基于红外视频的高实时性动目标检测识别算法研究与实现  

Study and Implementation of a Real-time Moving Target Detection and Recognition Algorithm Based on Infrared Video Stream

作　　者：戴玉成 温明[1] 武亚丽 袁滔 宋秋迪 DAI Yucheng;WEN Ming;WU Yali;YUAN Tao;SONG Qiudi(The 3th Research Institute of China Electronics Technology Group Corporation,Beijing 100015,China)

机构地区：[1]中国电子科技集团公司第三研究所,北京100015

出　　处：《电视技术》2020年第5期46-50,共5页Video Engineering

摘　　要：针对红外视频流中的远距离行人和车辆动目标自动检测和识别问题,提出了一种动目标自动检测和识别算法。首先,以无参数ViBe算法进行背景建模和运动前景检测;其次,利用连通区域识别算法分割和定位动目标对象区域,并解算对象的先验参数;最后,采用Softmax分类器识别目标类型。实验表明,该算法在测试集上的mAP达97%,能够准确有效地检测出红外视频流中的移动目标。此外,该算法在500 MHz低功耗的DSP上以27 fps运行,满足实时性要求,具有较高的应用价值。For the automatic detection and recognition of long-distance pedestrians and vehicles moving objects in infrared video streams,in this paper,we present an effective detection and recognition algorithm for moving objects.Our approach is to model the background and detect the moving foreground by using nonparametric ViBe algorithm firstly.Then,connected region recognition algorithm is employed to segment and locate the moving object proposal region and calculate a priori parameters of the object;finally,softmax classifier is used to predict the type of object.Evaluation on test set shows that the mAP of proposed algorithm reaches 97%,which can detect moving objects accurately and effectively in infrared video stream.In addition,the algorithm runs at 27fps on a 500MHz low-power DSP,which meets the real-time requirements and has high application value.

关 键 词：红外视频流 动目标检测 ViBe 线段编码 Softmax分类器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]