Issue link: http://winesandvines.uberflip.com/i/987926
72 WINES&VINES June 2018 GRAPEGROWING WINE EAST Natural Corks Champagne Corks Twinline Corks Bartops VISION Synthetic Corks G-Cap® Screw Caps Sales Representatives: Chris & Liz Stamp info@lakewoodcork.com lakewoodcork.com 4024 State Route 14 Watkins Glen, NY 14891 607-535-9252 607-535-6656 Fax PIONEER INNOVATOR PARTNER able to perform all calibration steps with mini- mal input by the user. These programs can take a set of images and automatically align them by feature identification, matching and bundle adjustment, which is refining a set of initial camera and structure parameter estimates to predict locations of observed points in a set of images. Previously, bundle adjustment, match- ing and feature identification were performed manually or semi-manually through a variety of computer programs. 4 The images acquired by different sensors (e.g., thermal, multispectral, visible, hyper- spectral) commonly used with UAVs also must be calibrated to avoid possible errors in the data. In one example, two 2 x 2 m leveled dark and white targets were placed in a central loca- tion within the UAV flight pattern. 2 These targets were measured in the field with a calibrated field spectrometer at the same time as UAV image acquisition. Hyperspectral sensors can be calibrated using measurements made with a calibrated uniform light source at several levels of illumination and integration times. 36 All im- ages, particularly hyperspectral, also require atmospheric correction. Aerosol optical depth was measured at 550 nm with a sun photometer in the study area during UAV flights. 36 The aero- sol optical depth measurement simulates total incoming irradiance at 1-nm increments, which is then used for the atmospheric correction of hyperspectral images. Another important factor in the processing of UAV images is the removal of the cover crop data from the space between the rows of vines. The data normally of interest is exclusively the spectral reflectance and/or thermal data of the vineyard canopy. Errors in the spectral and thermal canopy data can be caused by the in- clusion of the between-row soil data and/or cover-crop data. In most cases these data can be isolated by setting a threshold spectral re- flectance value, below which no data are in- cluded in maps. An example of a common error would be the inclusion of thermal bare soil data, since the soil would be expected to have a higher temperature than the vineyard canopy. The high-resolution nature of the images ac- quired by UAVs allows for the targeting of pure canopy pixels. Computer programs now can segregate the pure canopy pixels from other background pixels. Santesteban et al. 29 used Agisoft Photo- scan Professional (Agisoft LLC, St. Petersburg, Russia) to produce a digital elevation model that differentiates the image into two terms, row (vine canopy) and inter-row. The two terms were differentiated by their respective height, since the vine canopy is of a greater height than the ground or most inter-row cover crops. The extraction of pure vine vegetation pixels in the thermal, hyperspectral and mul- tispectral high-resolution images allows for increased accuracy and precision in the data acquired by a UAV. After the images are ap- propriately processed, the data can be used to evaluate various canopy variables of interest. An example of a processed image is shown in Figure 2b. Andrew G. Reynolds is professor of biological sciences/ viticulture at the Cool Climate Oenology and Viticulture Institute, Brock University, St. Catharines, Ontario, Canada; Elena Kotsaki, Hyun-Suk Lee and Patrick Kelly are also with the Cool Climate Oenology and Viticulture Institute, Brock University. Marilyne Jollineau is associ- ate professor in the department of geography at Brock University. Ralph Brown is professor of engineering at the School of Engineering, University of Guelph, Guelph, Ontario. Mehdi Shabanian is research assistant in the department of molecular and cellular biology at the University of Guelph. Adam Shemrock is with Air-Tech Solutions, Inverary, Ontario. To see references for this article, go to winesandvines.com and search under Magazine › Features › June 2018.