Study of Micro-vision Calibration Technique Based on SIFT Feature Matching - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2012

Study of Micro-vision Calibration Technique Based on SIFT Feature Matching

Tao Hu
  • Fonction : Auteur
  • PersonId : 988324
Hui-Lan Wu
  • Fonction : Auteur
  • PersonId : 988325
Guodong Liu
  • Fonction : Auteur
  • PersonId : 988326

Résumé

In the micro-vision system, precise calibration of the pixel equivalent is a prerequisite for accurate visual inspection. Traditional calibration process not only needs to take standard parts as the base, but also has strict requirements on their shape, manufacturing precision, placement and so on. Because of such shortcomings of traditional method, this paper gives an improved SIFT calibration algorithm for the pixel equivalent. Firstly, get the image’s characteristic points before and after its’ moving, using SIFT algorithm. Then filter the mismatched points through the second filtering algorithm so that the matching accuracy can be greatly improved. Compare moving distances of the characteristic point pixels to the reference distance of optical grating. Then accurate pixel equivalents can be obtained. Experiments show that the calibration algorithm is more accurate than traditional methods. So this algorithm can completely replace the standard-parts method in micro-vision system.
Fichier principal
Vignette du fichier
978-3-642-27275-2_30_Chapter.pdf (4 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01361148 , version 1 (06-09-2016)

Licence

Paternité

Identifiants

Citer

Tao Hu, Hui-Lan Wu, Guodong Liu. Study of Micro-vision Calibration Technique Based on SIFT Feature Matching. 5th Computer and Computing Technologies in Agriculture (CCTA), Oct 2011, Beijing, China. pp.270-277, ⟨10.1007/978-3-642-27275-2_30⟩. ⟨hal-01361148⟩
236 Consultations
52 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More