Face recognition analysis method used in monitoring video data

• Main Authors: GAO JIAN, JIANG WENJUN, LIANG GUANGLIN, FAN ZUOJUN
• Format: Patent
• Language: Chinese; English
• Published: 20.12.2019
• Subjects: CALCULATING; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a face recognition analysis method used in monitoring video data, which comprises the following steps of comparing and scoring the face high-latitude face features by using twoneural networks, and fusing a K adjacent algorithm and a screening mechanism taking the Euclidean distance as a threshold value to realize the face comparison and recognition so as to obtain an accurate face recognition result; further screening according to the time of the video data and the occurrence frequency of the face to obtain a recognition result finally, and further obtaining the more accurate face recognition result. According to the invention, the more accurate face recognition can be carried out on the video data or picture data shot by a real-time or historical monitoring camera. 本发明公开一种监控视频数据中使用的人脸识别分析方法，利用2个神经网络对人脸高纬人脸特征进行比对打分，融合K相邻算法以及欧式距离为阙值的筛选机制做到人脸比对识别，从而得到准确的人脸识别结果；此外，本发明还根据视频数据的时间和人脸的出现频次做进一步的筛选，最终得到识别结果，进而得到更为准确的人脸识别结果。本发明可以对实时或历史监控摄像头拍摄到的视频数据或图片数据进行比较准确的人脸识别。