Canadian Forest Service Publications
Automated detection of man-made disturbances in the forest from remotely sensed images. 2000. Iisaka, J. Pages II-315-II-322 in Second International Conference on Geospatial Information in Agriculture and Forestry, January 10-12, 2000, Lake Buena Vista, Florida. ERIM International Inc., Ann Arbor, MI.
Issued by: Pacific Forestry Centre
Catalog ID: 5399
Availability: Order paper copy (free)
Among various disturbances to the forest, man-made disturbances such as forest harvesting, land-use conversion to agriculture fields, and road development might be one of major causes of global environment change. A conventional image analysis method for multi-spectral images processes images on a pixel-by-pixel basis and assume that each ground cover-type has unique spectral characteristics. This paper describes some results using new methods for automatic detection of roads and trails in the forest from TM data integrating spectral and spatial information. A measure of local fractal dimension is used to identify the line-like objects in a scene and a new type of vegetation index was employed to present the spectral features of roads and trails in forest. Furthermore, the whole procedure was automated.