Abstract

Abstract This paper presents a novel image representation,which incorporates the principles of Laplacian Pyramidinto the irregular graph pyramid. The drawback of theregular Laplacian Pyramid is their lack to keep the topologicalstructure of the image, due to the contraction process inbuilding the Gaussian Pyramid. Irregular graph pyramid isable to hierarchically represent the topological structure ofan image with multiresolution, where each level is a graphdescribing the image with various resolutions by contractingthe graph from the level below. We build irregular Laplaciangraph pyramid by storing the difference of levels in irregulargraph pyramid. Experiments and results are presented in thepaper to show the characteristic of the irregular Laplaciangraph pyramid and some immediate advantages in computervision applications.

Reference

Shao, D., & Kropatsch, W. (2010). Irregular {L}aplacian {G}raph {P}yramid. In Proceedings of the 15th Computer Vision Winter Workshop (pp. 66–71). Libor {v S}pa{v c}ek and Vojt{v e}ch Franc. http://hdl.handle.net/20.500.12708/53398