Ship Classification Method for Point Cloud Images Based on Three-Dimensional Convolutional Neural Network

Yongmei Ren; Jie Yang; Zhiqiang Guo; Yilei Chen

doi:10.3788/LOP57.161022

Laser & Optoelectronics Progress, Volume. 57, Issue 16, 161022(2020)

Ship Classification Method for Point Cloud Images Based on Three-Dimensional Convolutional Neural Network

Yongmei Ren^1,2, Jie Yang^1、*, Zhiqiang Guo¹, and Yilei Chen³

¹Hubei Key Laboratory of Broadband Wireless Communication and Sensor Networks, School of Information Engineering, Wuhan University of Technology, Wuhan, Hubei 430070, China

²School of Electrical and Information Engineering, Hunan Institute of Technology, Hengyang, Hunan 421002, China

³School of Artificial Intelligence, Xidian University, Xi'an, Shaanxi 710071, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(12)

Fig. 1. Flow chart of proposed ship classification model

Download full size

Fig. 2. Structural diagram of typical CNN

Download full size

Fig. 3. Network structure of 3D CNN

Download full size

Fig. 4. CAD data and point cloud ship images. (a)(b) Cabin; (c)(d) rowing; (e)(f) sailing; (g)(h) cruise; (i)(j) cargo

Download full size

Fig. 5. Point cloud data samples in Sydney urban object dataset

Download full size

Table 1. Detail parameters of 3D CNN

View table

Table 1. Detail parameters of 3D CNN

Layer	Input size	Filter size	Stride	Output size	Number of parameters
Conv1	32×32×32×1	5×5×5×32	2	14×14×14×32	4032
Conv2	14×14×14×32	3×3×3×32	1	12×12×12×32	27680
Conv3	12×12×12×32	3×3×3×64	1	10×10×10×64	55360
Max Pooling 1	10×10×10×64	2×2×2	2	5×5×5×64	0
FC1	5×5×5×64	-	-	512	4096001
FC2	512	-	-	128	65537
FC3-Softmax	128	-	-	5	641

Table 2. Numbers of training and testing samples in self-build point cloud image ship dataset
View table
Table 2. Numbers of training and testing samples in self-build point cloud image ship dataset
No. Class Number of samplesin training set Number of samplesin testing set
1 Cabin 231 57
2 Rowing 231 57
3 Sailing 231 57
4 Cruise 231 57
5 Cargo 231 57
Total 1155 285

Table 3. Numbers of training and testing samples in ship dataset of point cloud images without noise
View table
Table 3. Numbers of training and testing samples in ship dataset of point cloud images without noise
No. Class Number of samplesin training set Number of samplesin testing set
1 Cabin 116 28
2 Rowing 116 28
3 Sailing 116 28
4 Cruise 116 28
5 Cargo 116 28
Total 580 140

Table 4. Classification accuracy of proposed 3D CNN model under each size of voxel grid
View table
Table 4. Classification accuracy of proposed 3D CNN model under each size of voxel grid
Size 32×32×32 48×48×48
Accuracy /% 97.14 95.71

Table 5. Classification accuracy, F1-score, and training time of each method on ship dataset of point cloud images without noise
View table
Table 5. Classification accuracy, F1-score, and training time of each method on ship dataset of point cloud images without noise
Method Accuracy /% F1-score Training time
PFH+BoW+SVM 96.43 0.9669 2.95 h
3D ShapeNets 90.71 0.9056 32.63 s
VoxNet 95.00 0.9500 13.44 s
Method in Ref. [17] 95.71 0.9566 62.25 s
Proposed method 97.14 0.9714 15.88 s

Table 6. Classification accuracy, F1-score, and training time of each method on self-built point cloud image ship dataset
View table
Table 6. Classification accuracy, F1-score, and training time of each method on self-built point cloud image ship dataset
Method Accuracy /% F1-score Training time/s
3D ShapeNets 90.17 0.9006 64.91
VoxNet 93.68 0.9358 28.01
Method in Ref. [17] 94.73 0.9471 144.47
Proposed method 96.14 0.9613 32.91

Table 7. Classification accuracy, F1-score, and training time of each method on Sydney urban object dataset
View table
Table 7. Classification accuracy, F1-score, and training time of each method on Sydney urban object dataset
Method Accuracy /% F1-score Training time/s
GFH+SVM^[12] 73.58 - -
VoxNet 89.51 0.8939 77.11
Method in Ref. [15] 84.00 - -
Method in Ref. [17] 87.37 0.8661 445.05
Proposed method 91.58 0.9153 90.45

Tools

Get Citation

Copy Citation Text

Yongmei Ren, Jie Yang, Zhiqiang Guo, Yilei Chen. Ship Classification Method for Point Cloud Images Based on Three-Dimensional Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2020, 57(16): 161022

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Image Processing

Received: Dec. 10, 2019

Accepted: Jan. 16, 2020

Published Online: Aug. 5, 2020

The Author Email: Jie Yang (jieyang@whut.edu.cn)

DOI:10.3788/LOP57.161022

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Detail parameters of 3D CNN

Table 1. Detail parameters of 3D CNN

Table 2. Numbers of training and testing samples in self-build point cloud image ship dataset

Table 2. Numbers of training and testing samples in self-build point cloud image ship dataset

Table 3. Numbers of training and testing samples in ship dataset of point cloud images without noise

Table 3. Numbers of training and testing samples in ship dataset of point cloud images without noise

Table 4. Classification accuracy of proposed 3D CNN model under each size of voxel grid

Table 4. Classification accuracy of proposed 3D CNN model under each size of voxel grid

Table 5. Classification accuracy, F1-score, and training time of each method on ship dataset of point cloud images without noise

Table 5. Classification accuracy, F1-score, and training time of each method on ship dataset of point cloud images without noise

Table 6. Classification accuracy, F1-score, and training time of each method on self-built point cloud image ship dataset

Table 6. Classification accuracy, F1-score, and training time of each method on self-built point cloud image ship dataset

Table 7. Classification accuracy, F1-score, and training time of each method on Sydney urban object dataset

Table 7. Classification accuracy, F1-score, and training time of each method on Sydney urban object dataset