Image Target 3 D Information Of Rapid Extraction And Vrml Technology Research Of 3 D Reconstruction

This paper uses binocular distance to extract three-dimension information of the target quickly and using VRML language description method to rebuid the target.In order to eliminate the noise of image, a fast-mean filtering method is proposed in this paper, the speed of this method has nothing to do with the template size, only with the image size, is faster than the traditional method an order of magnitude.Stereo matching is need to extract three-dimensional information of the target, With the image captured by paralleled camera in the stereo matching module, using the proposed threshold-based matching points screening method, the speed of this method several orders of magnitude faster compared with RANSAC algorithm. The speed of new method accord with the requirement of real-time processing.In order to segment target quickly in image-segmentation, this paper applies region growing way to complete the mark by one scan, this method consuming less time compared with the traditional twice-scanning method. For the whole three-dimensional extraction this process save time.VRML language is used to rebuild objects, you can see the special relationship of objects from computer which installed VRML browser.