Using colour features of cv. ‘Gala’ apple fruits in an orchard in image processing to predict yield

New apple fruit recognition algorithms based on colour features are presented to estimate the number of fruits and develop models for early prediction of apple yield, in a multi-disciplinary approach linking computer science with agricultural engineering and horticulture as part of precision agriculture. Fifty cv. ‘Gala’ apple digital images were captured twice, i.e. after June drop and during ripening, on the preferred western side of the tree row with a variability of between 70 and 170 fruit per tree, under natural daylight conditions at Bonn, Germany. Several image processing algorithms and fruit counting algorithms were used to analyse the apple images. Finally, an apple recognition algorithm with colour difference R − B (red minus blue) and G − R (green minus red) was developed for apple images after June drop, and two different colour models were used to segment ripening period apple images. The algorithm was tested on 50 images of trees in each period. Close correlation coefficients R2 of 0.80 and 0.85 were obtained for two developmental periods between apples detected by the fruit counting algorithm and those manually counted. Two sets of data in each period were used for modelling yield prediction of the apple fruits. In the calibration data set, the R2 values between apples detected by the fruit counting algorithm and actual harvested yield were from 0.57 for young fruit after June drop to 0.70 in the fruit ripening period. In the validation data set, the R2 value between the number of apples predicted by the model and actual yield at harvest ranged from 0.58 to 0.71. The proposed model showed great potential for early prediction of yield for individual trees of apple and possibly other fruit crops.