A framework of bin-based classifier fusion for score level fusion of multibiometrics is proposed. It recovers identity information lost in feature extraction and template matching phase. Recovered information is well combined by ensemble learning methods.