• Laser & Optoelectronics Progress
  • Vol. 58, Issue 2, 0210020 (2021)
Baoyan Shan1、2、3, Zhencai Zhu1、*, Yonghe Zhang1、3, and Chengbo Qiu1、2、3
Author Affiliations
  • 1Innovation Academy for Microsatellites of Chinese Academy of Sciences, Shanghai 201203, China
  • 2University of Chinese Academy of Sciences, Beijing 100049, China
  • 3Key Laboratory of Microsatellite, Chinese Academy of Sciences, Shanghai 201203, China
  • show less
    DOI: 10.3788/LOP202158.0210020 Cite this Article Set citation alerts
    Baoyan Shan, Zhencai Zhu, Yonghe Zhang, Chengbo Qiu. A Real-Time SIFT Algorithm for Planetary Surface Feature Extraction[J]. Laser & Optoelectronics Progress, 2021, 58(2): 0210020 Copy Citation Text show less
    Flow chart of proposed algorithm
    Fig. 1. Flow chart of proposed algorithm
    Gaussian kernel function implementation process
    Fig. 2. Gaussian kernel function implementation process
    Key point extraction parallelization process
    Fig. 3. Key point extraction parallelization process
    Error distribution
    Fig. 4. Error distribution
    Feature point matching results of four algorithms. (a) SIFT algorithm; (b) SiftGPU algorithm; (c) SURF algorithm; (d) FG-SIFT algorithm
    Fig. 5. Feature point matching results of four algorithms. (a) SIFT algorithm; (b) SiftGPU algorithm; (c) SURF algorithm; (d) FG-SIFT algorithm
    Time-consuming comparison of feature point extraction. (a) Time-consuming statistics; (b) algorithm speedup ratio
    Fig. 6. Time-consuming comparison of feature point extraction. (a) Time-consuming statistics; (b) algorithm speedup ratio
    Time-consuming comparison of building Gaussian pyramid. (a) Time-consuming statistics; (b) algorithm speedup ratio
    Fig. 7. Time-consuming comparison of building Gaussian pyramid. (a) Time-consuming statistics; (b) algorithm speedup ratio
    Bank0Bank1Bank2Bank3Bank4
    (0,0)→(0,7)(1,0)→(1,7)(2,0)→(2,7)(3,0)→(3,7)(4,0)→(4,7)
    (0,8)→(0,15)(1,8)→(1,15)(2,8)→(2,15)(3,8)→(3,15)(4,8)→(4,15)
    Table 1. Shared memory layout in block
    Numberσ valueAverage error /%Maximum error /%
    11.9030.8107.059
    22.2630.4877.059
    32.6930.3337.451
    43.2000.8237.059
    53.8050.5067.843
    64.5250.4938.627
    Table 2. Error statistics of different σ values
    AlgorithmImageGroup 1Group 2Group 3Group 4
    Real angle /(°)4.278.8214.6024.00
    Total matches693869063344830
    Effective matches679865362867428
    OpenCv_SIFTMatching accuracy /%98.094.685.751.6
    Calculated angle /(°)4.349.0314.3124.31
    Angel error /%1.642.381.991.29
    Total matches676664083147734
    Effective matches660160142715379
    SiftGPUMatching accuracy /%97.693.986.351.6
    Calculated angle /(°)4.219.0314.9124.29
    Angel error /%1.412.382.121.21
    Total matches118441258276081578
    Effective matches764459391902161
    SURFMatching accuracy /%64.547.225.010.2
    Calculated angle /(°)4.547.5010.4427.61
    Angel error /%6.3214.9728.4915.04
    Total matches691663262580726
    Effective matches648159642180379
    FG-SIFTMatching accuracy /%93.794.384.552.2
    Calculated angle /(°)4.368.5914.2624.70
    Angel error /%2.112.612.332.92
    Table 3. Matching effect of four algorithms
    Baoyan Shan, Zhencai Zhu, Yonghe Zhang, Chengbo Qiu. A Real-Time SIFT Algorithm for Planetary Surface Feature Extraction[J]. Laser & Optoelectronics Progress, 2021, 58(2): 0210020
    Download Citation