The identification of the polymer's behaviour and especially rubber-like materials remains a challenging task. In this work, we propose a method for the identification which combines strain field obtained by digital image analysis and the optimisation of an approximated stress field adapted to the specimen geometry. A biaxial stretching test is performed on a crosshair rubber specimen. With the proposed approach, heterogeneity of the strain field during this equi-biaxial tension test becomes an advantage. It allows the simultaneous identification of several strain states: uniaxial, biaxial and planar elongations as well as shear. The stress field is approximated by the sum of a homogeneous field and an additional field. The latter, checking the boundary conditions on the free edge is decreasing when entering the sample. The characteristic decreasing length is optimized so that the approximate field verifies the equilibrium equations. Combining image analysis with an optimized stress field, we manage the identification the hyperelastic potential by calculating explicitly the two derivatives f=∂W/∂I1 and g=∂W/∂I2 and conclude on their dependence on I1 and I2 the two first invariants of the rigth Cauchy-Green tensor. Finally, a specific finite element algorithm has been developed to similate a 2D-incompresible hyperelastic material. This algorithm is used to validation the identification potentiel by comparing simulation results and experimental data
