Three-dimensional Object Recognition Based On Moir(?) Fringe

With the development of information optics, the technology of optical information processing shows obvious advantage and good qualities compared with electronic image processing since 70s. Technology of photoelectricity processing was originated from optical and electronic image processing. Using photoelectricity image processing, high speed and resolving power of optical processing was combined with flexibleness, accuracy and the programming ability. Accordingly, a new technology — photoelectric image processing and optical pattern recognition came into being, which belongs to crossed subject of computer, electronics and optics information. Photoelectric pattern recognition, a burgeoning and advanced technology in 80s, has been applied in many fields, such as: robot vision, object tracking, fingerprint recognition, image encrypting, information security etc. Most of the existing technology deals with intensity image of two-dimensional object. However object information obtains not only in the intensity image but also in the object form and structure. Losing the form and structure increases the difficulties of three-dimensional object recognition. Three-dimensional object recognition based on structured light projection is to project a sinusoidal fringe pattern to the object's surface. The fringe pattern deforms due to the modulation of the object's height. Because object's height information is obtained in the deformed picture' foundation frequency, the correlation of the deformed picture is able to recognize three-dimensional objects. The essence of this method is encoding the object'sthree-dimensional character into two-dimensional fringe through the method of structured light illumination and recognizing three-dimensional object through the method of two-dimensional recognition.Moire fringe is another method encoding the object's three-dimensional character into the two-dimensional fringe. Because moire fringe is the object's contour lines, the height distributing information of the object is encoded in the moire fringe. So the three-dimensional object recognition based on moire fringe has the characteristic of intrinsic three-dimensional recognition. Principles, method and experiment results are proposed in this paper. Original work in this paper includes:1. Three-dimensional Object Recognition based on moire Fringe: In the shadow-moire system, moire fringe corresponding to the 3D distribution of object is obtained after a filtering in the frequency domain. It depends on the correlation peak to discriminate different objects when the moire fringes modulated by the object's height are input images. Because moirS fringe is contour lines, the height distributing information of the object is encoded in the moire fringes. So the correlation of referenced fringe and the fringe waiting for recognition has the characteristic of intrinsic three-dimensional recognition. The method is tested to have the probability of using in three-dimensional object recognition through computer simulation.2. Three-dimensional object recognition with rotation-invariance based on moire fringe: Rotation-invariance is one of major aspects of optical pattern recognition (OPR) and the key of its practicability. The rotation invariance and good recognition is achieved by using SDF (Synthetic Discrimination Functions) as the reference in the joint transform correlator. The method is tested to have the efficiency of testing three-dimensional object with rotation-invariance through computer simulation and on-the-spot experiments.