基于多算法融合的移动相机视频识别技术研究
|
摘要
传统移动相机视频识别技术存在识别准确度较低,易受外界干扰的问题,为了提高苏通综合管廊工程特高压GIL设备SF6密度继电器视频图像的自动识别和监测能力,采用移动相机视频识别技术进行仪表识别,提出基于去模糊算法和SURF算法多算法融合的移动相机视频识别技术。采用模糊视频特征分析方法进行SF6密度继电器的视频采集和帧重组,进行SF6密度继电器的边缘图像特征提取,将采集的移动相机视频进行多传感信息融合,结合SF6密度继电器的边缘轮廓特征信息进行多算法的融合处理,采用去模糊算法进行特高压GIL设备SF6密度继电器的模糊去噪,降低视频识别中的模糊度,采用SURF算法进行边缘角点检测,根据角点检测结果进行移动相机视频信息的状态点标注,实现移动相机视频识别。仿真结果表明,在视频码率为120的时候,本文方法的准确率就已达到45%,采用该方法进行特高压GIL设备SF6密度继电器移动相机视频识别的准确度较高,抗扰动能力较强。
The traditional video recognition technology of mobile camera has the problems of low recognition accuracy and easy to be interfered by the outside world. In order to improve the automatic recognition and monitoring ability of the SF6 density relay video image of the ultra-high voltage GIL equipment of Sutong integrated pipe gallery project.This paper presents a new video recognition technique for mobile camera based on the combination of de-blur algorithm and SURF algorithm,which is based on the video recognition technology of mobile camera. The method of fuzzy video feature analysis is used to collect video and reorganize the frame of SF6 density relay,the edge image feature of SF6 density relay is extracted,and the collected video of mobile camera is fused with multi-sensor information. The edge profile feature information of SF6 density relay is fused with multiple algorithms,and the fuzzy denoising of SF6 density relay of UHV GIL equipment is carried out by using de-fuzzy algorithm to reduce the ambiguity in video recognition.The edge corner is detected by SURF algorithm,and the status point of mobile camera video information is annotated according to the result of corner detection,and the video recognition of mobile camera is realized. The simulation results show that when the video code rate is 120,the accuracy of this method has reached 455. The accuracy of using this method for video recognition of SF6 density relay mobile camera in UHV GIL equipment is high. The disturbance ability is strong.
The traditional video recognition technology of mobile camera has the problems of low recognition accuracy and easy to be interfered by the outside world. In order to improve the automatic recognition and monitoring ability of the SF6 density relay video image of the ultra-high voltage GIL equipment of Sutong integrated pipe gallery project.This paper presents a new video recognition technique for mobile camera based on the combination of de-blur algorithm and SURF algorithm,which is based on the video recognition technology of mobile camera. The method of fuzzy video feature analysis is used to collect video and reorganize the frame of SF6 density relay,the edge image feature of SF6 density relay is extracted,and the collected video of mobile camera is fused with multi-sensor information. The edge profile feature information of SF6 density relay is fused with multiple algorithms,and the fuzzy denoising of SF6 density relay of UHV GIL equipment is carried out by using de-fuzzy algorithm to reduce the ambiguity in video recognition.The edge corner is detected by SURF algorithm,and the status point of mobile camera video information is annotated according to the result of corner detection,and the video recognition of mobile camera is realized. The simulation results show that when the video code rate is 120,the accuracy of this method has reached 455. The accuracy of using this method for video recognition of SF6 density relay mobile camera in UHV GIL equipment is high. The disturbance ability is strong.
引文
[1]朱映映,文振焜,杜以华,等.基于视频感知哈希的视频篡改检测与多粒度定位[J].中国图象图形学报,2013,18(8):924-932.
[2]许琮,宋立锋.可分级视频编码差错控制技术综述[J].电视技术,2014,38(1):6-10.
[3]申小霞,张桦,高赞.基于Kinect和金字塔特征的行为识别算法[J].光电子·激光,2014,(2):357-363.
[4]田国会,尹建芹,韩旭,等.一种基于关节点信息的人体行为识别新方法[J].机器人,2014,36(3):285-292.
[5]李松,谢新新,刘东林,等.基于Dirichlet过程的无线视频码率变化识别算法[J].高技术通讯,2016,26(10):833-840.
[6]陈利峰.舞蹈视频图像中人体动作识别技术的研究[J].现代电子技术,2017,40(3):51-53.
[7] AMOR B B,SU J,SRIVASTAVA A. Action recognition using rate-invariant analysis,of skeletal shape trajectories[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence,2016,38(1):1-13.
[8]李松林,贾勇,郭勇,等.基于改进Camshift的穿墙雷达运动人体目标成像跟踪算法[J].计算机应用,2018,38(2):528-532.
[9] TAN Q Y,LEUNG H,SONG Y,et al. Multipath ghost suppression for through-the-wall-radar[J]. IEEE Transactions on Aerospace and Electronic Systems,2014,50(3):2284-2292.
[10] GENNARELLI G,SOLDOVIERI F. Multipath ghosts in radar imaging:physical insight and mitigation strategies[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2014,8(3):1078-1086.
[11]程振宇,张灿,和智涛,等.基于3G网络视频传输的一种Qo S控制方法[J].中国科学院大学学报,2014,31(1):117-123.
[12]王军,陈翠琴.基于RFID信息与视频图像的人员识别的研究[J].物联网技术,2015,5(3):30-31.
[13]焦义文,王元钦,马宏,等.群时延失真对天线组阵合成信噪比的影响分析[J].信号处理2015,31(2)145-153.
[14]周亚兰.基于FPGA的多通道数据解码技术研究[J].物联网技术,2015,5(3):32-34.
[15]宋晓丽,刘冀伟,张晓星.分布式视频编码的关键帧提取算法[J].智能系统学报,2011,6(6):539-543.
[16]宋彬,贺红,刘海华等. Wyner-Ziv视频编码中边信息生成算法研究[J].通信学报,2010,31(12):97-103.
[17] R. Puri,K. Ramchandran. PRISM:an uplink-friendly multimedia coding paradigm:Proceedings of 2003 IEEE international conference on acoustics,speech,and Signal Processing[C]. Hong Kong,2003:856-859.
[18]孙一南,韩晓明.基于多算法融合的运动目标定位研究[J].微电子学与计算机,2017,34(8):123-127.
[19]孙丰,秦开怀,孙伟,等.一种针对移动相机的实时视频背景减除算法[J].计算机辅助设计与图形学学报,2016,28(4):573-579.
[20]罗会兰,赖泽云,孔繁胜.动作切分和流形度量学习的视频动作识别[J].中国图象图形学报,2017,22(8):1106-1119.
[2]许琮,宋立锋.可分级视频编码差错控制技术综述[J].电视技术,2014,38(1):6-10.
[3]申小霞,张桦,高赞.基于Kinect和金字塔特征的行为识别算法[J].光电子·激光,2014,(2):357-363.
[4]田国会,尹建芹,韩旭,等.一种基于关节点信息的人体行为识别新方法[J].机器人,2014,36(3):285-292.
[5]李松,谢新新,刘东林,等.基于Dirichlet过程的无线视频码率变化识别算法[J].高技术通讯,2016,26(10):833-840.
[6]陈利峰.舞蹈视频图像中人体动作识别技术的研究[J].现代电子技术,2017,40(3):51-53.
[7] AMOR B B,SU J,SRIVASTAVA A. Action recognition using rate-invariant analysis,of skeletal shape trajectories[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence,2016,38(1):1-13.
[8]李松林,贾勇,郭勇,等.基于改进Camshift的穿墙雷达运动人体目标成像跟踪算法[J].计算机应用,2018,38(2):528-532.
[9] TAN Q Y,LEUNG H,SONG Y,et al. Multipath ghost suppression for through-the-wall-radar[J]. IEEE Transactions on Aerospace and Electronic Systems,2014,50(3):2284-2292.
[10] GENNARELLI G,SOLDOVIERI F. Multipath ghosts in radar imaging:physical insight and mitigation strategies[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing,2014,8(3):1078-1086.
[11]程振宇,张灿,和智涛,等.基于3G网络视频传输的一种Qo S控制方法[J].中国科学院大学学报,2014,31(1):117-123.
[12]王军,陈翠琴.基于RFID信息与视频图像的人员识别的研究[J].物联网技术,2015,5(3):30-31.
[13]焦义文,王元钦,马宏,等.群时延失真对天线组阵合成信噪比的影响分析[J].信号处理2015,31(2)145-153.
[14]周亚兰.基于FPGA的多通道数据解码技术研究[J].物联网技术,2015,5(3):32-34.
[15]宋晓丽,刘冀伟,张晓星.分布式视频编码的关键帧提取算法[J].智能系统学报,2011,6(6):539-543.
[16]宋彬,贺红,刘海华等. Wyner-Ziv视频编码中边信息生成算法研究[J].通信学报,2010,31(12):97-103.
[17] R. Puri,K. Ramchandran. PRISM:an uplink-friendly multimedia coding paradigm:Proceedings of 2003 IEEE international conference on acoustics,speech,and Signal Processing[C]. Hong Kong,2003:856-859.
[18]孙一南,韩晓明.基于多算法融合的运动目标定位研究[J].微电子学与计算机,2017,34(8):123-127.
[19]孙丰,秦开怀,孙伟,等.一种针对移动相机的实时视频背景减除算法[J].计算机辅助设计与图形学学报,2016,28(4):573-579.
[20]罗会兰,赖泽云,孔繁胜.动作切分和流形度量学习的视频动作识别[J].中国图象图形学报,2017,22(8):1106-1119.