Integrated system for recognizing human face based on categorizer and method thereof

The invention discloses a face identifying system based on sorter integration and a method thereof. The system and the method aim at obtaining a better identifying rate to a face identifying system by solving the weighting coefficients of a plurality of sub-sorters to integrate a plurality of sub-sorters to output. The whole system includes a training system and a sorting system; wherein, the training system accomplishes: carrying out characteristic extracting on a face image; selecting a plurality of sub-sorters which output posterior probability, inputting different training samples to the sub-sorters to train, obtaining the posterior probability of the original training sample as well as adopting linear planning and optimizing to solve the weighting coefficients of each sub-sorter; the sorting system accomplishes: inputting the samples to be sorted after the characteristic extracting into the sub-sorters after training to obtain the posterior probability of the samples to be sorted, designing the sorting rules according to the posterior probability and the weighting coefficients of the sub-sorters as well as outputting the sorting result. The system and the method has a high identifying rate and can be used for the face identifying in the categories of machine learning and mode identifying.