Face detection result set optimization system and face detection result set optimization method

Abstract

The invention provides a face detection result set optimization system and a face detection result set optimization method. The system and the method are characterized in that a video stream is extracted from each set of image photographing equipment; through early-period processing of a tracking detecting module, a characteristic analyzing module and an optimization processing module, a high-quality face queue which comprises an optimal face data set is obtained; and then, the high-quality face queue is transmitted to a server through a face uploading module by means of network communication, thereby realizing subsequent face comparison processing. Compared with prior art in which an original human image which is directly acquired through monitoring software is utilized for subsequent face comparison processing, the high-quality face queue supplied by the invention has advantages of reducing face data volume and greatly improving quality. Therefore, the face detection result set optimization system and the face detection result set optimization method have advantages of greatly reducing network flow uploaded through network, preventing error warning in a large number of comparisons, greatly improving comparison effect and reducing CPU resource consumption.

Description

technical field [0001] The present invention relates to the technical field of video image calculation, in particular to an optimal method and system for a human face detection result set. Background technique [0002] Usually, the original portraits collected by video surveillance software are obtained by decoding the real-time streaming protocol (Real Time Streaming Protocol; RTSP) network video stream, and then upload the collected original portraits through the network without processing and provide them to The background application does the comparison. [0003] While the existing network high-definition camera captures video at a speed of 25 frames per second, as long as the monitoring target stays in the collection area for 1 second, there will be 25 faces. If the target stays for 1 minute, there will be 1,500 faces. Face calculation will generate a large amount of data to be compared and recognized. The deflection angles of these massive faces are large and small, a...

AI Technical Summary

Problems solved by technology

[0003] While the existing network high-definition camera captures video at a speed of 25 frames per second, as long as the monitoring target stays in the collection area for 1 second, there will be 25 faces. If the target stays for 1 minute, there will be 1,500 faces. Face calculation will generate a large amount of data to be compared and recognized. The deflection angles of these massive faces are large and small, and the quality of the photos is also uneven. All of them are used for recognition and comparison, and the CPU resource consumption is very large. Moreover, if the above-mentioned massive data is uploaded to the background for comparison and use through the network, it will also consume a lot of network traffic; in addition, the photos collected by the Rtsp camera will be affected by various factors such as posture, light, and weather. The quality of the photos is poor. If all the collected portraits are used for comparison, there will be a large number of false alarms

Images