Figures (5)  Tables (3)
    • Figure 1. 

      Mango individual identity anti-counterfeiting system based on biometric fingerprints.

    • Figure 2. 

      Image processing. (a) Original image. (b) Scale normalized image. (c) Gray normalized image. (d) Binary image.

    • Figure 3. 

      The ROC curves of conventional mango identification methods.

    • Figure 4. 

      Fuzzy C-means clustering and clustering center of 350 normalized images.

    • Figure 5. 

      ROC curves of the mango identification method based on Fuzzy C-means clustering.

    • Positive sampleNegative sample
      Correct identificationTP (True positive)FP (False positive)
      Error identificationFN (False negative)TN (True negative)

      Table 1. 

      Prediction results of the identification method.

    • Verification time (s)AUCEER (%)
      VGI0.61950.99930.43
      VRI0.60580.98534.80
      VRIE0.39930.98056.23

      Table 2. 

      Performance of conventional mango identification methods.

    • Verification time (s)AUCEER (%)
      VRIC0.63260.99502.86
      VRIEC0.42120.99303.43

      Table 3. 

      Performance of the mango identification method based on Fuzzy C-means clustering.