An Approach for Simultaneous Integration of Range Images in Binocular Stereo Vision

A new integration algorithm is proposed for combining multi-view range images into a single mesh surface to completely describe the object measured.By the algorithm,each range image is in turn treated as a benchmark image.Based on the benchmark images the overlapped correspondence relationship of multi-view is defined.Then the overlapped points are merged and the mesh on the benchmark image is kept.Finally a uniform waterproof surface is generated by stitching the gaps between the benchmark images.The algorithm can make the final integrated surface closer to the actual measured object by simultaneously integrating multi-view range images and by optimizing the overlapped regions.The integration experiments demonstrate the validity and efficiency of the approach.