海面弱小目标红外检测算法的高速实现
|
摘要
为了实现红外图像中海面弱小目标的精确检测,提出了一种基于局部峰值检测和管道滤波的红外图像处理算法。首先采取局部峰值检测提取疑似目标,然后根据自适应域值处理去除多数非目标峰值,最后通过管道滤波法排除残留干扰以准确识别目标。针对算法中包括大量条件判断和并行计算的特点,通过比对CPU和GPU的工作特性,最终采用CPU-GPU协作的异构计算模型对算法进行了加速。实验结果表明,在大量海面杂波的干扰下,该加速检测算法运行后的目标检测漏警率不高于3.5%,虚警率不高于5%,加速比为26,处理分辨率为640×512图像的速率不低于32帧/秒,具有很高的工程应用价值。
In order to realize the accurate detection of small dim target in the sea through infrared image,a detection algorithm based on local peak detection and pipeline-filtering are put forward by my team.The algorithm first extracts some suspected targets by local peak detection,and then removes most of the non-target peaks according to the self-adaptive threshold processing.Finally,the residual interference is eliminated by the pipeline filtering method to identify the target accurately.In view of the characteristics of the algorithm including a large number of conditions and parallel computing,this paper compares the work characteristics of CPU and GPU,and finally speeds up the algorithm by using the heterogeneous computing model of CPU-GPU collaboration.The experimental results show that the leakage alarm rate of the target detection is not higher than 3.5%,the false alarm rate is lower than 5%,the acceleration ratio is 26,the rate of resolution processing 640×512 images is not less than 32 frames per second,and it has high engineering application value.
In order to realize the accurate detection of small dim target in the sea through infrared image,a detection algorithm based on local peak detection and pipeline-filtering are put forward by my team.The algorithm first extracts some suspected targets by local peak detection,and then removes most of the non-target peaks according to the self-adaptive threshold processing.Finally,the residual interference is eliminated by the pipeline filtering method to identify the target accurately.In view of the characteristics of the algorithm including a large number of conditions and parallel computing,this paper compares the work characteristics of CPU and GPU,and finally speeds up the algorithm by using the heterogeneous computing model of CPU-GPU collaboration.The experimental results show that the leakage alarm rate of the target detection is not higher than 3.5%,the false alarm rate is lower than 5%,the acceleration ratio is 26,the rate of resolution processing 640×512 images is not less than 32 frames per second,and it has high engineering application value.
引文
[1] WANG Jun,JIANG Zhi,SUN Hui-ting et al.Detection and tracking of weak infrared targets based on noise variance estimation[J].Journal of Optoelectronics·Laser,2018,29(3):305-311.王军,姜志,孙慧婷,等.基于噪声方差估计的红外弱小目标检测与跟踪方法[J].光电子·激光,2018,29(3):305-311.
[2] XU Song,HAN Guang-liang.Infrared small target fast detection based on local saliency[J].Acta Photonica Sinica,2013,42(2):228-233.薛松,韩广良.基于局部峰值的红外弱小目标快速检测[J].光子学报,2013,42(2):228-233.
[3] SUN Guo-dong,JI Shu-peng,ZHOU Zhen.Dim small IR sea target detection based on wavelet and context model[J].Infrared Technology,2010,32(2):97-100.孙国栋,吉书鹏,周桢.基于小波和Context模型的海面红外弱小目标检测[J].红外技术,2010,32(2):97-100.
[4] Diani M,Acito N,Corsini G.Dim target detection in IR maritime surveillance systems[J].2003 IEEE International Geoscience and Remote Sensing Symposium,2003,4:2650-2652.
[5] WANG Bin,DONG Li-li,ZHAO Ming,Xu Wen-hai.A small dim infrared maritime target detection algorithm based on local peak detection and pipeline-filtering[J].International conference on Graphic & Image Processing,2015,9817:98170U.
[6] YUAN Tao,MA Yan,LIU Ding-sheng.Review of GPU applications in remote sensing image processing[J].Remote Sensing Information,2012,27(6):110-117.袁涛,马艳,刘定生.GPU在遥感图像处理中的应用综述[J].遥感信息,2012,27(6):110-117.
[7] LU Feng-shui,SONG Jun-qiang,YIN Fu-kang,et al.Survey of CPU/GPU synergetic parallel computing[J].Computer Science,2011,38(3):5-9.卢风顺,宋君强,银福康,等.CPU/GPU协同并行计算研究综述[J].计算机科学,2011,38(3):5-9.
[8] JIA Hai-peng,ZHANG Yun-chuan,YUAN Liang,et al.Research of viola-jones face detection algorithm performance optimization based on openCL[J].Chinese Journal of Compluters,2016,39(9):1775-1788.贾海鹏,张云泉,袁良,等.基于OpenCL的Viola-Jones人脸检测算法性能优化研究[J].计算机学报,2016,39(9):1775-1788.
[9] Li Peng-long,Ding Yi,Duan Song-jiang,et al.A method of rapid distortion correction for UAV image based on GPU-CPU Co-processing technology[J].IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium,2017,5720-5723.
[10] Ming W,Feng-ming Z,Hong-li L,et al.A novel detection method of infrared dim and small target based on cross window[J].Acta Photonica Sinica,2011,40(5):764-768.
[11] Qiang Z,Jing-ju C,Qi-heng Z,et al.Small dim infrared targets segmentation method based on local maximum[J].Infrared Technology,2011,33(1):41-44.
[12] FAN Lu,GU Guo-hua,CHEN Qian,et al.Fhreshold updating CFAR detection algorithm of infrared small target under complex background[J].Infrared Technology,2011,33(10):559-563.范璐,顾国华,陈钱,等.复杂背景下的红外弱小目标阈值更新恒虚警算法[J].红外技术,2011,33(10):559-563.
[13] LIU Pan-zhi,HAN Chong-zhao,JIE Jing.A method to accurately estimate threshold factors for radar constant false alarm ratio detectors[J].Journal of Xi′an Jiaotong University,2009,43(2):67-71.刘盼芝,韩崇昭,介婧.一种精确估计恒虚警检测器阈值因子的方法[J].西安交通大学学报,2009,43(2):67-71.
[14] TAN Jie,XIONG Wei,ZHOU Wei.Dual-threshold detection method based on CFAR[J].Radar Science and Technology,2015,13(2):154-158.谭洁,熊伟,周伟.基于恒虚警率的双阈值检测方法[J].雷达科学与技术,2015,13(2):154-158.
[15] LIAN Ke,WANG Hou-jun,LI Dan.Pipeline filtering method based on feature analysis of local grey level of small infrared target[J].Journal of Projectiles,Rockets,Missiles and Guidance,2011,31(4):200-206.连可,王厚军,李丹.基于红外目标局部灰度特性分析的管道滤波方法[J].弹箭与制导学报,2011,31(4):200-206.
[16] ZHAO Xiao-ming,YUAN Sheng-chun,MA Xiao-li,et al.Research on infrared small target detection technique based on moving pipeline filtering[J].Infrared Technology,2009,31(5):295-297.赵小明,袁胜春,马晓丽,等.基于移动式管道滤波的红外小目标检测方法研究[J].红外技术,2009,31(5):295-297.
[17] WANG Chang-cheng,SHEN Yu-heng,ZHANG Di,et al.A dynamic queue based pipeline filter for infrared dimsmall target detection[J].Chinese Control Conference (CCC),2015,34:3770-3775.
[18] LI Ming-jie,HU Ming-yong,ZHANG Jian,et al.Real-time imaging destortion correction of large-field objective lens based on CPU+GPU hybrid platform[J].Acta Photonica Sinica,2018,47(6):0611002-1～8.李明杰,胡明勇,张健,等.基于CPU+GPU的大视场物镜成像畸变实时校正[J].光子学报,2018,47(6):0611002-1～8.
[19] LIU Xin,JIANG Chao,FENG Cun-yong.Image parallel processing based on CUDA and OpenCV[J].Science of Surveying and Mapping,2012,37(4):123-125.刘鑫,姜超,冯存永.CUDA和OpenCV图像并行处理方法研究[J].测绘科学,2012,37(4):123-125.
[20] SUN Li-chao,ZHANG Sheng-bing,CHENG Xun-tao,et al.Fast face detection algorithm based on CUDA[J].Jisuanji Yu Xiandaihua,2013,8:12-14.孙立超,张盛兵,程训焘,等.基于CUDA的快速人脸检测算法[J].计算机与现代化,2013,8:12-14.
[2] XU Song,HAN Guang-liang.Infrared small target fast detection based on local saliency[J].Acta Photonica Sinica,2013,42(2):228-233.薛松,韩广良.基于局部峰值的红外弱小目标快速检测[J].光子学报,2013,42(2):228-233.
[3] SUN Guo-dong,JI Shu-peng,ZHOU Zhen.Dim small IR sea target detection based on wavelet and context model[J].Infrared Technology,2010,32(2):97-100.孙国栋,吉书鹏,周桢.基于小波和Context模型的海面红外弱小目标检测[J].红外技术,2010,32(2):97-100.
[4] Diani M,Acito N,Corsini G.Dim target detection in IR maritime surveillance systems[J].2003 IEEE International Geoscience and Remote Sensing Symposium,2003,4:2650-2652.
[5] WANG Bin,DONG Li-li,ZHAO Ming,Xu Wen-hai.A small dim infrared maritime target detection algorithm based on local peak detection and pipeline-filtering[J].International conference on Graphic & Image Processing,2015,9817:98170U.
[6] YUAN Tao,MA Yan,LIU Ding-sheng.Review of GPU applications in remote sensing image processing[J].Remote Sensing Information,2012,27(6):110-117.袁涛,马艳,刘定生.GPU在遥感图像处理中的应用综述[J].遥感信息,2012,27(6):110-117.
[7] LU Feng-shui,SONG Jun-qiang,YIN Fu-kang,et al.Survey of CPU/GPU synergetic parallel computing[J].Computer Science,2011,38(3):5-9.卢风顺,宋君强,银福康,等.CPU/GPU协同并行计算研究综述[J].计算机科学,2011,38(3):5-9.
[8] JIA Hai-peng,ZHANG Yun-chuan,YUAN Liang,et al.Research of viola-jones face detection algorithm performance optimization based on openCL[J].Chinese Journal of Compluters,2016,39(9):1775-1788.贾海鹏,张云泉,袁良,等.基于OpenCL的Viola-Jones人脸检测算法性能优化研究[J].计算机学报,2016,39(9):1775-1788.
[9] Li Peng-long,Ding Yi,Duan Song-jiang,et al.A method of rapid distortion correction for UAV image based on GPU-CPU Co-processing technology[J].IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium,2017,5720-5723.
[10] Ming W,Feng-ming Z,Hong-li L,et al.A novel detection method of infrared dim and small target based on cross window[J].Acta Photonica Sinica,2011,40(5):764-768.
[11] Qiang Z,Jing-ju C,Qi-heng Z,et al.Small dim infrared targets segmentation method based on local maximum[J].Infrared Technology,2011,33(1):41-44.
[12] FAN Lu,GU Guo-hua,CHEN Qian,et al.Fhreshold updating CFAR detection algorithm of infrared small target under complex background[J].Infrared Technology,2011,33(10):559-563.范璐,顾国华,陈钱,等.复杂背景下的红外弱小目标阈值更新恒虚警算法[J].红外技术,2011,33(10):559-563.
[13] LIU Pan-zhi,HAN Chong-zhao,JIE Jing.A method to accurately estimate threshold factors for radar constant false alarm ratio detectors[J].Journal of Xi′an Jiaotong University,2009,43(2):67-71.刘盼芝,韩崇昭,介婧.一种精确估计恒虚警检测器阈值因子的方法[J].西安交通大学学报,2009,43(2):67-71.
[14] TAN Jie,XIONG Wei,ZHOU Wei.Dual-threshold detection method based on CFAR[J].Radar Science and Technology,2015,13(2):154-158.谭洁,熊伟,周伟.基于恒虚警率的双阈值检测方法[J].雷达科学与技术,2015,13(2):154-158.
[15] LIAN Ke,WANG Hou-jun,LI Dan.Pipeline filtering method based on feature analysis of local grey level of small infrared target[J].Journal of Projectiles,Rockets,Missiles and Guidance,2011,31(4):200-206.连可,王厚军,李丹.基于红外目标局部灰度特性分析的管道滤波方法[J].弹箭与制导学报,2011,31(4):200-206.
[16] ZHAO Xiao-ming,YUAN Sheng-chun,MA Xiao-li,et al.Research on infrared small target detection technique based on moving pipeline filtering[J].Infrared Technology,2009,31(5):295-297.赵小明,袁胜春,马晓丽,等.基于移动式管道滤波的红外小目标检测方法研究[J].红外技术,2009,31(5):295-297.
[17] WANG Chang-cheng,SHEN Yu-heng,ZHANG Di,et al.A dynamic queue based pipeline filter for infrared dimsmall target detection[J].Chinese Control Conference (CCC),2015,34:3770-3775.
[18] LI Ming-jie,HU Ming-yong,ZHANG Jian,et al.Real-time imaging destortion correction of large-field objective lens based on CPU+GPU hybrid platform[J].Acta Photonica Sinica,2018,47(6):0611002-1～8.李明杰,胡明勇,张健,等.基于CPU+GPU的大视场物镜成像畸变实时校正[J].光子学报,2018,47(6):0611002-1～8.
[19] LIU Xin,JIANG Chao,FENG Cun-yong.Image parallel processing based on CUDA and OpenCV[J].Science of Surveying and Mapping,2012,37(4):123-125.刘鑫,姜超,冯存永.CUDA和OpenCV图像并行处理方法研究[J].测绘科学,2012,37(4):123-125.
[20] SUN Li-chao,ZHANG Sheng-bing,CHENG Xun-tao,et al.Fast face detection algorithm based on CUDA[J].Jisuanji Yu Xiandaihua,2013,8:12-14.孙立超,张盛兵,程训焘,等.基于CUDA的快速人脸检测算法[J].计算机与现代化,2013,8:12-14.