An image processing method and device thereof

An image processing and image technology, applied in the field of image processing, can solve the problems of unsuitable real-time calculation, huge amount of calculation, large difference in results, etc., and achieve the effect of improving the recognition pass rate and enriching the recognition methods.

Active Publication Date: 2020-04-28
腾讯征信有限公司
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Through the above process, it can be seen that the existing estimation methods either have a large amount of calculation and are not suitable for real-time calculation; or they rely too much on feature operators, resulting in large differences in the results obtained from different training sets.

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
  • An image processing method and device thereof
  • An image processing method and device thereof
  • An image processing method and device thereof

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0033] This embodiment provides an image processing method, and the method is applied to an image processing device; specifically, as figure 1 As shown, the method includes:

[0034] Step 101: Determine an image set; the image set includes a first image corresponding to each head pose of at least two head poses of the first object;

[0035] specifically, figure 2 It is a schematic diagram of the first images of different head postures of the first target contained in the image set according to the embodiment of the present invention; figure 2 As shown, the image set contains four head-oriented first images for the first target object, which are the first image-1 with the head facing up, the first image-2 with the head facing down, and the first image-2 with the head facing down. The first image-3 with the head facing left and the first image-4 with the head facing right. Here, in practical applications, the elements in the image set are represented by the vectors of the f...

Embodiment 2

[0049] Based on the method described in Embodiment 1, this embodiment provides a method for determining pose matching model groups and head recognition model groups based on correlation filter technology; here, the correlation filter technology is derived from correlation operations, so The correlation operation described above is an operation to calculate the similarity between two models; for example, suppose there are two pictures R and T, and they are defined as two modes in two dimensions, namely R(x,y) and T (x,y), the correlation operation between two modes is: (T*R)(x,y)=∫∫T(x,y)R(x-Γ x ,y-Γ y )dΓ x dΓ y ; Convert the two modes to the frequency domain and use R(x, y) as a filter, then the correlation operation can be considered as filtering T(x, y) with the R(x, y) filter, that is, the correlation filter. Due to the advantages of time-shift invariance, good degeneration and closed-form solution, the correlation filter has been effectively applied in face recognitio...

Embodiment 3

[0054] Based on the method described in Embodiment 1, this embodiment also provides a method for determining a pose matching model group and a head recognition model group based on correlation filter technology. That is, according to the HOG features and Gabor features under the same head pose of different objects (taking the first object and the second object as an example), the pose matching model group and the head recognition model group are respectively determined.

[0055] Specifically, acquire the posture features corresponding to each of the at least two head postures corresponding to the second target body, and the local features of the head details under the head posture; according to the corresponding head postures for each head posture The posture features of the first target body and the second target body (this posture feature can be determined based on correlation filter technology), determine the posture matching model for identifying the head posture of the det...

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 embodiment of the present invention discloses an image processing method, including: determining an image set; extracting the posture features of the head posture corresponding to each first image in the image collection, and extracting the posture features of the head posture corresponding to each first image; The local features of the head details; according to the pose features of the first image in the image set, determine the pose matching model group for recognizing the head pose of the detection object in the image to be detected, and according to the local first image in the image set A head recognition model group for identifying the head details under the head posture of the detection object in the image to be detected is determined based on the feature; wherein, the pose matching model group includes a model group for the first object Pose models of at least two head poses; the head recognition model group includes head models of head details for at least two head poses of the first object. The embodiment of the invention also discloses an image processing device.

Description

technical field [0001] The present invention relates to image processing technology, in particular to an image processing method and device thereof. Background technique [0002] The current head orientation estimation methods are roughly divided into two categories: the first category, model-based methods; the second category, appearance-based methods. Among them, the model-based head orientation estimation method generally judges the head orientation by reconstructing the 3D model of the head. This type of method is relatively accurate, but requires a huge amount of calculation, so it is not suitable for real-time calculation; while the appearance-based There are two methods for head orientation estimation: one is to estimate the head orientation based on the robustness of feature representation; the other is to estimate the head orientation based on the feature points of the face and head; here, based on the robustness of feature representation The head orientation estim...

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 Patents(China)
IPC IPC(8): G06K9/00G06K9/46
CPCG06V40/10G06V10/50G06V10/462
Inventor 晏栋
Owner 腾讯征信有限公司
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