<title>Filtering and edge detection of remote sensing images by Hermite integration</title>
Lists
Only you see this
Abstract
Image smoothing and edge detection by use of Gaussian filters are much used in remote sensing image processing. In the present paper, we propose Hermite integration method to realize Gaussian filters and their derivatives by use of orthogonal polynomial theory and interpolation. We analyze at first 1-D cases and show that the output of a Gaussian filter can be calculated by the weighted sum of the input signal sampled at the positions corresponding to the Hermite polynomial roots, which gives a much better algebraic precision and a less important complexity than the classical mask convolution method. The digital implementation is then presented. The Hermite integration method is then generalized to the calculation of Gaussian-filtered derivatives and to multidimensional cases, such as 2-D image processing in remote sensing. Our method shows the following advantages: (1) Better algebraic precision. (2) Constant and reduced computational complexity independent of the filter window size. (3) Processing completely in parallel. (4) The possibility to detect edges with subpixel precision. The method is implemented and tested for artificial data and real remote sensing images and is compared with the classical mask convolution method, the experimental results are reported as well.
BibTeX
@inproceedings{shen1994lt,
author = {Jun Shen and Weifeng Shen},
title = {{\<title\>Filtering and edge detection of remote sensing images by Hermite integration\</title\>}},
booktitle = {Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE},
year = {1994},
doi = {10.1117/12.196776},
}Loading related content...