Translator Disclaimer
1 September 2015 Intelligent image processing for vegetation classification using multispectral LANDSAT data
Author Affiliations +
We propose an intelligent computational technique for analysis of vegetation imaging, which are acquired with multispectral scanner (MSS) sensor. This work focuses on intelligent and adaptive artificial neural network (ANN) methodologies that allow segmentation and classification of spectral remote sensing (RS) signatures, in order to obtain a high resolution map, in which we can delimit the wooded areas and quantify the amount of combustible materials present into these areas. This could provide important information to prevent fires and deforestation of wooded areas. The spectral RS input data, acquired by the MSS sensor, are considered in a random propagation remotely sensed scene with unknown statistics for each Thematic Mapper (TM) band. Performing high-resolution reconstruction and adding these spectral values with neighbor pixels information from each TM band, we can include contextual information into an ANN. The biggest challenge in conventional classifiers is how to reduce the number of components in the feature vector, while preserving the major information contained in the data, especially when the dimensionality of the feature space is high. Preliminary results show that the Adaptive Modified Neural Network method is a promising and effective spectral method for segmentation and classification in RS images acquired with MSS sensor.
© (2015) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Stewart R. Santos, Jorge L. Flores, and G. Garcia-Torales "Intelligent image processing for vegetation classification using multispectral LANDSAT data", Proc. SPIE 9608, Infrared Remote Sensing and Instrumentation XXIII, 96081G (1 September 2015);

Back to Top