Journal of Infrared and Millimeter Waves, Volume. 39, Issue 2, 263(2020)

Superpixel segmentation method of high resolution remote sensing images based on hierarchical clustering

Liang HUANG1,2, Bing-Xiu YAO1、*, Peng-Di CHEN1, Ai-Ping REN1, and Yan XIA1
Author Affiliations
  • 1Faculty of Land Resource Engineering, Kunming University of Science and Technology, Kunming650093, China
  • 2Surveying and Mapping Geo-Informatics Technology Research Center on Plateau Mountains of Yunnan Higher Education, Kunming650093, China
  • show less
    Figures & Tables(12)
    Flow chart of remote sensing image segmentation
    Gradient distribution of experimental images(a) remote sensing image, (b) sobel gradient image, (c) gradient distribution, (d) remote sensing image, (e) sobel gradient image, (f) gradient distribution
    Segmentation results using AMR-WT by changing the value of s and m(a) s = 1, m = 1; (b) s = 2, m = 1; (c) s = 3, m = 1; (d) s = 4, m = 1; (e) S1s = 1, m =1; (f) S2s = 1, m = 6; (g) s = 1, m = 12; (h) s = 1, m = 20
    Simulation graph of clustering(a)clustering simulation, (b)segmentation results, (c)clustering simulation, (d)segmentation results
    Test data sets (a) S1 imag, (b) S2 image, (c) S3 image, (d) S4 image, (e) S1 image, (f) S2 reference image, (g) S3 reference image, (h) S4 reference image
    Experimental results of over-segmentation(a) S1 proposed superpixel algorithm, (b) S1 SLIC algorithm, (c) S1 LSC algorithm, (d) S1 Meanshift algorithm, (e) S2 proposed superpixel algorithm, (f) S2 SLIC algorithm, (g) S2 LSC algorithm, (h) S2 MeanShift algorithm, (i) S3 proposed superpixel algorithm, (j) S3 SLIC algorithm, (k) S3 LSC algorithm, (l) S3 Meanshift algorithm, (m) S4 proposed superpixel algorithm, (n) S4 SLIC algorithm, (o) S4 LSC algorithm, (p) S4 MeanShift algorithm
    The time of over-segmentation
    Segmentation results of S1 and S2 (a) S1 FNEA 50, (b) S1 FNEA 100, (c) S1 SH method, (d) result of S1 by proposed method, (e) S2 FNEA 50, (f) S4 FNEA 100, (g) S2 SH method, (h) result of S2 by proposed method
    Segmentation results of Test1
    Segmentation results of S3 and S4(a) S3 FNEA 50, (b) S3 FNEA 100, (c) S3 SH method, (d) S3 result by proposed method, (e) S4 FNEA 50, (f) S4 FNEA 100, (g) S4 SH method, (h) S4 result by proposed method
    Segmentation results of Test2
    • Table 1. Information of segmentation data sets

      View table
      View in Article

      Table 1. Information of segmentation data sets

      序号传感器大小/像素分辨率
      S1QuickBird433×5500.61 m
      S2QuickBird2480×9750.61 m
      S3无人机726×4680.05 m
      S4无人机979×5860.05 m
    Tools

    Get Citation

    Copy Citation Text

    Liang HUANG, Bing-Xiu YAO, Peng-Di CHEN, Ai-Ping REN, Yan XIA. Superpixel segmentation method of high resolution remote sensing images based on hierarchical clustering[J]. Journal of Infrared and Millimeter Waves, 2020, 39(2): 263

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Remote Sensing Technology and Application

    Received: Jul. 25, 2019

    Accepted: --

    Published Online: Apr. 29, 2020

    The Author Email: Bing-Xiu YAO (1366711008@qq.com)

    DOI:10.11972/j.issn.1001-9014.2020.02.014

    Topics