To investigate the effects of skin stretching stress on tissue expansion and its mechanotransduction mechanism, an animal model of skin stretching was established in our previous work. However, the distribution of stress in the skin cannot be precisely determined in real-time, which leads to an imprecise correlation analysis between stress distribution and protein levels of mechanosensitive molecules. To address this problem, a real-time suture reconstruction method with a multi-ocular stereo vision measurement system was developed in this study. The shape of the suture passing through the rat’s skin can be measured in real-time, and the distribution of skin stress can be precisely determined. The key to real-time suture reconstruction is suture-matching in multi-ocular images, therefore a suture-matching method based on the nearest-neighbor rule and region growing was proposed. First, color markers were fixed on the suture nodes and the frame, secondly, pixels in the frame region are sampled, and the frame inner region is segmented by the nearest neighbor rule. Furthermore, the pixels within the color markers in the region were selected as the seed points for growth, and the markers were segmented using a edge and color integrated detection method. Finally, the markers on the frame were fitted to form a reference line, and the distance between each marker on the suture and the reference line was calculated as the matching feature to realize image matching of nodes in the stereo camera. The three-dimensional coordinates of the nodes were calculated based on region growth, and the sutures were reconstructed using a B-spline curve. The experimental results confirmed the feasibility and high robustness of the proposed method, which ensured accurate three-dimensional reconstruction of the suture.