Multispectral detection of floral buds for automated thinning of pear

Niels Wouters, Bart De Ketelaere, Tom Deckers, Josse De Baerdemaeker, Wouter Saeys
2015 Computers and Electronics in Agriculture  
Thinning of pome and stone fruit involves the reduction of tree crop load in order to 14 regulate fruit set and quality. As it is typically carried out through manual labor, thinning comprises a 15 large part of a grower's production costs. Mechanized thinning has been shown to be a cost-effective 16 alternative but the performance of existing thinning devices needs to be further improved by taking 17 the variation in bearing capacity of the individual trees into account. 18 In this work, a
more » ... ispectral camera system is developed to detect the floral buds of pear (cv. 19 Conference) during the growth stages prior to bloom. During a two-year field trial, the multispectral 20 system was used to measure orchard scenes in six distinct optical wavebands under controlled 21 illumination. These wavebands are situated in the visible and near infrared region of the spectrum 22 and were selected based on hyperspectral laboratory measurements described in previous work. 23 The recorded multispectral images were converted to a database containing the spatial-spectral 24 signatures of the objects present in the orchard. Subsequently, canonical correlation analysis was 25 applied to create a spectral discriminant model that detects pixels originating from floral buds. This 26 model was then applied to the recorded data after which an image analysis algorithm was designed 27 and optimized to predict the number of floral buds. In total, approximately 87% of the visible floral 28 buds were detected correctly with a low false discovery rate (<16 %). Therefore, it is expected that 29 the multispectral sensor can be used to improve the efficiency of existing thinning devices. 30 Additionally, it could as well be used as a stand-alone sensor for early-season yield estimation. 31 32
doi:10.1016/j.compag.2015.01.015 fatcat:ifxvnegwkzdphkwchrmw4ycroe