Anterior|Próximo

Método de registro de nuvem de pontos baseado em extração multifuncional de pontos-chave e otimização de pares de pontos por triângulos semelhantes

  • role： First author第一作者
  • Affiliation：

    College of Geography and Environment, Jiangxi Normal University, Nanchang 330022, China

    Nanchang Base of International Centre on Space Technologies for Natural and Cultural Heritage under the Auspices of UNESCO, Nanchang 330022, China

    Research Center for Linguistic Spatial Information Science, Jiangxi Normal University, Nanchang 330022, China

  • Email：tuhaowen0204@126.com
  • Introduction：涂浩文，研究方向为点云配准与建模。E-mail： tuhaowen0204@126.com
TU Haowen123,  
  • role： Corresponding author通信作者
  • Affiliation：

    College of Geography and Environment, Jiangxi Normal University, Nanchang 330022, China

    Nanchang Base of International Centre on Space Technologies for Natural and Cultural Heritage under the Auspices of UNESCO, Nanchang 330022, China

    Research Center for Linguistic Spatial Information Science, Jiangxi Normal University, Nanchang 330022, China

  • Email：wjhgis@126.com
  • Introduction：吴建华，研究方向为空间数据智能集成、地理信息工程、语言空间智能。E-mail： wjhgis@126.com
WU Jianhua123*,  
  • Affiliation：

    College of Geography and Environment, Jiangxi Normal University, Nanchang 330022, China

    Nanchang Base of International Centre on Space Technologies for Natural and Cultural Heritage under the Auspices of UNESCO, Nanchang 330022, China

WANG Yuan12

Resumo

Nos algoritmos existentes de registro de nuvem de pontos, o cálculo de características dos pontos geralmente utiliza vizinhança fixa, dificultando a aplicação ao cálculo de características de nuvens de pontos complexas, o que resulta em extração de pontos-chave de baixa qualidade e presença de muitos pontos externos nos pontos correspondentes, afetando a precisão do registro. Para isso, este artigo propõe um método de registro de nuvem de pontos baseado na extração de pontos-chave multifuncionais e na otimização de pares de pontos por triângulos semelhantes. O cerne do método é extrair pontos-chave por meio de uma estratégia adaptativa de vizinhança e otimizar o modelo de similaridade de correspondência. Primeiro, utiliza-se a amostragem descendente por voxels e filtragem mista para pré-processar os dados, determina-se a vizinhança ótima da nuvem de pontos com base na função de entropia de características, e os pontos-chave são extraídos combinando o desvio padrão do ângulo dos vetores normais no vizinho, anisotropia e curvatura; em seguida, as características dos pontos-chave são calculadas usando o histograma direcional (SHOT); depois, pares iniciais de pontos correspondentes são construídos usando a razão da distância dos vizinhos mais próximos bidirecionais (BNNDR), e um modelo computacional baseado em triângulos semelhantes é proposto para otimizar os pares e realizar o registro bruto. Por fim, o algoritmo ICP ponto a plano com restrição do ângulo dos vetores normais é usado para realizar o registro fino. O método foi testado com o conjunto de dados Stanford, e os resultados mostram que apresenta o menor erro em comparação com os algoritmos clássicos ICP, K-4PCS baseado em pontos-chave e SAC-IA combinados com ICP. Além disso, o método também demonstrou vantagens significativas em aplicações de registro com dados reais.

Palavras-chave

vizinhança adaptativa;registro de nuvem de pontos;triângulos semelhantes;ICP ponto a plano

References

  1. 1.
    Abdi H and Williams L J. 2010. Principal component analysis. WIREs: Computational Statistics, 2(4): 433-459
  2. 2.
    Besl P J and McKay N D. 1992. Method for registration of 3-D shapes//Proceedings Volume 1611, Sensor Fusion IV: Control Paradigms and Data Structures. Boston: SPIE: 586-606
  3. 3.
    Chen Y and Medioni G. 1992. Object modelling by registration of multiple range images. Image and Vision Computing, 10(3): 145-155
  4. 4.
    Chetverikov D, Svirko D, Stepanov D and Krsek P. 2002. The trimmed iterative closest point algorithm//2002 International Conference on Pattern Recognition. Quebec City: IEEE: 545-548
  5. 5.
    Demantké J, Mallet C, David N and Vallet B. 2012. Dimensionality based scale selection in 3D lidar point clouds. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXXVIII-5-W12: 97-102
  6. 6.
    Huang Z J, Wen Y X, Wang Z H, Ren J J and Jia K. 2024. Surface reconstruction from point clouds: a survey and a benchmark. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(12): 9727-9748
  7. 7.
    Jia C Q, Pang M and Han X. 2023. A feature extraction algorithm based on optimal neighborhood size. Circuits, Systems, and Signal Processing, 42(4): 2193-2214
  8. 8.
    Li Y X, Guo J M, Pan S Y, Lü L L, Lu Z X and Zhang D. 2020. A point cloud registration algorithm based on ISS-SHOT features. Bulletin of Surveying and Mapping, (4): 21-26
  9. 9.
    Liu R F, Wang F, Ren H W, Wang M Y and Yang J B. 2022. Road scene laser point cloud registration method based on geographical object features. Chinese Journal of Lasers, 49(18): 1810002
  10. 10.
    Luo R, Zhou Z X, Chu X, Ma W and Meng J H. 2022. 3D deformation monitoring method for temporary structures based on multi-thread LiDAR point cloud. Measurement, 200: 111545
  11. 11.
    Salti S, Tombari F and Di Stefano L. 2014. SHOT: unique signatures of histograms for surface and texture description. Computer Vision and Image Understanding, 125: 251-264
  12. 12.
    Tombari F, Salti S and Di Stefano L. 2013. Performance evaluation of 3D keypoint detectors. International Journal of Computer Vision, 102(1): 198-220
  13. 13.
    Wang M J, Yi F, Li L and Huang C J. 2022. Local neighborhood feature point extraction and matching for point cloud alignment. Infrared and Laser Engineering, 51(5): 20210342
  14. 14.
    Wang W B, Tian M Y, Yu J Y, Song C H, Li J R and Zhou M L. 2022. Improved iterative nearest point point cloud alignment method. Laser and Optoelectronics Progress, 59(2): 0215007
  15. 15.
    Weinmann M, Jutzi B and Mallet C. 2014. Semantic 3D scene interpretation: a framework combining optimal neighborhood size selection with relevant features. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, II-3: 181-188
  16. 16.
    Xiong F G. 2018. Research on Registration Technology of 3D Point Cloud. Taiyuan: North University of China
  17. 17.
    Xu G X, Pang Y J, Bai Z X, Wang Y L and Lu Z W. 2021. A fast point clouds registration algorithm for laser scanners. Applied Sciences, 11(8): 3426
  18. 18.
    Yang J L, Li H D and Jia Y D. 2013. Go-ICP: solving 3D registration efficiently and globally optimally//Proceedings of the 2013 IEEE International Conference on Computer Vision. Sydney: IEEE: 1457-1464
  19. 19.
    Yang J Q, Zhang S K, Fan S C and Cao Z G. 2022. Survey on multi-view point cloud registration algorithm. Journal of Huazhong University of Science and Technology (Nature Science Edition), 50(11): 16-34, 43
  20. 20.
    Yang S, Hou M L and Li S N. 2023. Three-dimensional point cloud semantic segmentation for cultural heritage: a comprehensive review. Remote Sensing, 15(3): 548
  21. 21.
    Zeybek M and Şanlıoğlu İ. 2019. Point cloud filtering on UAV based point cloud. Measurement, 133: 99-111
  22. 22.
    Zhang J Y, Yao Y X and Deng B L. 2022b. Fast and robust iterative closest point. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(7): 3450-3466
  23. 23.
    Zhang S J, Ma Y Z and Zhao H F. 2019. Elimination algorithm of mismatched point pairs based on point cloud geometric feature. Computer Engineering, 45(4): 163-168
  24. 24.
    Zhang Z Y, Zheng J Y, Tao Y Y, Xiao Y, Yu S M, Asiri S, Li J C and Li T S. 2022a. Traffic sign based point cloud data registration with roadside LiDARs in complex traffic environments. Electronics, 11(10): 1559
  25. 25.
    Zhu J H, Jiang Z T, Evangelidis G D, Zhang C Q, Pang S M and Li Z Y. 2019. Efficient registration of multi-view point sets by K-means clustering. Information Sciences, 488: 205-218

Leia o texto completo

The above content is generated by Large Model Translation. The translated content is for reference only. We do not assume any commercial or legal responsibilty for any consequences arising from the use of our website