• Laser & Optoelectronics Progress
  • Vol. 56, Issue 9, 090005 (2019)
Xuegang Luo1、2, Junrui Lü1, and Zhenming Peng2、*
Author Affiliations
  • 1 School of Mathematics and Computer Science, Panzhihua University, Panzhihua, Sichuan 617000, China
  • 2 School of Information and Communication Engineering, University of Electronic Science and Technology of China, Chengdu, Sichuan 610054, China
  • show less
    DOI: 10.3788/LOP56.090005 Cite this Article Set citation alerts
    Xuegang Luo, Junrui Lü, Zhenming Peng. Recent Research Progress of Superpixel Segmentation and Evaluation[J]. Laser & Optoelectronics Progress, 2019, 56(9): 090005 Copy Citation Text show less
    Overall process of DES algorithm
    Fig. 1. Overall process of DES algorithm
    Region update strategy for KMM algorithm. (a) Positions of superpixels C1 and C2 after t iterations; (b) movement of center point; (c) region update
    Fig. 2. Region update strategy for KMM algorithm. (a) Positions of superpixels C1 and C2 after t iterations; (b) movement of center point; (c) region update
    Overall processing process of CAS
    Fig. 3. Overall processing process of CAS
    Difference between circular metric and SRC metric. (a) Example of circular metric; (b) example of SRC metric
    Fig. 4. Difference between circular metric and SRC metric. (a) Example of circular metric; (b) example of SRC metric
    Comparison of BR values of algorithms under different numbers of superpixels
    Fig. 5. Comparison of BR values of algorithms under different numbers of superpixels
    Comparison of UER values of algorithms under different numbers of superpixels
    Fig. 6. Comparison of UER values of algorithms under different numbers of superpixels
    Comparison of CO values of algorithms under different numbers of superpixels
    Fig. 7. Comparison of CO values of algorithms under different numbers of superpixels
    Image segmentation results with strong regularity structure by several algorithms (about 300 superpixels). (a) Original image; (b) LSC; (c) SNIC; (d) DBSCAN; (e) FLIC; (f) CAS; (g) GMMS; (h) MBS; (i) SCALP
    Fig. 8. Image segmentation results with strong regularity structure by several algorithms (about 300 superpixels). (a) Original image; (b) LSC; (c) SNIC; (d) DBSCAN; (e) FLIC; (f) CAS; (g) GMMS; (h) MBS; (i) SCALP
    Image segmentation results with rich texture structure by several algorithms (about 500 superpixels).(a) Original image; (b) LSC; (c) SNIC; (d) DBSCAN; (e) FLIC; (f) CAS; (g) GMMS; (h) MBS; (i) SCALP
    Fig. 9. Image segmentation results with rich texture structure by several algorithms (about 500 superpixels).(a) Original image; (b) LSC; (c) SNIC; (d) DBSCAN; (e) FLIC; (f) CAS; (g) GMMS; (h) MBS; (i) SCALP
    Xuegang Luo, Junrui Lü, Zhenming Peng. Recent Research Progress of Superpixel Segmentation and Evaluation[J]. Laser & Optoelectronics Progress, 2019, 56(9): 090005
    Download Citation