Evaluation of remote sensing techniques for mapping transborder trails

This paper evaluates the utility of various image processing methods for mapping smuggler trail networks crossing the U.S.-Mexico border. Very high spatial-resolution, digital, multispectral imagery was acquired in three visible and one near-infrared wavelength bands along the U.S.-Mexico border using an Airborne Data Acquisition and Registration (ADAR) digital camera system mounted on a helicopter. Four image enhancement methods and one neural-network based automated feature extraction technique were tested. Measures of trail length were compared with a ground-based GPS trail survey to evaluate accuracy. The optimal image enhancement method for trail mapping varied with topography and vegetation structure. The green vegetation component from spectral mixture analysis (SMA) and the normalized difference vegetation index (NDVI) were most useful for enhancing trails for subsequent interactive visual interpretation and digitizing. The neural network feature extraction technique produced superior results when combined with interactive digitizing.