Kinect 3D reconstruction for quantification of grape bunches volume and mass

In the last decade, three-dimensional reconstruction of plants has gained a noticeable importance, in particular for the possibility of collecting data correlated to biomass, leaves area, etc. Different sensing technologies are available for 3D imaging, as, for instance, laser scanning, or stereoscopic reconstruction: however, practical application is still limited by high costs, or high speed data processing demand. For the present work the depth sensing camera technology is implemented. Such technique has gained in the last years a wide diffusion in particular thanks to recent introduction in the video games industries. Specifically, a Microsoft Kinect™ RGB-depth was used. The Kinect sensor can be applied in fruitculture in order to allow fast reconstruction, allowing characterization of the canopy or estimation of the fruit volume. In this paper it is proposed specifically for vineyard applications, where strict national or regional regulations limit plant management and grape production. The work focuses in particular on the possibility of implementing such sensor in order to rapidly estimate the grape mass and allow foreseeing production before harvesting. To this end, measurements are carried out in the case of both black and white berries, with different mass grapes (ranging between 50 and 1000 g) and scanned at different distances with two orientations of the sensor.