Adaptive smooth principal curves design

In the paper, a new principal curves design is presented which is based on a new subdivision scheme. Principal curves are extensions of principal component analysis and are smooth curves, which pass through the middle of data sets. The core of principal curves design is to iteratively average the current local projections of the data sets which produce smooth ´middle´ curves. The adaptive principal curves design is generated in two steps: forming the fold line by joining up the nodes in order, generating scheduled smooth curves by iteratively subdividing the fold lines. The experiments demonstrate the validity and facility of our algorithm.