Head posture estimation method based on multi-feature-point set active shape model (ASM)

Abstract

The invention relates to a head posture estimation method based on a multi-feature-point set active shape model (ASM). The method includes the steps that firstly, face samples are trained to obtain the global ASM and a local texture model; secondarily, according to the models obtained through training, face feature point fitting is performed on an obtained face image sequence, feature point coordinates are stored, and reference coordinates are updated periodically; then, the displacement of all feature points is calculated, and the number of the feature points exceeding a displacement threshold value is counted; finally, according to the counted number of the feature points and the displacement direction, head postures are estimated. Influences, caused by inaccurate positioning on a small number of feature points, on head posture estimation can be reduced, meanwhile, the head posture estimation method has a high robustness effect on illumination, various head postures, such as the front face, left turning, right turning, head raising and head lowering, can be estimated, and the head posture estimation method has great application prospects in the fields of intelligent video monitoring, virtual reality, mode recognition, man-machine interaction and others.

Description

technical field [0001] The invention belongs to the fields of computer pattern recognition, computer vision and human-computer interaction, and relates to a method for estimating the head posture of a face image, in particular to a method for estimating the head posture based on a multi-feature point set active shape model. Background technique [0002] Head pose contains rich human emotional information, which can well express the true thoughts deep in the human heart. Therefore, head pose estimation has great research and application value in the field of computer vision. It is an important link in the research of intelligent video surveillance, virtual reality, pattern recognition and human-computer interaction. The result of the pose estimation will directly affect the final System stability. [0003] At present, it is still very difficult to make computers have the same recognition ability as humans. In machine vision, head pose estimation needs to be able to detect a...

AI Technical Summary

Problems solved by technology

[0004] However, in these processes, due to the ever-changing head posture, camera distortion, lighting and other environmental factors, as well as the character's appearance and expression, and factors such as hats and glasses occlusion, the positioning of feature points is often not stable enough. Affects the accuracy of the final head pose estimation

Because of this, head pose estimation has always been a very challenging subject

Images