3D Reconstruction based on the matching method of GA

Stereo matching is an important part of three-dimensional reconstruction (3D). Because of the problem that the traditional matching methods have a lower matching precision, and it causes a poor robustness of reconstructed result, this paper presents a matching method that is combined the method of Euclidean distance between feature points as the similarity measurement and genetic algorithm (GA). Through the Euclidean distance between feature points, it gets the initial set of matching points, and then uses GA to estimate the transformation matrix between the two to be matched images. After that it can recover the translation, rotation and scaling relations between the images accurately, and eliminate the false matching points from the initial set of matching points, then obtain the accurate set of matching points. And the experimental result shows that because of the increased matching accuracy, the robustness of the reconstructed result also has been improved.