This study addresses the issue of feature point mismatches in stereo vision measurement caused by the lack of directionality and insufficient feature distinctiveness of circular markers. It proposes an innovative stereo vision measurement strategy that integrates active and passive techniques. By mapping and solving on the correction plane, it restores the true contour and center position of the circles. The method uses light projection technology and virtual textures to accurately locate feature points on circles, reducing mismatches and improving measurement efficiency and accuracy. Experiments validated that this method outperforms solely active or passive measurements in both simple and complex environments. Results show that at an 8000 mm testing distance, the average spatial error for high-precision targets is less than 0.05 mm, and for rod cross-sections, the absolute error in detecting hole distances is within 0.1 mm.