Evaluation criterion of thermal light ghost imaging based on the receiver operating characteristic analysis.

The performances of different thermal ghost imaging (GI) algorithms are compared in an experiment of computational GI using a digital micromirror device. Here we present a rather different evaluation criterion named receiver operating characteristic (ROC) analysis that serves as the performance of merit for the quantitative comparison. A ROC curve is created by plotting the true positive rate against the false positive rate at various threshold settings. Both theoretical analysis and experimental results demonstrate that the ROC curve and the area under the curve are better and more intuitive indicators of the performance of the GI, compared with conventional evaluation methods. Additionally, for examining gray-scale objects, the calculation of the volume under the ROC surface is analyzed and serves as a performance metric. Our scheme should attract general interest and open exciting prospects for ROC analysis in thermal GI.