Three-dimensional human body posture estimation method based on monocular camera

Abstract

The invention belongs to the technical field of computer vision, and discloses a three-dimensional human body posture estimation method based on a monocular camera. The method comprises the followingsteps that 1, a three-dimensional human body model is built on the basis of standard proportion data of a normal human body, and a two-dimensional projection of the three-dimensional human body modelis acquired on the basis of a weak perspective projection method; 2, a video of the human body in motion is acquired by means of the monocular camera, and the two-dimensional contour feature of each human body image frame of the video is extracted; and 3, the two-dimensional contour feature of each human body image frame is compared with the two-dimensional projection of the three-dimensional human body model, cost functions are established, and an optimal solution of the cost functions is calculated, so that the most matched three-dimensional model is obtained. According to the method, the defect of the high requirement of a general generative model method on model generation is overcome, and the method can be widely applied to posture and size estimation on human body objects in different postures and different shapes and has the advantages of being high in adaptability and robustness.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a method for estimating a three-dimensional posture of a human body based on a monocular camera. Background technique [0002] Vision-based human pose estimation has gradually become a research hotspot in the field of computer vision. Accurate pose estimation can provide better data basis and guidance for action recognition, behavior analysis, and behavior understanding. [0003] Today, there are two main categories of vision-based methods for 3D human pose estimation, including discriminative models and generative models. The discriminative model requires a large amount of training data to obtain an accurate pose estimation structure, but processing a large amount of data will reduce the efficiency of the algorithm, and when the real image is quite different from the training sample, the algorithm effect of the discriminative model is very unsatisfactory . ...

AI Technical Summary

Problems solved by technology

[0006] The invention provides a method for estimating the three-dimensional pose of a human body based on a monocular camera, which solves the problems that the existing methods rely too much on the pre-established three-dimensional human body model, have low computing efficiency, and poor robustness.

Images