Title :
A 2D autoregressive method for deformation evaluation in registration of remote sensing data
Author :
Fung, Ko Bong ; Ionescu, Daniela
Author_Institution :
Dept. of Electr. Eng., Ottawa Univ., Ont., Canada
Abstract :
A registration method is developed based on an A* algorithm in conjunction with a 2D autoregressive model used for image representation. The A* algorithm serves for obtaining the initial pixels which used further for determining the innovation process. Prior to this step the 2D autoregressive parameters are calculated using an initial set of pixels from the original image. The error measured between the innovation and the primitive pixels of the subject image is then considered as the exact function for the A* algorithm. The process is recursively applied until the whole set of primitives is registered. An example illustrates the approach
Keywords :
image processing; remote sensing; stochastic processes; time series; 2D autoregressive method; 2D autoregressive parameters; A* algorithm; deformation evaluation; error; exact function; image representation; innovation; primitive pixels; registration; remote sensing data; subject image; Data engineering; Geometry; Image processing; Image registration; Layout; Pixel; Polynomials; Remote sensing; Resonance light scattering; Testing;
Conference_Titel :
Electrical and Computer Engineering, 1993. Canadian Conference on
Conference_Location :
Vancouver, BC
Print_ISBN :
0-7803-2416-1
DOI :
10.1109/CCECE.1993.332393