Delineation and Extraction of Information on Farm-to-Market Roads (FMR) using Very High-Resolution (VHR) Satellite Image

Abstract. This study introduces a semi-automatic method for extracting farm-to-market roads (FMRs) using very high-resolution (VHR) satellite imagery with only RGB bands. The workflow combines edge detection, thresholding, and morphological operations with vector-based transect analysis to delineate roads and estimate width information. Applied in Nueva Ecija, Philippines, the method produced reliable results despite the absence of multispectral data and processed large images in under a minute. This speed makes the approach practical for monitoring FMR construction, especially in remote areas where site visits are difficult. While VHR imagery provides accuracy and frequent coverage, its use is limited by cost and licensing. To address scalability, the method can be adapted for medium-resolution sources such as PlanetScope or Sentinel-2, with expected trade-offs in precision. Future comparison with UAV-based monitoring may further clarify the balance between cost, coverage, and accuracy, strengthening its application in broader infrastructure planning.