Due to security and oversight requirements, identity verification of personnel is needed in industrial operation scenes. However, in electric power operation, it is difficult to realize accurate identity verification because of special dress and the complex scene. Current verification methods are mostly based on face verification technology, which is susceptible to factors like clothing occlusion, detection distance and detection angles. Comparative to face features, body features are more stable and reliable. In this thesis, a multi-feature identity verification method, combining face and body features, is proposed. It effectively increases accuracy of identity verification in single person scenes, and achieves 99.17% classification accuracy under the four-folds cross-validation set of multi-view case on the CASIA-A dataset, where the classification accuracy of the three viewing angles are 98.75%, 100% and 98.75% respectively.