Analysis and classification of planar shapes using spatially varying autoregressive models

A spatially variant circular autoregressive (SVCAR) model is introduced for the analysis and classification of closed shape boundaries. The model treats a shape silhouette representation sequence as the output of a nonstationary all-pole linear system whose coefficient's spatial evolution can be expressed as a truncated function expansion. Features derived from the SVCAR model are shown to be invariant to shape scaling, rotation, and translation. A shape-matching algorithm is developed to optimally adjust the SVCAR model coefficient for changes in contour sequence starting point. A comparative experimental classification study is presented.<>