ESTIMATING CROP DENSITY FROM MULTI-SPECTRAL UAV IMAGERY IN MAIZE CROP
- 1IREA CNR, Istituto per il Rilevamento Elettromagnetico dell’Ambiente, Consiglio Nazionale delle Ricerche, 20133 Milano, Italy
- 2Department of Agricultural and Forestry scieNcEs (DAFNE), University of Tuscia, Via San Camillo de Lellis, 01100 Viterbo, Italy
- 3IRPI-CNR, Istituto di Ricerca per la Protezione Idrogeologica, Consiglio Nazionale delle Ricerche, 10135, Torino, Italy
- 4IBF Servizi S.p.a., 44037 Jolanda di Savoia, Ferrara, Italy
Keywords: maize field, Multi-spectral UAV, Vegetation Fractional Cover, image enhancement
Abstract. In this study we exploit UAV data for estimating Fractional Vegetation Cover (FVC) of maize crop at the early stages of the growing season. UAV survey with a MicaSense RedEdge multispectral sensor was carried out on July 13th, 2017 over a maize field in Italy; simultaneous RGB in situ pictures were collected to build a reference dataset of FVC over 15 ESU (Elementary Sampling Units) distributed over the field under investigation. The approach proposed for classification of UAV data is based on local contrast enhancement techniques applied to a vegetation index (NDVI-Normalized Difference Vegetation Index) to capture signal from small plants at the early development stage. The output fc map is obtained over grid cells over 70 × 70 cm size. The approach proposed here, based on contextual analysis, reduced artefacts due to illumination conditions by better enhancing signal from vegetation compared to, for example, simple band combination such as vegetation index alone (e.g. NDVI). Validation accomplished by a point comparison between estimated (from UAV) and in situ measured FVC values provided R2 = 0.73 and RMSE = 6%.