Crosstalk cancellation results – Train
Original images (left, right):
——————-
Images with added crosstalk: In order to validate the proposed model’s accuracy, images with crosstalk superposed digitally according to model (2) from the paper have been generated and compared in single-eye test with images containing naturally-superposed crosstalk (leakage).
——————-
Images after crosstalk cancellation (with “anti-crosstalk”): Images below have been processed by three variants of the proposed algorithm: non-negativity constraint only (no pre-processing), non-linear adjustment and linear adjustment of intensity range.
No pre-processing:
Non-linear pre-processing with “alpha”=30:
Non-linear pre-processing with “alpha”=60:
Linear pre-processing with “alpha”=30:
Linear pre-processing with “alpha”=60:
