A Study of Measures for Contour-based Recognition and Localization of Known Objects in Digital Images

Baptiste Magnier (1) , Hasan Abdulrahman (2)

1 LGI2P - Laboratoire de Génie Informatique et Ingénierie de Production
2 Northern Technical University

Usually, the most important structures in an imageare extracted by an edge detector. Once extracted edges arebinarized, they represent the shape boundary information ofan object. For the edge-based localization/matching process,the differences between a reference edge map and a candidateimage are quantified by computing a performance measure. Thisstudy investigates supervised contour measures for determiningthe degree to which an object shape differs from a desiredposition. Therefore, several distance measures are evaluated fordifferent shape alterations: translation, rotation and scale change.Experiments on both synthetic and real images exhibit whichmeasures are accurate enough for an object pose or matchingestimation, useful for robot task as to refine the object pose.

Mots clés: contours; similarity measures; Distance transform

Domaines: Traitement des images [eess.IV]

Baptiste Magnier, Hasan Abdulrahman. A Study of Measures for Contour-based Recognition and Localization of Known Objects in Digital Images. IPTA 2018 - Image Processing Theory, Tools and Applications, Nov 2018, Xi'an, China. ⟨10.1109/IPTA.2018.8608165⟩. ⟨hal-01942291⟩