基于分类的感兴趣目标检测算法及仿真验证
|
摘要
针对在弹丸冲击靶板后效过程中,由于靶板破片对弹丸的遮挡,影响对弹丸姿态等信息的获取的问题,提出了一种群目标分类的合成孔径成像算法,利用分类结果筛选多视角图像中感兴趣目标,以某一视角为参考进行合成孔径成像,减少前景遮挡对目标信息探测的干扰。仿真实验结果表明:该算法能够有效去除弹丸前景的遮挡物,提高了弹丸目标的重建质量,验证了所提算法的有效性。
During the aftereffect of the bullet impacting target plate,due to the occlusion of the bullet by fragments of the target plate,acquisition of information such as how the bullet posture is affected. Aiming at this problem,a synthetic aperture imaging algorithm for group target classification is proposed. The classification results are used to filter the interested target in multi-view images,and synthetic aperture imaging is performed with reference to a certain angle of view to reduce the interference of foreground occlusion on target information detection. The simulation results show that the occlusion in front of the bullet can be effectively removed and the reconstruction quality of the bullet target is improved sufficiently,which verify the effectiveness of the proposed algorithm.
During the aftereffect of the bullet impacting target plate,due to the occlusion of the bullet by fragments of the target plate,acquisition of information such as how the bullet posture is affected. Aiming at this problem,a synthetic aperture imaging algorithm for group target classification is proposed. The classification results are used to filter the interested target in multi-view images,and synthetic aperture imaging is performed with reference to a certain angle of view to reduce the interference of foreground occlusion on target information detection. The simulation results show that the occlusion in front of the bullet can be effectively removed and the reconstruction quality of the bullet target is improved sufficiently,which verify the effectiveness of the proposed algorithm.
引文
[1] Vaish V,Levoy M,Szeliski R,et al. Reconstructing occludedsurfaces using synthetic apertures:Stereo,focus and robustmeasures[C]∥Proceedings of IEEE Computer Socity Conferenceon Computer Vision and Pattern Recognition,2006:2331-2338.
[2] Boykov Y,Jolly M P. Interactive graph cuts for optimal boundaryand region segmentation of objects in N-D images[C]∥Procee-dings of the 8th IEEE International Conference on ComputerVision,Vancouver,Canada,IEEE,2001:105-112.
[3] Pei Z,Zhang Y,Chen X,et al. Synthetic aperture imaging usingpixel labeling via energy minimization[J]. Pattern Recognition,2013,46(1):174-187.
[4] Ma W,Yang T,Zhang Y,et al. Unstructured synthetic aperturephotograph based occluded object imaging[C]∥2013 TheSeventh International Conference on Image and Graphics,IEEE,2013:34-39.
[5]刘严羊硕,刘宾,潘晋孝.基于前景移除的合成孔径成像算法[J].光学学报,2018,38(6):154-161.
[6] Song Z,Zhang Y,Yang T,et al. High-quality synthetic apertureauto-imaging under occlusion[J]. International Conference onIntelligent Science and Intelligent Data Engineering,2012:407-416.
[7] Park J H,Min S W,Jung S,et al. New stereovision scheme usinga camera and a lens array[C]∥Algorithms and Systems forOptical Information Processing,San Diego:2001:73-80.
[8] Park J H,Jung S,Choi H,et al. Depth extraction by use of arectangular lens array and one-dimensional elemental imagemodification[J]. Appl Opt,2004,43(25):4882-4895.
[9] Choi H,Park J H,Hong J,et al. An improved stereovision schemeusing one camera and a composite lens array[C]∥StereoscopicDisplays&Virtual Reality Systems,California:InternationalSociety for Optics and Photonics,2004:196-203.
[10] Klaus Halbach. Matrix representation of Gaussian optics[J].American Journal of Physics,1964,32(2):90-108.
[11] Rodriguez A,Laio A. Clustering by fast search and find of densitypeaks[J]. Science,2014,344:1492-1496.
[12] Arthur D,Arthur D,Vassilvitskii S,et al. K-means++:Theadvantages of careful seeding[C]∥The Eighteenth ACM-SIAMSymposium on Discrete Algorithms,New Orleans:Society forIndustrial and Applied Mathematics,2007:1027-103.
[13]李昌兴,黄艳虎,支晓斌,等.基于加速K均值的谱聚类图像分割算法改进[J].传感器与微系统,2016,35(9):137-140.
[2] Boykov Y,Jolly M P. Interactive graph cuts for optimal boundaryand region segmentation of objects in N-D images[C]∥Procee-dings of the 8th IEEE International Conference on ComputerVision,Vancouver,Canada,IEEE,2001:105-112.
[3] Pei Z,Zhang Y,Chen X,et al. Synthetic aperture imaging usingpixel labeling via energy minimization[J]. Pattern Recognition,2013,46(1):174-187.
[4] Ma W,Yang T,Zhang Y,et al. Unstructured synthetic aperturephotograph based occluded object imaging[C]∥2013 TheSeventh International Conference on Image and Graphics,IEEE,2013:34-39.
[5]刘严羊硕,刘宾,潘晋孝.基于前景移除的合成孔径成像算法[J].光学学报,2018,38(6):154-161.
[6] Song Z,Zhang Y,Yang T,et al. High-quality synthetic apertureauto-imaging under occlusion[J]. International Conference onIntelligent Science and Intelligent Data Engineering,2012:407-416.
[7] Park J H,Min S W,Jung S,et al. New stereovision scheme usinga camera and a lens array[C]∥Algorithms and Systems forOptical Information Processing,San Diego:2001:73-80.
[8] Park J H,Jung S,Choi H,et al. Depth extraction by use of arectangular lens array and one-dimensional elemental imagemodification[J]. Appl Opt,2004,43(25):4882-4895.
[9] Choi H,Park J H,Hong J,et al. An improved stereovision schemeusing one camera and a composite lens array[C]∥StereoscopicDisplays&Virtual Reality Systems,California:InternationalSociety for Optics and Photonics,2004:196-203.
[10] Klaus Halbach. Matrix representation of Gaussian optics[J].American Journal of Physics,1964,32(2):90-108.
[11] Rodriguez A,Laio A. Clustering by fast search and find of densitypeaks[J]. Science,2014,344:1492-1496.
[12] Arthur D,Arthur D,Vassilvitskii S,et al. K-means++:Theadvantages of careful seeding[C]∥The Eighteenth ACM-SIAMSymposium on Discrete Algorithms,New Orleans:Society forIndustrial and Applied Mathematics,2007:1027-103.
[13]李昌兴,黄艳虎,支晓斌,等.基于加速K均值的谱聚类图像分割算法改进[J].传感器与微系统,2016,35(9):137-140.