Study of vine foliage using multispectral imaging | Marue-Aure Bourgeon

This thesis manuscript presents the design and potential of onboard multispectral imaging on a terrestrial vehicle for characterizing vine foliage. After optimizing the equipment and image acquisition methods, a radiometric calibration method was developed. This method is based on using neutral patches from a MacBeth chart. The presented method involves estimating the correction parameters (a, b) of the images through linear interpolation and verifying the accuracy of the estimates through cross-validation. Using the images, the NDVI (Normalized Difference Vegetation Index) was determined and compared to that of the Greenseeker sensor. This comparison, while validating the capabilities of the device, highlighted the benefits of imaging: versatility in determining spectral indices, versatility in spatial characterization (focus on areas of interest), visualization of disease symptoms, offering interesting perspectives for identifying disease symptoms at the plot level.