial port I/O is poor. The lack of I/O has limited Java's application in the area of image processing systems. Our 3D scanner uses Java for the digitization and processing of image sequences. The digitization is enabled by two API's. The first, called QuickTime for Java is an Apple Computer product that has been made available for Macintosh and Windows platforms. The second, called javax.comm, consists of a specification from Sun, with implementations available for several platforms. The javax.comm package enables the control of serial-port based devices. We have written drivers for an x-y table and a frame grabber. Performance has been a primary concern with the use of Java for image sequence processing. We show that Java is both fast enough for 3D reconstructions and for display of wire-frame and smooth shaded models.