Abstract
This paper presents the clustering algorithm for detected points of slender objects in 3D space using a self-organizing map(SOM) and its test results in numerical simulations. In the 3D measurement of slender objects using digital holography, a lot of object points are detected for each slender object. The present method can identify the detected points for each object to know the object information, such as shape and inclination. The clustering performance is checked in numerical simulations. The simulations employ two different models in which the detected points are isotropically and linearly distributed in 3D space. The test results show that this algorithm can successfully identify several target clusters in 3D space.
