Application of Multi-target Pose Estimation in Wearable Monitoring System

摘要

The traditional examination monitoring has some disadvantages: monitor dead, blur distance monitoring pictures, monitoring environment restrictions and so on. This article researches wearable monitoring method. The method supplies multi-objective, multi-angle and close supervision, and breaks some technical barriers such as room space constrain, light interference. This article mentioned an integral projection algorithm basing on narrowing region, and this algorithm can focus face characters from video picture sequence, so it supplies precision guarantee for face pose estimate. Its basic thought is that establish the mapping relationship between position parameters and face three-dimensional pose by using the positions of face oval models and feature points matrixes in images, then apply the face pose in intelligent video examination monitoring system according to the gesture parameters obtained from inputted face images, and the experiment showed the system can run well.