Robust recognition of partially occluded 3-D objects from computationally reconstructed hologram by using a spatial filtering scheme

A novel approach for discrimination of partially occluded 3-D target objects from reconstructed digital hologram by using a spatial filtering scheme is proposed. Because of the distance condition among the target and occluding objects in the digital hologram system, the occluding objects would directly act as some noises to the reconstructed target image as a result performance of the digital hologram based 3-D target recognition system might be severely deteriorated. When we just focuse on the occluded image for recognition, sometimes we do not know the shape exactly. But if we extract the occluding objects from the hologram pattern, it is recognized more clearly when we reconstructed. Because of the property of hologram, some pixel does not have information what we want and other pixels have it. In accordance with it, we are able to discriminate the shape of occluded object even though the part of occluded object is not brightly and clearly. In this paper, by selectively subtraction the occluding object noises from the recorded hologram pattern using a spatial filtering scheme, robust discrimination of 3-D targets from the occluding objects can be realized with this spatial filtered hologram. Thus, performance of the proposed digital hologram based 3-D target recognition system can be dramatically improved. In order to show the feasibility of the proposed method, some experiments with test objects are carried out and the results are presented.