Applying image transformation and classification techniques to airborne hyperspectral imagery for mapping Ashe juniper infestations

Authors: Yang, Chenghai; Everitt, James; Johnson, Hyrum

Submitted to: International Journal of Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/23/2008
Publication Date: 6/15/2009
Citation: Yang, C., Everitt, J.H., Johnson, H.B. 2009. Applying image transformation and classification techniques to airborne hyperspectral imagery for mapping Ashe juniper infestations. International Journal of Remote Sensing. 30(11):2741-2758.

Interpretive Summary: Ashe juniper is a noxious, evergreen shrub or small tree that invades rangelands in central Texas. This study evaluated airborne hyperspectral imagery and five different image classification techniques for mapping Ashe juniper infestations. Image analysis and accuracy assessment showed that Ashe juniper could be distinguished from associated woody and herbaceous plant species. These results indicate that airborne hyperspectral imagery in conjunction with image processing techniques can be a useful tool for mapping Ashe juniper infestations.

Technical Abstract: Ashe juniper (Juniperus ashei Buchholz), in excessive coverage, reduces forage production, interferes with livestock management, and degrades watersheds and wildlife habitat in infested rangelands. The objective of this study was to apply minimum noise fraction (MNF) transformation and different classification techniques to airborne hyperspectral imagery for mapping Ashe juniper infestations. Hyperspectral imagery with 98 usable bands covering a spectral range of 475- 845 nm was acquired from two Ashe juniper infested sites in central Texas. MNF transformation was applied to the hyperspectral imagery to reduce data noise and spectral dimensionality. Transformed imagery with the first 10 spatially coherent MNF bands was classified using five whole-pixel classifiers: minimum distance, Mahalanobis distance, maximum likelihood, spectral angle mapper (SAM) and binary encoding. To determine if additional MNF bands could improve classification results, transformed imagery with the first 20 MNF bands was also classified. For comparison, the 10- and 20-band MNF imagery was inversely transformed to noise-reduced 98-band imagery in the original data space, which was then classified using the five classifiers. Accuracy assessment showed that the first 10 MNF bands were sufficient for distinguishing Ashe juniper from associated plant species (mixed woody species and mixed herbaceous species) and other cover types (bare soil and water). Although the 20-band MNF imagery provided better results for some classifications, the increase in overall accuracy was not statistically significant. Overall accuracy on the 10-band MNF imagery varied from 82% for binary encoding to 93% for minimum distance for site 1 and from 67% for binary encoding to 94% for maximum likelihood for site 2. The 98-band imagery derived from the 10-band MNF imagery resulted in overall accuracy ranging from 72% for binary encoding to 97% for maximum likelihood for site 1 and from 56% for binary encoding to 93% for minimum distance for site 2. Although both approaches produced comparable classification results, the MNF imagery required smaller storage space and less computing time. These results indicate that airborne hyperspectral imagery incorporated with image transformation and classification techniques can be a useful tool for mapping Ashe juniper infestations.