Snake-search algorithm for stereo vision reconstruction via monocular system

Inspired by the stereo vision of human-eyes, a novel method for the three-dimension reconstruction by zooming is proposed. Firstly, two images of the object are taken at distinct focal lengths by the camera with zoom lens. The detailed analysis is processed by utilizing the pinhole model and geometrical optics, which prove this method is theoretically effective. The contour of the object is extracted in the image, which is taken on a relative larger focal length. Then, that image is divided into the Macro Blocks (MB), which is the minimum unit during image processing, and the MB with sufficient contour pixels are selected as the Feature Macro Blocks (FMB). The searching and matching are carried out by an original strategy named as Snake Search (SS) and defined matching criterion. After matching, two coordinates of same object point in both images are obtained. By the geometrical computation introduced, the 3D information of the object is acquired. Digital Camera (DC) is selected as the equipment in experiment, due to it is widely used consumer electronics products which meet the requirements. Experimental results in two different situation show that the method proposed is available and the 3D reconstruction by single camera is also a potential research area for stereo reconstruction.