Optics and Precision Engineering, Volume. 31, Issue 4, 533(2023)
Quantitative evaluation method for structural similarity of multidimensional point cloud
Figures & Tables(11)
Fig. 2. Bunny， dragon and buddha models’ point clouds before add and remove Gaussian noise
Fig. 3.
Fig. 4. Similarity curves for cat models with different confidence levels regarding the
Fig. 5. Cat model of Top view of the point cloud with different confidence levels：
Fig. 6. Cat， car， man， head， aircraft carrier， plane point cloud datasets the pictures from top to bottom are：respectively， standard dataset， target extraction dataset， radius filtering， statistical filtering， low-pass filtering
Fig. 7. Correspondence between three-dimensional structure similarity values of bunny， dragon and armadillo and root mean square error of registration
Table 1. SSIM of bunny， dragon and Buddha before and after adding Gaussian noise
View tableView in ArticleTable 1. SSIM of bunny， dragon and Buddha before and after adding Gaussian noise
标准数据-标准数据 添加高斯噪声-标准数据 x-SSIM y-SSIM z-SSIM 3D-SSIM x-SSIM y-SSIM z-SSIM 3D-SSIM Bunny 1 1 1 1 0.921 86 0.994 69 0.960 3 0.880 56 Dragon 1 1 1 1 0.971 79 0.999 92 0.984 61 0.956 76 Buddha 1 1 1 1 0.981 97 0.999 94 0.989 42 0.971 52
Table 2. 3D-SSIM evaluation method of point cloud dataset based on cat， car， man， head， aircraft carrier and plane models
View tableView in ArticleTable 2. 3D-SSIM evaluation method of point cloud dataset based on cat， car， man， head， aircraft carrier and plane models
滤波方法 Cat点云数据集 Car点云数据集 x-SSIM y-SSIM z-SSIM 3DSSIM x-SSIM y-SSIM z-SSIM 3DSSIM 半径滤波 0.947 971 0.981 353 0.999 574 0.929 897 9 0.894 888 0.986 796 0.999 422 0.882 561 统计滤波 0.948 635 0.981 309 0.999 58 0.930 513 0.893 104 0.986 719 0.999 425 0.880 736 低通滤波 0.947 489 0.981 006 0.999 563 0.929 086 2 0.893 248 0.986 72 0.999 396 0.880 853 3 滤波方法 Man点云数据集 Head点云数据集 x-SSIM y-SSIM z-SSIM 3DSSIM x-SSIM y-SSIM z-SSIM 3DSSIM 半径滤波 0.999 304 0.951 542 0.996 985 0.948 012 8 0.997 347 0.995 341 0.998 968 0.991 675 9 统计滤波 0.999 246 0.967 378 0.996 866 0.963 619 0.996 317 0.998 451 0.999 68 0.994 455 低通滤波 0.999 227 0.936 821 0.996 373 0.932 701 6 0.996 306 0.996 21 0.996 373 0.988 930 1 滤波方法 Aircraft carrier点云数据集 Plane点云数据集 x-SSIM y-SSIM z-SSIM 3DSSIM x-SSIM y-SSIM z-SSIM 3DSSIM 半径滤波 0.992 45 0.994 53 0.998 78 0.985 817 1 0.964 051 0.994 53 0.998 78 0.957 607 9 统计滤波 0.997 52 0.998 324 0.999 236 0.995 087 0.964 13 0.999 62 0.996 931 0.960 805 8 低通滤波 0.996 304 0.996 207 0.999 029 0.991 561 3 0.983 494 0.996 23 0.996 91 0.976 759
Table 3. Point cloud registration of different 3D structure similarity values of bunny，dragon and armadillo
View tableView in ArticleTable 3. Point cloud registration of different 3D structure similarity values of bunny，dragon and armadillo
Bunny Dragon Armadillo 对照内容 KNN SOR 添加噪声 原图 KNN SOR 添加噪声 原图 KNN SOR 添加噪声 原图 匹配点对个数 1 820 2 097 2 182 1 567 2 660 3 160 2 173 1 994 1 725 1 929 2 039 1 424 距离最大值 3.384 67 3.447 44 12.007 48 3.622 26 1.785 31 1.758 23 1.478 26 1.327 44 4.115 62 3.606 46 3.704 68 3.430 48 距离最小值 0 0 0 0 0.006 55 0.006 81 0.017 53 0.012 78 0 0 0 0 均方根误差 1.562 44 1.856 02 1.966 59 1.465 46 2.782 92 2.645 83 2.454 16 2.080 55 2.308 17 1.832 08 2.023 54 1.759 13
Table 4. similarity values of different 3D structures of bunny，dragon and armadillo
View tableView in ArticleTable 4. similarity values of different 3D structures of bunny，dragon and armadillo
Bunny Dragon Armadillo 对照内容 KNN SOR 添加噪声 原图 KNN SOR 添加噪声 原图 KNN SOR 添加噪声 原图 x-SSIM 0.875 911 0.677 215 0.637 228 1 0.884 63 0.908 26 0.889 37 1 0.628 134 0.999 86 0.623 189 1 y-SSIM 0.971 333 0.971 365 0.902 463 1 0.988 59 0.988 59 0.988 27 1 0.329 717 0.350 26 0.340 384 1 z-SSIM 0.822 713 0.821 984 0.823 096 1 1 0.817 14 0.803 92 0.814 05 1 0.967 066 0.870 75 0.870 993 1 1 3D-SSIM 0.699 965 2 0.540 719 9 0.473 341 7 1 0.714 63 0.721 85 0.715 50 1 0.200 285 0.304 95 0.184 758 16 1
Tools
Get Citation
Copy Citation Text
Ziqian YANG, Yanqiu WANG, Fu ZHENG, Zhibin SUN. Quantitative evaluation method for structural similarity of multidimensional point cloud[J]. Optics and Precision Engineering, 2023, 31(4): 533
Paper Information
Category: Information Sciences
Received: Feb. 23, 2022
Accepted: --
Published Online: Mar. 7, 2023
The Author Email: SUN Zhibin (zbsun@nssc.ac.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20