Multiscales Edges Detection Based On Bubble Wavelet

Jia Tianxu,Zheng Nanning

1996-01-01

Abstract:Lateral inhibition described by Bubble function is a very common mechanism in living creature vision information processing system. In this paper, we use Bubble function as wavelet base and Gaussian function as smooth function to decompose signal using Mallat algorithm and the position of maximum decomposition value is just the position of edges. This algorithm has different ability to reduce noise and to extract edges according to different scales of Bubble function. When the scale of Bubble function is growing wide, then the ability of reducing noise increases and the ability of extracting detail edges decreases; otherwise, the ability of reducing noise decreases and the ability of extracting detail edge increases. Both the theory and experiment manifest that this algorithm is very efficient and practical.