摘要
对于婴幼儿与具有语言表达障碍的患者无法正常描述其疼痛感受的情况,可以采用通过视频采集设备,将图像交予计算机进行疼痛评估的方法。本文介绍在计算机视觉技术与人脸检测技术下,通过K近邻(KNN)算法,对特定人大量已标记的疼痛表情图片进行训练,获得疼痛表情识别模型,应用于视频流或图像中的疼痛表情的检测并进行0-4级分类。实验结果表明,训练得到的模型准确率可达93.2%,识别效率约为15帧每秒。该方法适用于患者医疗或康复过程的场景之中,具有一定的社会价值。
For infants and patients with language disorders who can not normally describe their pain feelings,through video capture equipment,images are sent to computers for pain assessment.This paper will introduce the computer vision technology and face detection technology.Using the k-Nearest Neighbor(KNN)algorithm,a large number of marked pain expression images of a specific person are trained to obtain the pain expression recognition model,and the trained model is then applied to the detection of pain expression in video stream or image and 0-4 level classification.The experimental results show that the accuracy of the trained model can reach 93.2%,and the recognition efficiency is about 15 frames per second.The model is suitable for patients in the scene of medical treatment or rehabilitation process,and has certain social value.
作者
王浩然
王彦博
张剑书
郑胜男
WANG Haoran;WANG Yanbo;ZHANG Jianshu;ZHENG Shengnan(School of Computer Engineering,Nanjing Institute of Technology,Nanjing,China,211167)
出处
《福建电脑》
2021年第8期36-38,共3页
Journal of Fujian Computer
基金
江苏省高等学校大学生创新创业训练计划项目(No.202011276050Y)
南京工程学院校级科研基金项目(No.QKJ201803)资助。
关键词
计算机视觉
人脸检测
K近邻
疼痛表情识别
Computer Vision
Face Detection
K-Nearest Neighbor
Pain Expression Recognition
作者简介
通信作者:王浩然,男,2000年生,主要研究领域为图像处理。E-mail:whrmailbox2020@163.com;王彦博,男,1999年生,主要研究领域为数据分析;张剑书,男,1992年生,主要研究方向为视频图像处理;郑胜男,女,1986年生,主要研究方向为数字图像处理。
