Video face detection method and device

A face detection and face position technology, which is applied to instruments, computing, character and pattern recognition, etc., can solve the problems of poor face detection effect, slow algorithm training, and slow detection, and achieves accelerated video face detection, The effect of increasing speed

Active Publication Date: 2018-10-19
SHENZHEN INFINOVA
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, this kind of algorithm has problems such as slow training and slow detection. The slow detection lies in the need to use many levels of adaboost classifier. is a supervised learning model, usually used for pattern recognition, classification, and regression analysis), although it can improve the speed, it is not difficult to verify through the test of the same detection sample set that this method will lead to the change of the face detection effect in the final application bad question

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
  • Video face detection method and device
  • Video face detection method and device
  • Video face detection method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0058] Among the above, the classifier training process is also included; the classifier training process includes: training the adaboost classifier and the SVM classifier; wherein, the training of the SVM classifier can use the same features as those used in other environments. In the training process of the adaboost classifier, the number of layers can be reduced to 1 / 2 to 2 / 3 of the number of layers when it is used alone due to the adoption of the subsequent SVM classifier. As a result, the reduction in the level of the adaboost classifier can effectively increase the speed of video face detection, thereby increasing the detection speed, and because the SVM classifier is used for secondary classification on the detection results of adaboost, it can also ensure that the acceleration does not affect Detection effect.

Embodiment 2

[0060] Further, in the above classifier training process, the training of the SVM classifier adopts features that are consistent with the calculations used in other environments, including (1) grayscale, (2) color, and (3) one or more of edge information as training features.

Embodiment 3

[0062] like figure 2 , the face detection flow process S1 and S3 in the above-mentioned video face detection method also includes S2) input acceleration processing; the input acceleration processing includes steps:

[0063] S21) traverse to determine whether the color of each pixel of the original image in the YCbCr color space satisfies Cb∈[80,135] and Cr∈[136,177] at the same time, and obtain the first binary mask image MASK1;

[0064] This step is actually a judgment on the color of the pixel in the YCbCr color space, and Cb∈[80,135] and Cr∈[136,177] are the optimal ones determined after combining a large number of experiments, which are in line with the human skin color under the YCbCr color space Cb, Cr chroma. Of course, the threshold of CbCr here is determined according to the skin color experience value, so it will be affected by the imaging equipment, and it should be adjusted according to the actual situation if necessary.

[0065] S22) performing image edge detec...

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 present invention provides a video human face detection method and device, wherein the method includes a human face detection process; the human face detection process includes steps, S1) inputting a video image frame, using the image frame as an original image; S3) using haar or The lbp feature uses the adaboost classifier to detect the original image to obtain the pre-sequence of the face position; S4) input the pre-sequence of the face position into the SVM classifier for detection, and obtain the final sequence of the face position. In the present invention, two classifiers, adaboost and SVM, are organically combined and applied to video face detection, so that the levels of adaboost classifiers can be greatly reduced compared with those used alone, thereby effectively improving the speed of video face detection, and further improving the detection results of adaboost The SVM classifier is used for secondary classification to ensure the detection effect of fast detection.

Description

technical field [0001] The invention relates to a video image processing method, in particular to a video face detection method and device. Background technique [0002] In real-time video surveillance devices, there are real-time detection and collection of face images from different angles, and then uploaded to the server database to store key information and then perform application requirements for criminal suspect face recognition. [0003] However, traditional face detection mostly uses the method of haar / lbp feature plus adaboost classifier to realize real-time face detection on embedded devices such as cameras. The above-mentioned adaboost is an iterative algorithm, and its core idea is to train different classifiers for the same training set, and then combine these weak classifiers to form a stronger final classifier. Using the adaboost classifier can exclude some unnecessary training data features and put them on the key training data. [0004] However, this kind...

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): G06K9/00G06K9/62
CPCG06V40/172G06V20/41G06F18/2411G06F18/214
Inventor 李杨莫平华刘军
Owner SHENZHEN INFINOVA
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