Digital Image Correlation (DIC) is one of the most used methods for displacement measurement. However, it was shown that the measured error increases in particular zones: sample edges and cracks. Until now, three techniques have been used to avoid this error: limit the correlation domain in the sample, use an adapted mesh and exclude background pixels with a mask. The purpose here is to use an edge position measurement method, the Virtual Image Correlation (VIC), to help the DIC. This method allows describing a contour in a digital image with an analytical model – such as a circle, BSpline, NURBS... – with an accuracy close to 1/1000 pixel. First, this boundary position knowledge is used to create automatically a mesh adapted to the sample geometry. Then, it allows generating an accurate pixel mask by excluding pixels using a criterion on the signed distance between the boundary position and their centers. A sensitivity analysis of this criterion is carried out to choose its optimal value. Compared to segmentation methods this technique allows controlling which pixels are excluded based on their position instead of their gray level. Synthetic tests show that the pixel mask allows a significant reduction of the displacement and strain errors close to the boundaries. Besides, the boundary position information in each deformed image was used to constrain the DIC problem. However, this constrained method revealed to be less effective than with a simple mask. Due to the accuracy of the VIC, the adapted mesh and the pixel mask can be created even if the image quality is not perfect.