A Preliminary Discrimination of Cluster Disqualified Shape for Table Grape by Mono-Camera Multi-Perspective Simultaneously Imaging Approach

Abstract

Sorting table grape based on external indices of cluster can not only assure consumer satisfaction but also enhance the industry’s competitiveness and profitability. This paper proposed a method of synchronously imaging from different perspectives to present the whole surface of “kyoho” table grape with the front-surface reflective mirrors. A preliminary grade was conducted to distinguish the disqualified cluster by multi-perspective imaging method using the mirrors and mono-camera. A grape image with three bunches in every 120° perspective was simultaneously imaged by once acquisition of mono-camera, and meanwhile, the weight of grape was weighed by a weighing senor. After series of image process stages, the grape bunches of one actual and two virtual regions were segmented out and the size of virtual regions was affined to the size of the actual region. The compactness of grape cluster was estimated by the deviating percentage between the sensing weight and the fitting weight which was regressed with the average area by exponential function. The ratio of contour curve’s width was extracted from the bunch region and was used to classify the disqualified shoulder pole of grape shape. Finally, the disqualified shape of table grape was classified by the weight, region size, compactness, and ratio and got a 95.5 % correct rate. Results show this preliminary grade is feasible to inspect the disqualified cluster shape of table grape by the multi-perspective simultaneously imaging method.