REAL TIME 3-D COMPUTER VISION

THE ROBOTIC LABORATORY AT NSWC IS CURRENTLY USING A VISION SYSTEM WHICH DIFFERS SIGNIFICANTLY FROM THOSE FOR WHICH MOST 3-D OBJECT RECOGNITION ALGORITHMS WERE DESIGNED. AS A RESULT, THE IMAGING SYSTEM PRESENTS A UNIQUE PROBLEM IN OBJECT RECOGNITION. LNK PROPOSES TO DEVELOP ALGORITHMS FOR THE RECOGNITION OF OBJECTS USING THE IMAGES SLICES RETURNED BY THE IMAGING SYSTEM AND/OR ORIGINAL GRAY LEVEL IMAGES. THE BASIC ALGORITHMS DEVELOPED WILL BE EXTENDABLE TO THE HANDLING OF MANY COMPLEX OBJECTS. THE DESIGNED ALGORITHMS WILL PRESENT APPROACHES TO UNDERTAKE THE DIFFICULT TASK OF SOLVING THE UNIQUE PROBLEMS THAT ARISE OUT OF THE UNCERTAINTY IN DEPTH PERCEPTION GENERATED BY THE NSWC IMAGING SYSTEM.