• Spectroscopy and Spectral Analysis
  • Vol. 41, Issue 9, 2980 (2021)
Zhi-fen WANG1、*, Wei-kuan JIA1、1; 2; *;, Shan-hao MOU1、1;, Su-juan HOU1、1; 2;, Xiang YIN3、3;, Ji ZE4、4;, and [in Chinese]4、4;
Author Affiliations
  • 11. School of Information Science and Engineering, Shandong Normal University, Ji'nan 250358, China
  • 33. School of Agricultural Engineering and Food Science, Shandong University of Technology, Zibo 255000, China
  • 44. School of Engineering, Cardiff University, Cardiff, CF24 3AA, United Kingdom
  • show less
    DOI: 10.3964/j.issn.1000-0593(2021)09-2980-09 Cite this Article
    Zhi-fen WANG, Wei-kuan JIA, Shan-hao MOU, Su-juan HOU, Xiang YIN, Ji ZE, [in Chinese]. KDC: A Green Apple Segmentation Method[J]. Spectroscopy and Spectral Analysis, 2021, 41(9): 2980 Copy Citation Text show less
    Flow chart of improved density peak clustering green apple segmentation algorithm
    Fig. 1. Flow chart of improved density peak clustering green apple segmentation algorithm
    Green apple image under various real conditions
    Fig. 2. Green apple image under various real conditions
    Green apple image and super pixel segmentation of green apple image(a): Green apple image, size: 6 000×4 000;(b): Green apple image after super pixel segmentation, a total of 3172 super pixel regions
    Fig. 3. Green apple image and super pixel segmentation of green apple image
    (a): Green apple image, size: 6 000×4 000;(b): Green apple image after super pixel segmentation, a total of 3172 super pixel regions
    (a) Original image; (b) Area under the R channel with a pixel value greater than 130; (c) Area under the G channel with a pixel value greater than 130; (d) Area under the B channel with a pixel value greater than 130
    Fig. 4. (a) Original image; (b) Area under the R channel with a pixel value greater than 130; (c) Area under the G channel with a pixel value greater than 130; (d) Area under the B channel with a pixel value greater than 130
    Numerical distribution of green apple target and background in R-B image and G-B image
    Fig. 5. Numerical distribution of green apple target and background in R-B image and G-B image
    The regional mean feature space of the green apple image area on the R-B image and the G-B image
    Fig. 6. The regional mean feature space of the green apple image area on the R-B image and the G-B image
    Local density distribution of data points
    Fig. 7. Local density distribution of data points
    Decision diagram (the horizontal axis is the local density and the vertical axis is the distance)
    Fig. 8. Decision diagram (the horizontal axis is the local density and the vertical axis is the distance)
    The ratio of the number of cluster pixels to the total number of pixels (percentage)
    Fig. 9. The ratio of the number of cluster pixels to the total number of pixels (percentage)
    Algorithm output imag(a): Original green apple image; (b): Improved density peak clustering result map;(c): Improved density peak clustering segmentation map
    Fig. 10. Algorithm output imag
    (a): Original green apple image; (b): Improved density peak clustering result map;(c): Improved density peak clustering segmentation map
    Segmentation effect diagram of the method
    Fig. 11. Segmentation effect diagram of the method
    Results of five clustering segmentation image segmentation algorithms
    Fig. 12. Results of five clustering segmentation image segmentation algorithms
    METHODFNRFPRSSSE
    Our method14.662.3287.6988.53
    K-means41.837.2562.6958.16
    Meanshift39.0613.5672.4854.83
    Fcm50.063.8766.5849.13
    Sifdp23.4410.5271.3876.06
    Table 1. Statistics of evaluation indicators of five clustering segmentation algorithms
    Zhi-fen WANG, Wei-kuan JIA, Shan-hao MOU, Su-juan HOU, Xiang YIN, Ji ZE, [in Chinese]. KDC: A Green Apple Segmentation Method[J]. Spectroscopy and Spectral Analysis, 2021, 41(9): 2980
    Download Citation