Characterizing the crack stability to predict the behaviour of ceramics designed for industrial use is a challenging issue. It requires accurate crack tip detection during the controlled crack propagation of notched bending tests. Different indirect methods are available, like for instance the compliance technique. Recently, techniques based on digital image correlation (DIC) have emerged: finite-element DIC (FE-DIC) with a finite element decomposition of the displacement field, integrated-DIC (I-DIC) based on Williams' series decomposition of the displacement field and regularized-DIC (R-DIC) for mechanical constraints. These full-field techniques enable the quantification of the crack length and the stress intensity factor K I . In this paper, these four methods are compared in terms of measurements of crack lengths and stress intensity factors during a notched bending test. The tested material is a damageable quasi-brittle ceramic at room temperature. The non linearity of the stress-strain law of this microcraked ceramic results in a complex behaviour that is not captured by the compliance method during the bending test. Therefore the linear elastic compliance method leads to a different estimation of crack lengths and stress intensity factors compared to DIC methods. On the other hand, the R-DIC approach handles the non linear material constitutive behaviour. It allows a deeper analysis of the mechanical fields, the energy dissipation and the damage mechanisms during the crack propagation.