The purpose of this invention is to analysis two dimensional image containing nonlinear wave forms, and inhomogeneous spatial scales. Image analysis has become increasingly important in pure research and practical applications. Routinely, we have to analyze the images from satellites or time series from spatial arrays. In image processing, scale separation is a necessary step toward an objective analysis of the information contents. Yet, most of the images contain patterns with inhomogeneous spatial scales which the current available image processing methods are ill-equipped to deal. The present method is based on the one dimensional Empirical Mode Decomposition (EMD) Method and Hilbert Spectral Analysis (HSA), which is an adaptive intrinsic basis generating method for one-dimensional data such as time series or profiles of two-dimensional images. The extension of this method to two dimensions is highly desirable for many of our present data are in the form of images. Unfortunately, there is no equivalence of the Hilbert Transform for two-dimensional variables. Furthermore, the two-dimensional spline will also encounter many difficulties to resolve the saddle points, ridges and valleys. To circumvent these difficulties and to offer a sharper scale resolving image processing method, we propose the 2D EMD and 2D HSA as an extension to the one-dimensional method.
Publications
Not Available
Technical Support Package:
Not Available
Related TOPS
Patent Number:
6311130
Licensing:
Available
Public Release Date:
3/22/2000
Reference Number: GSC-13909-1
|