Boundary detection in digital images based on spline functions and estimation theory

This paper presents a new method for segmentation of digital images based on estimation theory. A linear recursive filter is used to estimate individual boundary points. The system and measurement models are formulated by approximating boundaries in an image by spline functions. Two different spline models are considered in the paper. Both of these models yield good results in segmenting very noisy digital images. Moreover, the described method for boundary estimation is computationally efficient and well suited for real time processing.