Reflective filtered backprojection: numerical results

Jean-Baptiste Bellet (1) , Gérard Berginc (2)

1 IECL - Institut Élie Cartan de Lorraine
2 Thalès Optronique

This paper gives new insights on reflective tomography. It provides original numerical results both in 2D and 3D. They show that using the filtered backprojection on reflective images is a sensitivity analysis. By combination of the contrasts of the original images, this approach constructs indeed a volume whose most intense voxels are located near the surfaces of the original scene. Thus the combination of the FDK algorithm with the Maximum Intensity Projection is very relevant to generate new views of the considered scene. On numerical examples, we investigate the effect of the input contrasts, we show the robustness with respect to angle-dependancy and we show the stability. We also apply the method for a recognition problem from optronic images.

Mots clés: Reflective tomography; FDK algorithm; optronic recognition; Maximum Intensity Projection

Domaines: Traitement du signal et de l'image [eess.SP]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Optique / photonique

Jean-Baptiste Bellet, Gérard Berginc. Reflective filtered backprojection: numerical results. 2015. ⟨hal-01207522⟩