This study reports on the Heaviside-based digital image correlation (H-DIC) procedure and its application in fracture analysis. This improved DIC procedure was proposed to solve the uncertainty problems at the vicinity of the crack and to evaluate the opening and shear movements of crack lips and the orientations of cracks in the subset. Some tests were conducted to demonstrate the performance of the H-DIC algorithm. An application on argillaceous rock mass exhibiting multiple mixed-mode fractures is shown to validate the efficiency and robustness of the proposed method. This application consisted in processing images acquired from an experimental investigation performed in a gallery front submitted to climatic seasonal variations. The results illustrated how the H-DIC procedure enables to localise and to quantify the opening, shearing and orientation of subpixel cracks. A sensibility study performed on the opening and shear components demonstrated that the precision of crack aperture by H-DIC procedure is close to 0.14 pixel and the spatial resolution is equal to one pixel. Moreover, the crack area was calculated from local apertures on a monitoring duration of 1 year, and a maximum value of 595.8 mm 2 in winter was obtained. K E Y W O R D S crack area, crack orientation, digital image correlation, fractured structure, full-field measurement, lip displacements, mixed-mode fracture 1 | INTRODUCTION Desiccation cracking, which results from the heterogeneous spatial organisation of tensile stresses, the heterogeneities of mineralogy or the orientation of swelling clay particles inside geomaterials, modifies the hydraulic properties of clayey geomaterials (e.g., Vogel et al, 1 Coppola et al, 2 and Baer et al 3). Consequently, the water movement and solute transport in unsaturated cracked clayey soils become more complicated to describe and model. Indeed, models regarding water movement in cracked soils further require both dynamic crack patterns and varying hydraulic descriptions. 3,4 Thus, knowledge about the crack morphology seems indispensable for modelling the soil cracking mechanism and water transport in unsaturated cracked soils. 4,5 In recent years, the measurement of crack patterns has evolved from field measurements to automatic recognition and quantification using digital image analysis. Researchers frequently use digital imaging methods for crack quantification, 3-5 wherein superficial crack patterns are quantified using geometric and morphological indexes. Two limits of the previous approaches based on digital imaging methods are (i) the precision of desiccation crack detection (always greater than or close to 1 pixel) and (ii) their application being limited to