Fracture strain data provide essential information for material selection and serve as an important failure criterion in the computer simulations of crash events. Traditionally, the fracture strain was measured by evaluating the thinning at fracture using tools such as microscopes or a point micrometer. In the recent decades, digital image correlation (DIC) has evolved as a state-of-the-art optical system to record full-field strain history of materials. Using this method, a complete set of the fracture strains (including major, minor, and thinning strains) can be approximated for the material with the recorded data prior to a visible crack. However, results directly obtained with DIC can be dependent on experimental setup and evaluation parameters, which potentially introduce errors to the reported results. To evaluate the capability of the DIC for fracture strain measurement, a validation study was performed to compare the fracture strains of a 980GEN3 steel measured with DIC and microscope, respectively. Briefly, the specimens were formed to fracture using a testing configuration that was directly transferable from an established formability test. Marciniak cup tool was employed to deform the specimens with designed sample width to achieve different strain conditions (i.e., uniaxial tension, plane strain, and equi-biaxial stretch). The deviation was identified in fracture thinning strain measurement using DIC from that obtained with a microscope, especially under plane strain condition, demonstrating the limitation of the current DIC setup used in this work. In an effort to improve the accuracy of DIC fracture strain measurements, an optimized approach combining DIC with the thinning measurement is proposed and practiced. Using the newly developed method, fracture strain data for different advanced high strength steel grades were successfully generated and compared.