TMI Detailed instructions

Detailed instructions

Each folder (Stereo_SD_<ds>_<nb>) contains the following files

  • Stereo_SD_<ds>_<nb>_IMG_left: left camera image

  • Stereo_SD_<ds>_<nb>_IMG_right: right camera image

  • Stereo_SD_<ds>_<nb>_IMG_REC_left: left camera image rectified

  • Stereo_SD_<ds>_<nb>_IMG_REC_right: right camera image rectified

  • Stereo-SD_<ds>_<nb>_MASK_left: Mask for left image to be used for correspondence search. It excludes all markers and other artificial objects that could serve as features. (1: pixel to be excluded)

  • Stereo_SD_<ds>_ <nb>_MASK_right: Mask for right image to be used for correspondence search. It excludes all markers and other artificial objects that could serve as features. (1: pixel to be excluded)

  • Stereo_SD_<ds>_<nb>_MASK_REC_left: Mask for rectified left image to be used for correspondence search. It excludes all markers and other artificial objects that could serve as features. (1: pixel to be excluded)

  • Stereo_SD_<ds>_<nb>_MASK_REC_right: Mask for rectified right image to be used for correspondence search. It excludes all markers and other artificial objects that could serve as features. (1: pixel to be excluded)

  • Stereo_SD_<ds>_<nb>_CT_cut: Reference surface extracted from CT data

  • Stereo_SD_<ds>_<nb>_CT_cut_coverage.mps : Point set required to compute the surface coverage as defined in [1]

  • Stereo_SD_<ds>_ <nb>_MASK_eval: Mask for left image to be used for validation only. It includes only those pixels (1: pixel to be included) that have a corresponding point on the reference surface. Details can be found in [1].

  • Stereo_SD_<ds>_<nb>_MASK_REC_eval: Mask for rectified left image to be used for validation only. It includes only those pixels (1: pixel to be included) that have a corresponding point on the reference surface. Details can be found in [1].

  • Stereo_SD_<ds>-<nb>_Calibration.txt: contains the calibration parameters for every dataset in a format described here.

To use the validation data, we suggest using our validation tool. Alternatively, we propose applying the following procedure:

  1. Generate a disparity map/establish image correspondences using the original or rectified images. The masks (MASK_left/right or MASK_REC_left/right) should be applied to exclude the markers.

  2. Perform triangulation using the calibration data provided. In this step, only the region represented by the evaluation masks (MASK_eval or MASK_REC_eval) should be considered.

  3. Compare triangulated surface with reference surface (CT-cut) provided.

References:

[1] Maier-Hein L, Groch A, Bartoli A, Bodenstedt S, Boissonnat G, Chang PL, Clancy NT, Elson DS, Haase S, Heim E, Hornegger J, Jannin P, Kenngott H, Kilgus T, Müller-Stich B, Oladokun D, Röhl S, Dos Santos TR, Schlemmer HP, Seitel A, Speidel S, Wagner M, Stoyanov D. Comparative Validation of Single-shot Optical Techniques for Laparoscopic 3D Surface Reconstruction. IEEE T Med Imag (in press), 2014