The invention discloses an identity recognition method based on human body PPG signal segmentation which mainly solves the problem that the existing identity recognition rate based on PPG signal is low. The method comprises the steps: 1, acquiring training database and tested data; 2, pretreating the training data, and establishing a single-cycle wave-shaped library; 3, segmenting the single-cycle wave shape to acquire sub-wave-shaped sets and weight factors:4, disintegrating the sub-wave-shaped sets by using a DNMF method to acquire a base space and training sub-characteristic sets; 5, fusing the training sub-characteristics by using the weighting factors to acquire a training template library; 6, pretreating the tested data and segmenting to acquire the sub-wave-shaped sets and weight factors; 7, projecting the sub-wave shapes of the tested data on the base space to acquire test sub-characteristic sets; 8, carrying out the weighted fusion on the test sub-characteristic sets to acquire test characteristic sets, and completing identity recognition by using a SVM classifier. The identity recognition method based on the human body PPG signal segmentation has the advantages of being above 99.62% in recognition rate; capable of being applied to personal identification recognition in the fields of finance and government agencies and the like.