Citrus yield estimation based on images processed by an Android mobile phone

• Published in: Biosystems engineering Vol. 115; no. 2; pp. 162 - 170
• Main Authors: Gong, Aiping, Yu, Junlin, He, Yong, Qiu, Zhenjun
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.06.2013; Elsevier
• Subjects: Agricultural machinery and engineering; Agronomy. Soil science and plant productions; algorithms; Biological and medical sciences; Citrus; color; computer software; engineering; fruit yield; Fundamental and applied biological sciences. Psychology; Generalities. Biometrics, experimentation. Remote sensing; harvest date; image analysis; trees; Bioengineering; Mobile phone; Citrus fruit; Rutaceae; Process; Image; Citrus; Engineering; Dicotyledones; Angiospermae; Spermatophyta; Agriculture; Yield
• ISSN: 1537-5110; 1537-5129
• DOI: 10.1016/j.biosystemseng.2013.03.009

The advanced electronic technology makes the mobile phone suitable for citrus yield estimation with image processing techniques. This article proposes a new method to estimate the yield of citrus two weeks ahead of the harvesting season for an individual tree via image processing using an Android mobile phone (AMP). The procedure is essentially fruit-counting algorithm software developed in Java that uses AMP touch panel monitor to estimate citrus yield. The target photos were taken in natural conditions by the AMP. The image processing was used to choose colour, segment image, generate binarisation image, remove noise, and estimate fruit number. A modified 8-connectedness chain code to estimate the number of clustered citrus was introduced. The performance of the fruit-counting algorithm software was validated with 40 samples of citrus trees and the method achieved good precision and recognition ratio of 90%. The results indicated that the approach could be utilised to estimate the fruit yield of an individual tree which is valuable information to forecast yields, plan harvest schedules and generate prescription maps for site-specific management practices on an individual tree basis within the grove. Aided by the proposed approach, one can estimate the citrus yield without expensive equipment. •Android mobile phone vision technology is applied to estimate citrus yield.•The mechanism lies in the yellow pixels of citrus which can be discriminated easily.•Image segmentation suitable for handheld device with touch panel is introduced.•A modified Freeman chain code is used to estimate the number of clustered fruits.