Point Cloud Guided Filtering Algorithm Based on Optimal Neighborhood Feature Weighting

Zhibo Xu; Lü Qiujuan; Xinbin Gan; Jiamin Tan; Yongsheng Liu

doi:10.3788/LOP240827

Laser & Optoelectronics Progress, Volume. 61, Issue 14, 1415009(2024)

Point Cloud Guided Filtering Algorithm Based on Optimal Neighborhood Feature Weighting

Zhibo Xu^1,2, Lü Qiujuan³, Xinbin Gan¹, Jiamin Tan¹, and Yongsheng Liu^1,2、*

¹Key Laboratory of Road Construction Technology and Equipment of Ministry of Education, School of Construction Machinery, Chang'an University, Xi'an 710064, Shaanxi , China

²AVIC JONHON Optronic Technology Co., LTD., Luoyang 471003, Henan , China

³Department of Basics, Rocket Force University of Engineering, Chinese People's Liberation Army, Xi'an 710025, Shaanxi , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Point cloud guided filtering flow chart based on optimal neighborhood feature weighting

Download full size

Fig. 2. Based on two-parameter feature point recognition effect. (a) Bunny; (b) Horse; (c) Dragon; (d) Armadillo

Download full size

Fig. 3. Influence of parameter ε on the model fairing results. (a) Noise model; (b) ε=1.00; (c) ε=0.75; (d) ε=0.50; (e) ε=0.25; (f) ε=0.01

Download full size

Fig. 4. Horse triangulation model under small scale noise with different scales. (a) n=0.05; (b) n=0.10; (c) n=0.15; (d) n=0.20

Download full size

Fig. 5. Fairing effect of guided filtering algorithm with neighborhood adaptive feature preservation. (a) n=0.05; (b) n=0.10; (c) n=0.15; (d) n=0.20

Download full size

Fig. 6. Fairing effect of each method on Dragon model. (a) Dragon noise model; (b) Laplace algorithm; (c) WLOP algorithm; (d) BF algorithm; (e) MLS algorithm; (f) proposed algorithm

Download full size

Fig. 7. Fairing effect of each method on Armadillo model. (a) Armadillo noise model; (b) Laplace algorithm; (c) WLOP algorithm; (d) BF algorithm; (e) MLS algorithm; (f) proposed algorithm

Download full size

Table 1. Feature point recognition statistics based on different feature descriptions

View table

Table 1. Feature point recognition statistics based on different feature descriptions

Model	Point	Based on normal variation		Based on surface variation		Double-parameter extraction
Model	Point	Feature number	Ratio /%	Feature number	Ratio /%	Feature number	Ratio /%
Bunny	35947	5223	14.52	6247	17.38	4543	12.64
Horse	48485	6404	13.21	8989	18.54	5687	11.73
Dragon	437645	27166	6.27	35628	8.14	25096	5.73
Armadillo	172974	18654	10.78	20564	11.90	15234	8.81

Table 2. Running time of each fairing algorithm at different point cloud models
View table
Table 2. Running time of each fairing algorithm at different point cloud models
Model Time /s
Armadillo Dragon
WLOP 96.532 215.612
BF 18.560 48.251
Laplace 17.254 30.361
MLS 23.532 80.254
Proposed algorithm 40.812 154.881

Table 3. Error statistics of different models by different fairing methods

View table

Table 3. Error statistics of different models by different fairing methods

Model	Error index	WLOP	BF	Laplace	MLS	Proposed algorithm
Armadillo	δ_m	4.012	3.624	4.125	4.356	3.014
	δ_max	10.125	8.345	11.235	9.521	5.851
	d_max	0.424	0.362	0.284	0.251	0.126
	d_RMS	0.0962	0.0512	0.0745	0.0522	0.0450
Dragon	δ_m	20.125	23.313	18.341	17.315	16.316
	δ_max	26.754	30.012	27.514	25.684	26.075
	d_max	1.182	0.821	0.725	0.525	0.542
	d_RMS	0.308	0.185	0.161	0.099	0.065

Tools

Get Citation

Copy Citation Text

Zhibo Xu, Lü Qiujuan, Xinbin Gan, Jiamin Tan, Yongsheng Liu. Point Cloud Guided Filtering Algorithm Based on Optimal Neighborhood Feature Weighting[J]. Laser & Optoelectronics Progress, 2024, 61(14): 1415009

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites