Face pose estimation method, system and device based on feature matching and storage medium

A face posture and feature matching technology, applied in the field of image processing, can solve the problems of consuming hardware computing power and inaccurate estimation, and achieve the effect of short time consumption, simple implementation and high precision

Pending Publication Date: 2022-01-07
HANGZHOU XINHE SHENGSHI TECH
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] At present, the commonly used method is to estimate the pose based on the face key point algorithm. From the perspective of technical implementation, it is already very mature, but if it involves factors such as large poses and unsatisfactory lighting, it will affect the positioning accuracy of the face key point algorithm. , this error will be amplified in the process of calculating the face pose in the second stage, resulting in inaccurate estimation; in addition, if it is a high-precision face key point positioning algorithm, it will take a certain amount of time and consume hardware computing power

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Face pose estimation method, system and device based on feature matching and storage medium
  • Face pose estimation method, system and device based on feature matching and storage medium
  • Face pose estimation method, system and device based on feature matching and storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] A face pose estimation method based on feature matching, comprising the following steps:

[0042] S100. Acquire a face image containing a pose to be estimated, and perform face detection and cropping preprocessing on the face image to obtain a preprocessed image;

[0043] S200. Input the preprocessed image to a preset feature extraction network model to obtain contour feature information;

[0044] S300. Match the contour feature information with the preset face template to obtain the best matching degree template, the pose information of the best matching degree template is the input face image pose information, wherein the face template is Multi-pose face templates, each face template contains information of three rotation directions.

specific Embodiment approach

[0046] a) Prepare the network for multi-person multi-pose face template and feature extraction

[0047] The most important part of template matching is the richness of the template, and the template uses the BIWI head pose dataset. The BIWI data set contains 24 videos and about 15678 frames of face images. The acquisition process is that the collectors sit in front of the sensor and turn their heads freely. This method is used to collect head data of different people with different postures. The data set is usually As a benchmark for head pose estimation using deep methods, each face image provides information on three rotation directions.

[0048] Euler angles are a set of independent angular parameters used to determine the rotational position of a rigid body, which are pitch angle (pitch), yaw angle (yaw), and roll angle (roll). Here we take the python operation as an example to convert the information W provided by the BIWI dataset into Euler angles.

[0049] Specifically:...

Embodiment 2

[0075] A face pose estimation system based on feature matching, including a data acquisition module, a contour acquisition module and a matching comparison module;

[0076] The data acquisition module is used to acquire a face image containing a posture to be estimated, and perform face detection and cropping preprocessing on the face image to obtain a preprocessed image;

[0077] The contour acquisition module is used to input the preprocessed image to the preset feature extraction network model to obtain contour feature information;

[0078] The matching and comparing module is used to match the contour feature information with the preset face template to obtain the best matching degree template, and the pose information of the best matching degree template is the input face image pose information, wherein, The face templates are multi-pose face templates, and each face template contains information of three rotation directions.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a face pose estimation method based on feature matching, and the method comprises the following steps: obtaining a face image containing a to-be-estimated pose, and carrying out the face detection and cutting preprocessing of the face image, and obtaining a preprocessed image; inputting the preprocessed image into a preset feature extraction network model to obtain contour feature information; matching the contour feature information with a preset face template, and obtaining an optimal matching degree template, wherein posture information of the optimal matching degree template is input face image posture information, the face template is a multi-posture face template, and each face template comprises information of three rotation directions. According to the method, system and device, the pre-trained face feature point positioning model is used, only the face contour features of the middle layer are used, the whole set of system is realized based on python, the method comprises the steps of face detection algorithm calling, contour feature extraction, image alignment and feature matching, and the method is simple and easy to implement and high in starting speed.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a method, system, device and storage medium for face pose estimation based on feature matching. Background technique [0002] In the existing field of facial pattern recognition technology, face pose estimation refers to determining the rotation angle of the face in different dimensions in three-dimensional space through continuous video or a single picture, and usually uses two-dimensional images to solve the head pose. For classification problems, facial pose estimation has always been a challenging research project due to the reduction of one-dimensional information. With the application of intelligent algorithms, more and more people study face detection, face feature point location, face recognition, face attribute classification, face pose classification, etc., but no matter what kind of algorithm, high-quality human Face images will provide performance gu...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G06T7/00
CPCG06T7/0002G06T2207/20132G06T2207/30201
Inventor 吴彰鹏杨旭旨
Owner HANGZHOU XINHE SHENGSHI TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap