Three-dimensional face point cloud nose tip detection method and data processing device using the same

Abstract

The invention discloses a three-dimensional face point cloud nose tip detection method and a data processing device using the same. Super voxels are utilized to perform efficient segmentation on three-dimensional point cloud data, a super voxel neighborhood is utilized to enhance a matching capability between face partial structures, and relative geometric transformation between a face posture and a training sample face posture is estimated and tested in combination with a partial reference frame, thereby correcting a generalized Hoff voting process of a three-dimensional face nose tip position, and improving face nose tip position estimation accuracy and reliability of a system.

Description

technical field [0001] The invention relates to the field of three-dimensional point cloud data processing, in particular to a method for detecting a nose tip in a three-dimensional human face point cloud and a data processing device using the same. Background technique [0002] With the maturity of 3D imaging technologies such as structured light, stereo vision and laser scanning, real-time 3D data acquisition has become a reality. More and more face recognition researchers are turning their attention to the field of 3D face recognition technology. 3D face recognition technology is expected to fundamentally solve the problems of "illumination, posture and expression" faced by 2D face recognition. [0003] 3D face detection is the basis of 3D face recognition technology. Point clouds collected by sensors or stereo vision often contain a large amount of non-face point cloud data. These areas are useless information for 3D face recognition, which will affect the later face ...

AI Technical Summary

Problems solved by technology

Other 3D face detection technologies either only rely on the geometric features at the contour of the tip of the nose, or combine two-dimensional texture information to locate the tip of the nose, and the positioning accuracy is relatively low

In the existing technology, the nose tip detection technology that only relies on the geometric features of the nose tip contour does not make full use of the spatial relationship between other structures of the face, such as eyes, mouth, etc., and the nose tip

However, the combination of 2D texture information technology will fail when only 3D point cloud data can be obtained.

Images