Issue No. 09 - September (1989 vol. 11)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/34.35501
<p>Time series modeling techniques are adapted to represent or describe two-dimensional closed contours. Both linear and nonlinear models are fitted. It is found that to detect small changes in shape nonlinear modeling is necessary, even though linear models may be sufficient to differentiate between shapes which differ widely. A nonlinear model called the noncausal quadratic Volterra model is developed for the purpose. Implementation is illustrated with shapes of aircraft.</p>
2D closed contours; shape description; pattern recognition; picture processing; time series; modeling; noncausal quadratic Volterra model; pattern recognition; picture processing; time series
A. Sarkar and B. Kartikeyan, "Shape Description by Time Series," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 11, no. , pp. 977-984, 1989.