抄録
This paper presents an optical technique for fibrous object measurement in 3D space based on digital holography. The technique consists of two steps : one is object identification using clustering based on self-organization algorithm and the other is shape evaluation using curvature and torsion as evaluation indexes. On identification, candidate points of fibrous objects are clustered to each fiber. Then constitution point sequences are made from the clustered points to evaluate curvature and torsion quantitatively representing object shape. The performance of the present technique is checked by numerical simulation. It is found from the numerical simulation results that curvature and torsion obtained in shape evaluation are in good agreement with the true values.
