A facial feature point tracking method and system

Abstract

The present invention provides a facial feature point tracking method, comprising the following steps: step A: initialize parameters; step B: receive the first image; step C: if the first image includes a facial image, generate a second image based on the first image A facial image; step D: generate first feature points based on the first facial image; step E: generate first prediction information based on the first feature points; step F: receive the second image; step G: based on the second image and The first prediction information generates a third image; step H: if the third image includes a second facial image corresponding to the facial image, generate a second feature point based on the second facial image, use the second feature point as the first feature point and Go to step E. By predicting the second feature point at the next moment through the first feature point at the previous moment, the calculation amount of the facial feature point tracking technology is reduced, the calculation difficulty is reduced, and the real-time tracking of the facial feature point is realized.

Description

technical field [0001] The present invention relates to the technical field of image processing, and more specifically, to a facial feature point tracking method and system. Background technique [0002] With the development of technology in the field of computer vision, the technology for locating facial feature points is becoming more and more perfect. However, the existing excellent feature point locating algorithms are mostly aimed at static pictures, and are rarely used in dynamic images such as video files. Feature points refer to specific points such as the tip of the nose, the corners of the mouth, and the corners of the eyes. [0003] In the prior art, the feature point tracking technology of dynamic images mainly uses repetitive calculation tracking, that is, each frame detects and locates feature points on the face of the collected image. This method requires a large amount of calculation and is difficult to achieve real-time tracking. [0004] Therefore, how to ...

AI Technical Summary

Problems solved by technology

[0003] In the existing technology, the feature point tracking technology of dynamic images mainly uses repeated calculation tracking, that is, each frame detects and locates feature points on the face of the collected image. This method requires a large amount of calculation and is difficult to achieve real-time tracking.

Images