Kiwifruit yield estimation using image processing by an Android mobile phone

Abstract

Advanced electronic technology makes mobile phone suitable for image processing technology on fruit yield estimation. A new method was proposed to estimate yield of kiwifruit orchard via image processing using an Android mobile phone. Image processing was used to select significant color channels and their threshold for segmentation, generate binarized image, and remove noise. A fruit-counting algorithm was developed to calculate the number of kiwifruit by identifying the number of fruit calyxes. Image area was estimated by the pixels proportion of a reference board hung under the canopy. Finally, yield was calculated by multiplying the proportion of the image in three-dimensional space with the number of kiwifruit in an image and the kiwifruit orchard area. Under natural conditions, the performance of the fruit counting algorithm software was validated with 100 kiwifruit images. The proposed method achieved a good precision and recognition ratio of 76.4%. Results showed the pixels area of the reference board could be correctly detected. It is reliable to estimate the yield of kiwifruit orchard by recognizing the number of kiwifruit and the area of the reference board. The results indicated that the approach could be utilized to estimate the fruit yield of a kiwifruit orchards, which is valuable information to forecast yield, and plan harvest schedules.