摘要
由于火焰在视频图像中的运动表现形式不同于一般物体的目标运动,具有一定的特殊性。为了在视频图像中复杂背景下更加准确检测出火焰信息,本文在火焰颜色特征基础上提出了基于火焰运动区域特征的火焰检测算法。该算法通过提取火焰运动轨迹曲率的极值点并将其相对位置表示出来作为火焰运动区域的特征向量,从而获得运动火焰轨迹的几何特征。仿真实验结果表明,此方法在进行火焰相似性配比查询时,降低了图像的维度,在检测中能够保持较高的火焰检测率和较低火焰误检率,具有良好的检测效果。
The motion form of flame is particular and different from the motion form of general object. In order to detect the flame information more accurately in the complex background of video images,a flame detection algorithm based on the characteristics of the flame motion region is proposed,which is derived from the color characteristics of the flame. In this algorithm,the extreme points of the curvature of the flame trajectory are extracted,and the relative positions are represented as the eigenvectors of the flame motion area,then the geometric characteristics of the flame trajectory are acquired. The simulation results show when the flame similarity ratio query using this method is processed,the dimension of the image is reduced,and flame detection rate is maintained the higher level,meanwhile flame false detection rate is maintained the lower level. Therefore,good detection effect is reflected.
作者
耿庆田
张晶
赵宏伟
王闯
GENG Qing - tian;ZHANG Jing;ZHAO Hong - wei;WANG Chuang(Department of Computer Science and Technology, Changchun Normal University, Changchun Jilin 130032,China;Department of Computer Science and Technology, Jilin University,Changchun Jilin 130012, China;Comprehensive Practice Education Base School of Kuancheng District of Changchun,Changchun Jilin 130052, China)
出处
《长春师范大学学报》
2018年第2期35-38,共4页
Journal of Changchun Normal University
基金
吉林省省级产业创新专项资金项目“基于大数据的人类在通信网络上的行为动力学研究”(2016C078)
吉林省产业技术研究和开发专项项目“基于物联网技术的智能仓储平台开发与应用”(2017C031-2)
吉林省教育厅“十三五”科学技术研究项目“基于人工智能的汽车自适应巡航控制技术研究”(2018269)
关键词
计算机应用
图像识别
运动检测
火焰识别
computer application
image recognition
motion detection
flame recognition