11 May 2009 Effects of image restoration in classification and visual analysis of LANDSAT imagery over Puerto Rico
Author Affiliations +
Abstract
In image processing, restoration is performed in order to enhance salient features or remove degradation effects to assist in subsequent human or machine analysis. Restoration algorithms deal with removing the effect of spatial blurring introduced by the imaging system which is introduced in multispectral remote sensing by limitations of the optical system, scanning, and electronics. This result in that the support of the signal received at a particular pixel comes from an area larger than the projection of the pixel on the surface ground. This hinders the ability to derive surface information from satellite images on a per-pixel basis. Image deblurring is the problem of recovering the original image from its blurred noisy observation. In this study, we present how image reconstruction algorithms can help visual and machine-based analysis of Landsat imagery taken over coastal areas in southern Puerto Rico. Image restoration improves image contrast near edges and improves visual appeal of different color composites. Improvements in infrared bands were more significant than in bands in the visual range. Performance of classifiers over nearly homogeneous areas used for training and testing areas was not significantly changed. However, small features that were not classified correctly in the original image were classified correctly in the restored image.
© (2009) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
E. Veronica Morales-Irizarry, E. Veronica Morales-Irizarry, Miguel Vélez-Reyes, Miguel Vélez-Reyes, } "Effects of image restoration in classification and visual analysis of LANDSAT imagery over Puerto Rico", Proc. SPIE 7334, Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XV, 733427 (11 May 2009); doi: 10.1117/12.823753; https://doi.org/10.1117/12.823753
PROCEEDINGS
10 PAGES


SHARE
RELATED CONTENT

TM/LANDSAT thermal image unmixing
Proceedings of SPIE (August 03 1997)
Restoration For Sampled Imaging Systems
Proceedings of SPIE (December 09 1986)
Time series tropical forest change detection a visual and...
Proceedings of SPIE (November 03 1996)

Back to Top