A bearded face detection method for surveillance video

Abstract

The invention relates to a monitoring video-oriented face detection method with beards. This method is mainly for monitoring video processing. First, the computer reads the video file, decodes it and uses the improved background difference method combined with the upper body HOG classifier to obtain the area where there are moving targets in the upper body of the human body; In one part, the face Haar classifier is used to detect the possible face area, and the lower part is intercepted as the beard detection area; finally, the horizontal texture complexity is calculated after normalizing the scale of the area, and the judgment Whether there is a beard. The method utilizes the motion and morphological features of the bearded faces in the surveillance video, and can reliably detect the bearded faces in walking and running upright people.

Description

technical field [0001] The invention belongs to the field of video image processing for public security early warning, and in particular relates to a bearded face detection method. Background technique [0002] Video surveillance is widely used in the field of public security, providing strong data and technical support for early warning and verification in public security management. Bearded face detection is mainly used to quickly screen out bearded personnel targets from massive surveillance videos, which can enhance the efficiency of public security departments in screening and searching for specific targets, and is useful for preventing and cracking down on crimes, tracking down suspects, and maintaining social security. important role. However, the level of intelligent analysis for surveillance video is relatively low at present, and there is no effective technical means for the application of beard face detection for surveillance video. Bearded face detection for su...

AI Technical Summary

Problems solved by technology

[0003] For the various links in beard face detection, in terms of moving target detection, existing methods such as patent 201410110812.1 use the ViBe algorithm to establish a background model for the video image frame, and combine the frame difference method to segment the foreground area. This method consumes a lot of time for background update , resulting in slower overall processing speed; patent 201110253323.8 uses edge detection and frame difference method for motion detection, and patent 201310586151.5 combines adjacent frame difference method and mixed Gaussian model to realize moving targets. The disadvantage of the above method is that it is prone to cavity areas; in the human body In terms of detection, patent 201010218630.8 uses fuzzy templates to detect multi-pose human bodies, and the speed is slow; patent 201310415544.X realizes human body detection based on color and depth information, but the color information in surveillance videos is greatly affected by light; patent 201110026465.0 is based on depth images Human body detection is not suitable for conventional surveillance video images; in terms of face texture detection, patent 201210351040.1 uses the Gabor filter to extract the texture features of the face, but only extracts the overall contour of the face and cannot be used to represent the beard features; to a published patent on facial beard detection

Images