Recognition of 3D thin-plate objects using orthogonal AR model

A three-dimensional (3D) autoregressive model is introduced for modeling, invariant feature extraction and recognition of arbitrary 3D thin-plate objects. Am orthogonal estimation algorithm is derived for the estimation of the coefficient matrices associated with the object boundary and error reduction ratio is used to determine the sufficient order of the model to describe the boundary. It is shown that the eigenvalues, trace and the determinant of the orthogonal coefficient matrices are invariant to rotation around the origin, to the choice of starting point in tracing the boundary and to scale and translation, and classification based on orthogonal parameters is shown to be less susceptible to incorrect model order specification.