Chengdu Rainpoo Technology Co., Ltd.

Chengdu Rainpoo Technology Co., Ltd.

Chengdu Rainpoo Technology Co., Ltd.

Corporate News

Article

Article
How GSD and carrier drone types affect the relative accuracy of 3D model

In the last article How ground control points and PPK data affect the relative accuracy of 3D model,we have mentioned that in addition to GCPs and PPK datas, the GSD and carrier drone also have a certain degree of effects on the relative accuracy of the 3D model.So we do another test to verify it.

Experiment 1: the effect of GSD on the relative accuracy of the 3D model;

Condition 1

Carrier drone

Oblique camera

GCPs

PPK

 CW10 VTOL

DG4pros

NO

NO

Result Table 1:

Number of Objects

measurement length(L0)

3D model length(L1)GSD=2cm

3D model length(L2)GSD=1.5cm

DS1(L0-L1)

DS2(L0-L2)

1

7.210

7.18

7.20

0.030

0.010

 

8.706

8.69

8.65

0.016

0.056

 

10.961

10.89

10.90

0.071

0.061

2

7.010

6.88

6.86

0.130

0.150

3

1.822

1.76

1.76

0.062

0.062

4

10.410

10.38

10.37

0.030

0.040

5

10.718

10.65

10.68

0.068

0.038

6

13.787

13.72

13.71

0.067

0.077

7

11.404

11.41

11.38

-0.006

0.024

 

12.147

12.13

12.12

0.017

0.027

8

7.526

7.44

7.49

0.086

0.036

 

13.797

13.83

13.77

-0.033

0.027

9

10.374

10.35

10.35

0.024

0.024

10

2.109

1.98

2.09

0.129

0.019

 

4.281

4.14

4.19

0.141

0.091

11

14.675

14.55

14.64

0.125

0.035

 

8.600

8.58

8.57

0.020

0.030

12

13.394

13.36

13.38

0.034

0.014

13

12.940

12.95

12.91

-0.010

0.030

14

7.190

7.21

7.20

-0.020

-0.010

15

13.371

13.36

13.38

0.011

-0.009

 

6.435

6.37

6.43

0.065

0.005

16

3.742

3.74

3.70

0.002

0.042

17

6.022

6.03

5.99

-0.008

0.032

18

3.937

3.91

3.94

0.027

-0.003

19

8.120

8.09

8.12

0.030

0.000

 

14.411

14.40

14.37

0.011

0.041

20

6.077

6.06

6.04

0.017

0.037

21

13.696

13.68

13.65

0.016

0.046

RMSE: DS1=0.0397m,DS2=0.0356m

Condition 2

Carrier drone

Oblique camera

GCPs

PPK

CW10 VTOL

DG4pros

Yes

No

Result Table 2:

Number of Objects

measurement length(L0)

3D model length(L1)GSD=2cm

3D model length(L2)GSD=1.5cm

DS1(L0-L1)

DS2(L0-L2)

1

7.210

7.18

7.21

0.030

0

 

8.706

8.68

8.68

0.026

0.026

 

10.961

10.91

10.87

0.051

0.091

2

7.010

6.92

6.88

0.090

0.13

3

1.822

1.79

1.76

0.032

0.062

4

10.410

10.39

10.38

0.020

0.03

5

10.718

10.66

10.72

0.058

-0.002

6

13.787

13.77

13.79

0.017

-0.003

7

11.404

11.38

11.38

0.024

0.024

 

12.147

12.12

12.13

0.027

0.017

8

7.526

7.47

7.53

0.056

-0.004

 

13.797

13.83

13.78

-0.033

0.017

9

10.374

10.34

10.37

0.034

0.004

10

2.109

2.03

2.11

0.079

-0.001

 

4.281

4.18

4.28

0.101

0.001

11

14.675

14.59

14.67

0.085

0.005

 

8.600

8.57

8.58

0.030

0.02

12

13.394

13.35

13.39

0.044

0.004

13

12.940

12.92

12.9

0.020

0.04

14

7.190

7.21

7.19

-0.020

0

15

13.371

13.36

13.37

0.011

0.001

 

6.435

6.43

6.42

0.005

0.015

16

3.742

3.72

3.7

0.022

0.042

17

6.022

6.00

5.98

0.022

0.042

18

3.937

3.94

3.91

-0.003

0.027

19

8.120

8.09

8.07

0.030

0.05

 

14.411

14.41

14.38

0.001

0.031

20

6.077

6.03

6.04

0.047

0.037

21

13.696

13.65

13.65

0.046

0.046

RMSE:  DS1=0.0328m,DS2=0.0259m

Condition 3

Carrier drone

Oblique camera

GCPs

PPK

CW10 VTOL

DG4pros

NO

YES

 

Result Table 3:

Number of Objects

measurement length(L0)

3D model length(L1)GSD=2cm

3D model length(L2)GSD=1.5cm

DS1(L0-L1)

DS2(L0-L2)

1

7.210

7.20

7.20

0.01

0.01

 

8.706

8.71

8.69

0.00

0.02

 

10.961

10.87

10.88

0.09

0.08

2

7.010

6.89

6.87

0.12

0.14

3

1.822

1.79

1.75

0.03

0.07

4

10.410

10.39

10.38

0.02

0.03

5

10.718

10.67

10.69

0.05

0.03

6

13.787

13.75

13.78

0.04

0.01

7

11.404

11.39

11.38

0.01

0.02

 

12.147

12.13

12.12

0.02

0.03

8

7.526

7.51

7.49

0.02

0.04

 

13.797

13.79

13.78

0.01

0.02

9

10.374

10.35

10.34

0.02

0.03

10

2.109

2.03

2.02

0.08

0.09

 

4.281

4.15

4.21

0.13

0.07

11

14.675

14.61

14.65

0.07

0.03

 

8.600

8.60

8.57

0.00

0.03

12

13.394

13.37

13.40

0.02

-0.01

13

12.940

12.88

12.89

0.06

0.05

14

7.190

7.20

7.18

-0.01

0.01

15

13.371

13.36

13.38

0.01

-0.01

 

6.435

6.40

6.46

0.03

-0.03

16

3.742

3.75

3.75

-0.01

-0.01

17

6.022

5.97

5.97

0.05

0.05

18

3.937

3.93

3.91

0.01

0.03

19

8.120

8.10

8.08

0.02

0.04

 

14.411

14.40

14.38

0.01

0.03

20

6.077

6.04

6.05

0.04

0.03

21

13.696

13.65

13.67

0.05

0.03

RMSE: DS1=0.0342m,DS2=0.0328m

Condition 4

Carrier drone

Oblique camera

GCPs

PPK

CW10 VTOL

DG4pros

YES

YES

Result Table 4:

Number of Objects

measurement length(L0)

3D model length(L1)GSD=2cm

3D model length(L2)GSD=1.5cm

DS1(L0-L1)

DS2(L0-L2)

1

7.21

7.21

7.21

0.00

0.00

8.706

8.68

8.7

0.03

0.01

10.961

10.90

10.89

0.06

0.07

2

7.01

6.98

6.99

0.03

0.02

3

1.822

1.80

1.78

0.02

0.04

4

10.41

10.40

10.39

0.01

0.02

5

10.718

10.70

10.69

0.02

0.03

6

13.787

13.77

13.76

0.02

0.03

7

11.404

11.39

11.39

0.01

0.01

12.147

12.13

12.12

0.02

0.03

8

7.526

7.49

7.51

0.04

0.02

13.797

13.81

13.8

-0.01

0.00

9

10.374

10.36

10.35

0.01

0.02

10

2.109

2.02

2.11

0.09

0.00

4.281

4.16

4.28

0.12

0.00

11

14.675

14.66

14.68

0.02

0.00

8.6

8.54

8.53

0.06

0.07

12

13.394

13.35

13.37

0.04

0.02

13

12.94

12.89

12.91

0.05

0.03

14

7.19

7.18

7.2

0.01

-0.01

15

13.371

13.35

13.35

0.02

0.02

6.435

6.41

6.4

0.02

0.03

16

3.742

3.72

3.71

0.02

0.03

17

6.022

5.98

5.98

0.04

0.04

18

3.937

3.89

3.89

0.05

0.05

19

8.12

8.12

8.1

0.00

0.02

14.411

14.40

14.39

0.01

0.02

20

6.077

6.03

6.03

0.05

0.05

21

13.696

13.66

13.64

0.04

0.06

RMSE: DS1=0.0308m,DS2=0.0249m

Conclusions

The variables of experiment 1 are:

1:With/Without PPK data.

2:With/Without GCPs.

3:different GSD:1.5 cm or 2cm

After analysis of the four sets of experimental data, we can draw the following conclusions:

 

When the oblique camera is DG4Pros:

(1) With/Without ground control point (GCP) and PPK data involved in the AT triangulation, whether it is a VTOL or a multi-rotor drone, whether the GSD is 2cm or 1.5cm, the 3D model built by oblique camera DG4Pros can meet the requirements of relative accuracy error Ds≤10cm(Actually ≤5cm).

(2) When the GSD is the single variable, the relative accuracy of the 3D model with GSD=1.5cm is better than that GSD=2cm.

Experiment 2: the effect of carrier drone types on the relative accuracy of the 3D model

Condition 1

Oblique camera

GCPs

PPK

GSD

DG4pros

NO

NO

1.5cm

Result Table 1:

Number of Objects

measurement length(L0)

3D model length(L1)CW10

3D model length(L2)M600Pro

DS1(L0-L1)

DS2(L0-L2)

1

7.21

7.20

7.19

0.010

0.02

 

8.706

8.65

8.70

0.056

0.01

 

10.961

10.90

10.91

0.061

0.05

2

7.01

6.86

6.98

0.150

0.03

3

1.822

1.76

1.79

0.062

0.03

4

10.41

10.37

10.39

0.040

0.02

5

10.718

10.68

10.69

0.038

0.03

6

13.787

13.71

13.76

0.077

0.03

7

11.404

11.38

11.38

0.024

0.02

 

12.147

12.12

12.12

0.027

0.03

8

7.526

7.49

7.50

0.036

0.03

 

13.797

13.77

13.77

0.027

0.03

9

10.374

10.35

10.33

0.024

0.04

10

2.109

2.09

2.02

0.019

0.09

 

4.281

4.19

4.20

0.091

0.08

11

14.675

14.64

14.65

0.035

0.03

 

8.6

8.57

8.57

0.030

0.03

12

13.394

13.38

13.35

0.014

0.04

13

12.94

12.91

12.92

0.030

0.02

14

7.19

7.20

7.17

-0.010

0.02

15

13.371

13.38

13.35

-0.009

0.02

 

6.435

6.43

6.41

0.005

0.02

16

3.742

3.70

3.70

0.042

0.04

17

6.022

5.99

5.99

0.032

0.03

18

3.937

3.94

3.89

-0.003

0.05

19

8.12

8.12

8.08

0.000

0.04

 

14.411

14.37

14.36

0.041

0.05

20

6.077

6.04

6.06

0.037

0.02

21

13.696

13.65

13.65

0.046

0.05

RMSE:  DS1=0.0356m,DS2=0.342m

Condition 2

Oblique camera

GCPs

PPK

GSD

DG4pros

YES

NO

1.5cm

 

Result Table 2:

Number of Objects

measurement length(L0)

3D model length(L1)CW10

3D model length(L2)M600Pro

DS1(L0-L1)

DS2(L0-L2)

1

7.21

7.21

7.21

0

0.00

 

8.706

8.68

8.70

0.026

0.01

 

10.961

10.87

10.91

0.091

0.05

2

7.01

6.88

6.98

0.13

0.03

3

1.822

1.76

1.80

0.062

0.02

4

10.41

10.38

10.39

0.03

0.02

5

10.718

10.72

10.70

-0.002

0.02

6

13.787

13.79

13.75

-0.003

0.04

7

11.404

11.38

11.38

0.024

0.02

 

12.147

12.13

12.13

0.017

0.02

8

7.526

7.53

7.49

-0.004

0.04

 

13.797

13.78

13.79

0.017

0.01

9

10.374

10.37

10.35

0.004

0.02

10

2.109

2.11

2.07

-0.001

0.04

 

4.281

4.28

4.21

0.001

0.07

11

14.675

14.67

14.66

0.005

0.02

 

8.6

8.58

8.57

0.02

0.03

12

13.394

13.39

13.35

0.004

0.04

13

12.94

12.9

12.93

0.04

0.01

14

7.19

7.19

7.18

0

0.01

15

13.371

13.37

13.36

0.001

0.01

 

6.435

6.42

6.42

0.015

0.01

16

3.742

3.7

3.71

0.042

0.03

17

6.022

5.98

6.00

0.042

0.02

18

3.937

3.91

3.91

0.027

0.03

19

8.12

8.07

8.10

0.05

0.02

 

14.411

14.38

14.35

0.031

0.06

20

6.077

6.04

6.06

0.037

0.02

21

13.696

13.65

13.67

0.046

0.03

RMSE:  DS1=0.0259m,DS2=0.256m

Conclusions

The variables of experiment 2 are:

1:With/Without GCPs.

2:Different Carrier drone:a VTOL drone or a multi-rotor drone

After analysis of the two sets of experimental data, we can draw the following conclusions:

 

When the oblique camera is DG4Pros:

(1) With/Without ground control point (GCP) whether it is a VTOL or a multi-rotor drone, the 3D model built by oblique camera DG4Pros can meet the requirements of relative accuracy error Ds≤10cm(Actually ≤5cm).

(2) When the carrier drone types is the single variable, the relative accuracy of the 3D model with multi-rotor  is better than that with VTOL,but the difference is not very big.