Almost one-thirdof the selected samples are selected randomly as testing samples,and the other selected samples are used as training samples. Thenthe MEC-BP-Adaboost neural network-based color correction algo-rithm generates the mapping model with training samples. Hencea number of training processes are conducted, producing multipleBP neural network-based color correction models with the opti-cal weights and thresholds that are acquired by mind evolutionarycomputation