Browsing by Author "Marques, J."
Now showing 1 - 2 of 2
Results Per Page
Sort Options
- Grapevine yield prediction using image analysis - improving the estimation of non-visible bunchesPublication . Victorino, G.; Maia, G.; Queiroz, J.; Braga, R.; Marques, J.; Santo-Victor, J.; Lopes, C.M.Yield forecast is an issue of utmost importance for the entire grape and wine sectors. There are several methods for vineyard yield estimation. The ones based on estimating yield components are the most commonly used in commercial vineyards. Those methods are generally destructive and very labor intensive and can provide inaccurate results as they are based on the assessment of a small sample of bunches. Recently, several attempts have been made to apply image analysis technologies for bunch and/or berries recognition in digital images. Nonetheless, the effectiveness of image analysis in predicting yield is strongly dependent of grape bunch visibility, which is dependent on canopy density at fruiting zone and on bunch number, density and dimensions. In this work data on bunch occlusion obtained in a field experiment is presented. This work is set-up in the frame of a research project aimed at the development of an unmanned ground vehicle to scout vineyards for non-intrusive estimation of canopy features and grape yield. The objective is to evaluate the use of explanatory variables to estimate the fraction of non-visible bunches (bunches occluded by leaves). In the future, this estimation can potentially improve the accuracy of a computer vision algorithm used by the robot to estimate total yield. In two vineyard plots with Encruzado (white) and Syrah (red) varieties, several canopy segments of 1 meter length were photographed with a RGB camera and a blue background, close to harvest date. Out of these images, canopy gaps (porosity) and bunches’ region of interest (ROI) files were computed in order to estimate the corresponding projected area. Vines were then defoliated at fruiting zone, in two steps and new images were obtained before each step. Overall the area of bunches occluded by leaves achieved mean values between 67% and 73%, with Syrah presenting the larger variation. A polynomial regression was fitted between canopy porosity (independent variable) and percentage of bunches not occluded by leaves which showed significant R2 values of 0.83 and 0.82 for the Encruzado and Syrah varieties, respectively. Our results show that the fraction of non-visible bunches can be estimated indirectly using canopy porosity as explanatory variable, a trait that can be automatically obtained in the future using a laser range finder deployed on the mobile platform
- Hydrogeological study of a high mountain area (Serra da Estrela, Central Portugal): a multidisciplinary approachPublication . Marques, J.; Espinha Marques, J. M.; Chaminé, H. I.; Afonso, M. J.; Carreira, P. M.; Fonseca, P. E.; Cabral, J.; Santos, F. A. Monteiro; Vieira, Goncalo; Mora, Carla; Gomes, A.; Teixeira, J.; Samper, Javier; Pisani, B. J.; Aguiar, C.; Gonçalves, J. A.; Almeida, P.The results ofa preliminary hydrogeological study ofthe river Zêzere catchment upstream ofManteigas(Serra da Estrela Natural Park,Central Portugal) are presented.In this mountain region,different types of groundwater and surface water (used in several economic activities) occur.The methodology adopted in thisstudy emphasizes the way how Geology,Geomorphology,Geophysics,Geochemistry,Soil Science andHydrogeology contribute to the description ofthe hydrological phenomena taking place in the catchment,such as infiltration and aquifer recharge and groundwater flow and geochemistry — allowing to develop bet-ter hydrogeologic conceptual models.The hydrological modelling in course includes the use ofthe VISU- AL BALAN code,which is being coupled to a GIS.The hydrogeochemical techniques are highlighted as well as its preliminary results concerning major and minor elements.The thermomineral water study includes the identification ofthe reservoir’s geologic material,the characterization ofwater-rock interactionand geothermometry.