An Approach to Thai Decorative Pattern Recognition Using Bezier Curve Representation with Progressive Iterative Approximation

This paper presents a method to convert Thai decorative pattern (Lai-Thai) from a hand-drawn picture into a vector graphic image. The proposed method samples a series of points from a scanned image. Then a Progressive Iterative Approximation (PIA) algorithm is used to plot a corresponding Bézier curve that fits sampling points. Examples of curve fitting to Thai famous decorative patterns, Kra-Jung-Taa-Aoy and Pra-jam-yam patterns are demonstrated. The results show that more number of sampling points on the input path yields better quality in pattern recognition but the process may consume longer computational time. A challenge of this method is to choose an optimal number of sampling points to yield good quality of the approximation with minimal computational time. Quality of the output also depends on the methods of selecting sampling points.