Accurate identification of individual all-red Red Fuji apple fruit based on digital image processing

Individual traceability of agricultural products is an important research field in modern agricultural informatization. However, existing identification technologies, such as QR codes and barcodes, have drawbacks such as susceptibility to information tampering and resource wastage. Therefore, this paper proposes a digital identification solution to address these issues. Using the commonly available all-red Red Fuji apples as experimental materials, we propose an accurate individual identification solution based on the uniqueness and invariance of apple surface features. It combines the Speeded-Up Robust Features (SURF) algorithm with the Fast Library for Approximate Nearest Neighbors (FLANN) algorithm, improved by Random Sample Consensus (RANSAC). First, 30 images of all-red Red Fuji apples in the experimental group are acquired and pre-processed. The images are then tagged, matched, and analyzed using the proposed method. The results show that the method achieved accurate identification of individual apples, with recognition accuracy unaffected by changes in shooting angles or storage duration. The method was then extended to a test group of 160 apples for further validation, yielding consistent results. The results confirm that the solution presented in this paper demonstrates robust stability and accuracy, enabling accurate identification of individual all-red Red Fuji apples.