Point cloud denoising method based on image segmentation

Table 1. Image segmentation result of object 1
View table
View in Article
Table 1. Image segmentation result of object 1
区域分类区域数量/个区域面积/pixel
所有区域 95 189 121
无噪声的参考区域 3 169 811
分割后的噪声区域 66 3 828
分割后待定区域 26 15 482

Table 2. Noise region analysis results in the pending region of object 1

View table

View in Article

Table 2. Noise region analysis results in the pending region of object 1

参考区域序号	参考区域深度值 $Z_{2}$	待定区域深度值 $Z_{1}$	欧氏距离 $D$	是否为噪声区域
1	159.818 3	215.272 7	55.454 4	是
2	144.241 1	75.346 5	68.894 6	是
3	143.559 6	-4.577 1	148.136 7	是
4	158.726 2	92.736 0	65.990 2	是
5	160.997 1	14.606 1	146.391 0	是
6	151.417 2	208.560 8	57.143 6	是
7	205.179 8	147.161 1	58.018 7	是
8	181.190 1	118.275 7	62.914 4	是
9	138.865 6	197.095 4	58.229 8	是
10	147.426 4	80.324 7	67.101 7	是
11	183.524 4	184.737 3	1.212 9	否
12	167.690 2	222.436 2	54.746 0	是
13	147.950 8	80.241 1	67.709 7	是
14	150.068 9	206.895 6	56.826 7	是
15	156.919 7	89.304 8	67.614 9	是
16	184.347 5	49.567 6	134.779 9	是
17	161.378 0	96.476 5	64.901 5	是
18	181.561 9	234.080 3	52.518 4	是
19	152.544 8	208.501 0	55.956 2	是
20	172.023 5	225.655 6	53.632 1	是
21	151.583 1	207.749 2	56.166 1	是
22	153.093 8	6.480 2	146.613 6	是
23	155.636 5	88.337 0	67.299 5	是
24	212.872 6	156.054 2	56.818 4	是
25	153.356 1	209.507 8	56.151 7	是
26	142.632 7	94.418 6	48.214 1	是

Table 3. Denoising effect of object 1

View table

View in Article

Table 3. Denoising effect of object 1

去噪比较参数	本文所提方法	半径滤波算法	体素滤波算法	文献［20］所提算法
去噪点云数量/pixel	169 854	187 940	90 878	167 795
去噪时间/s	0.954	0.755	0.081	0.909
$P_{1}$ /pixel	169 811	167 857	12 702	167 736
$P_{2}$ /pixel	43	20 083	78 176	60
$Q / %$	99.974	89.314	13.977	99.964

Table 4. Image segmentation result of object 2
View table
View in Article
Table 4. Image segmentation result of object 2
区域分类区域数量/个区域面积/pixel
所有区域 44 153 489
无噪声的参考区域 1 142 435
分割后的噪声区域 24 5 531
分割后待定区域 19 5 223

Table 5. Noise region analysis results in the pending region of object 2

View table

View in Article

Table 5. Noise region analysis results in the pending region of object 2

参考区域序号	参考区域深度值 $Z_{2}$	待定区域深度值 $Z_{1}$	欧氏距离 $D$	是否为噪声区域
1	87.956 3	214.845 6	126.889 3	是
2	72.970 7	195.816 2	122.845 5	是
3	159.195 5	46.145 4	113.050 1	是
4	161.944 1	308.982 3	147.038 2	是
5	181.188 8	331.732 3	150.543 5	是
6	244.452 3	114.121 1	130.331 2	是
7	210.227 2	86.303 6	123.923 6	是
8	50.125 9	143.183 9	93.058 0	是
9	168.068 5	317.413 3	149.344 8	是
10	164.477 3	49.584 0	114.893 3	是
11	191.471 8	346.683 9	155.212 1	否
12	228.729 9	101.040 1	127.689 8	是
13	262.979 7	128.326 2	134.653 5	是
14	234.222 7	401.880 4	167.657 7	是
15	181.577 2	331.502 1	149.924 9	是
16	190.888 6	70.496 3	120.392 3	是
17	49.100 7	44.079 3	5.021 4	否
18	169.123 6	317.352 8	148.229 2	是
19	55.677 4	172.789 0	117.111 6	是

Table 6. Denoising effect of object 2

View table

View in Article

Table 6. Denoising effect of object 2

去噪比较参数	本文所提方法	半径滤波算法	体素滤波算法	文献［20］所提算法
去噪点云数量/pixel	143 182	158 428	76 608	141 446
去噪时间/s	0.905	0.621	0.070	0.896
$P_{1}$ /pixel	142 435	140 796	10 704	140 634
$P_{2}$ /pixel	747	17 632	65 904	812
$Q / %$	99.436	88.871	13.972	99.426

Table 7. Image segmentation result of object 3
View table
View in Article
Table 7. Image segmentation result of object 3
区域分类区域数量/个区域面积/pixel
所有区域 22 120 613
无噪声的参考区域 1 112 871
分割后的噪声区域 9 1 423
分割后待定区域 12 6 319

Table 8. Noise region analysis results in the pending region of object 3

View table

View in Article

Table 8. Noise region analysis results in the pending region of object 3

参考区域序号	参考区域深度值 $Z_{2}$	待定区域深度值 $Z_{1}$	欧氏距离 $D$	是否为噪声区域
1	93.798 9	-4.994 9	98.793 8	是
2	63.862 6	176.253 3	112.390 7	是
3	5.103 3	106.992 2	101.888 9	是
4	5.946 4	88.094 5	82.148 1	是
5	-11.777 4	85.561 5	97.338 9	是
6	-12.030 2	72.359 2	84.389 4	是
7	82.316 5	203.897 7	121.581 2	是
8	126.799 5	35.294 4	91.505 1	是
9	200.582 4	356.998 5	156.416 1	是
10	79.323 2	74.390 4	4.932 8	否
11	178.757 3	329.011 4	150.254 1	是
12	-0.441 1	81.474 9	81.916 0	是

Table 9. Denoising effect of object 3

View table

View in Article

Table 9. Denoising effect of object 3

去噪比较参数	本文所提方法	半径滤波算法	体素滤波算法	文献［20］所提算法
去噪点云数量/pixel	112 934	124 960	60 432	111 565
去噪时间/s	0.875	0.574	0.058	0.862
$P_{1}$ /pixel	112 871	119 174	10 309	111 495
$P_{2}$ /pixel	63	5786	50 123	70
$Q / %$	99.944	95.370	17.059	99.937

Tools

Get Citation

Copy Citation Text

Jia-le LONG, Zi-hao DU, Jian-min ZHANG, Fu-jian CHEN, Hao-yuan GUAN, Ke-sen HUANG, Rui SUN. Point cloud denoising method based on image segmentation[J]. Chinese Journal of Liquid Crystals and Displays, 2023, 38(1): 104

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Research Articles

Received: May. 17, 2022

Accepted: --

Published Online: Feb. 20, 2023

The Author Email: Jian-min ZHANG (zjm99_2001@126.com)

DOI:10.37188/CJLCD.2022-0171

Topics

Table 1. Image segmentation result of object 1

Table 1. Image segmentation result of object 1

Table 2. Noise region analysis results in the pending region of object 1

Table 2. Noise region analysis results in the pending region of object 1

Table 3. Denoising effect of object 1

Table 3. Denoising effect of object 1

Table 4. Image segmentation result of object 2

Table 4. Image segmentation result of object 2

Table 5. Noise region analysis results in the pending region of object 2

Table 5. Noise region analysis results in the pending region of object 2

Table 6. Denoising effect of object 2

Table 6. Denoising effect of object 2

Table 7. Image segmentation result of object 3

Table 7. Image segmentation result of object 3

Table 8. Noise region analysis results in the pending region of object 3

Table 8. Noise region analysis results in the pending region of object 3

Table 9. Denoising effect of object 3

Table 9. Denoising effect of object 3