Similarity learning with joint transfer constraints for person re-identification.

•A novel similarity learning method under joint transfer constraints is proposed to learn a discriminative subspace with consistent data distributions.•The mid-level features are introduced in by defining the reconstruction matrix, by an optimal function addressed via the inexact augmented Lagrange multiplier (IALM) algorithm.•During the process of objective function solution for optimization problem, based on confinement fusion of multi-view and multiple sub-regions, and a solution strategy is proposed to solve the objective function using joint matrix transform.