Precision matching method for multiple depth image

Abstract

The present invention relates to an accurate registration method of multi-frame depth image. Said method includes the following steps: (1), judging boundary point, noise point and isolated point so as to remove point whose noise is large; (2), selecting reference point in overlay area of different depth images, creating correspondent point buffer zone for every reference point so as to search correspondent point in buffer zone; and (3), adopting iteration method to progressively optimize whole registration accuracy. Said invention also provides the concrete steps of said iteration method.

Description

Technical field [0001] The invention belongs to the technical field of computer virtual reality, and specifically it accurately registers the whole of multiple depth images, eliminates the cumulative error of pairwise registration, and is used for geometric modeling of a three-dimensional model. Background technique [0002] With the rapid development of 3D scanning technology in recent years, it has become more and more common to use 3D scanners to build 3D models of objects. The depth image with depth information scanned by the 3D scanner can only scan the model from one angle at a time, so the depth images scanned from different angles for multiple times need to be stitched, that is, registered, to obtain a complete object model . First, pairwise registration must be performed, but due to accumulated errors, the model may have serious "cracks" after pairwise registration. Based on the pairwise registration, this method accurately registers the entire model, eliminates the cumu...

AI Technical Summary

Problems solved by technology

Document 6-Pulli, K.Multiview Registration for LargeData Sets.In International Conference on 3D Digital Imaging and Modeling (3DIM), 1999. Use the results of pairwise registration to transform the reference point to the coordinate system of the corresponding depth image The position of the point in is used as its corresponding point, which is called "virtual" corresponding point in the paper. Only these points are used as corresponding points in the registration process of multiple depth images, and then use a method similar to R.Bergevin, and BenjemaaR. Because it saves the process of searching for corresponding points, it has high efficiency in time and space, but its corresponding points cannot update corresponding points in the registration process of multiple depth images, and it relies heavily on pairwise registration. result

Literature 7-Sharp, G., Lee, S., Wehe, D.Multiview Registration of 3D Scenes by Minimizing Error Between Coordinate Frames.IEEE Trans.PAMI, Vol.26, No.8, 2004. Pay special attention to the two The incompatibility of the two registration results, an adjustment method is proposed so that each loop can get a compatible pairwise registration result, but the disadvantage is that it does not fully consider the inconsistency of the different pairwise registration accuracy

[0005] In the various methods mentioned above, the search for corresponding points mostly searches for the nearest point or projected point in the whole world, and does not make full use of the results of pairwise registration, resulting in a large search range and affecting the calculation speed

Document 6 uses pairwise registration results to pre-calculate the corresponding points, which improves the calculation speed but has a certain impact on the accuracy.

The methods to minimize the error between corresponding points mainly include two types of methods, one-by-one optimization and simultaneous optimization. The one-by-one optimization method can only adjust the position of one depth image at a time, and the efficiency is low.

At the same time, the optimization method Document 5 can adjust the positions of all depth images at one time, so the efficiency is high. Document 5 uses the Ievenberg-Marquardt method to optimize the error at the same time, but it needs multiple iterations, and the calculation efficiency is still not high.

Images