Classification of planar shapes using multiresolution circular autoregressive models

A new multiresolution circular autoregressive (MCAR) shape boundary modeling technique is developed. The circular autoregressive model represents the shape contour sequence at short, medium, and long term boundary correlations. The model is independent of shape size, orientation, and location. The feature weighting (FW) recognition technique is used to classify the shapes. Experimental shape classification results are provided.<
>