Face library generation method and device for monitoring field and electronic equipment

Abstract

The invention relates to the technical field of computers, in particular to a face library generation method and device for the monitoring field and electronic equipment, and the method comprises the steps of obtaining a video stream monitored by a camera; performing face angle judgment on a first face image in the video stream; performing primary screening on the first face image according to a preset inclination angle to obtain a second face image; performing separation processing on the preset inclination angle, and performing secondary screening on the second face image according to the discretized preset inclination angle; and carrying out statistics on the second face image after secondary screening, wherein the second face image after secondary screening is used for carrying out face recognition on a person in the monitoring area. According to the invention, high-quality screening can be carried out on the pictures, repeated and unqualified pictures are filtered, manual focusing is not needed, and high-quality input of the face database is realized through monitoring.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method, device and electronic equipment for generating a face library used in the monitoring field. Background technique [0002] At present, the composition of the face database is composed of one or more photos of each person. If it is composed of a single face of each person, then this face is a frontal photo; and a face library with multiple faces will be able to include photos from more angles. The performance of the face database with multiple faces is better than that of a single face, because more face information will be entered, so people are more inclined to enter multiple faces on a single person. If this is the case, these pictures will be photos of faces from different angles or even different scenes. These photos are obtained by taking pictures separately or jumping a fixed number of frames in a video to capture pictures. However, if these different ...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, the face position and three-dimensional face angle information of the images in the video stream are obtained according to face tracking and detection. For face photos, a total of nine templates with different angles are stored for face recognition. However, when recognizing face angles, there is no mention of correcting the error of the camera’s overlooking angle in order to obtain more accurate face angle images. At the same time, the field of use is also limited to the field of face recognition, and there is no further limitation on the viewing angle of the camera.

Images