Building Contour Optimization Method Based on Main Direction

Meimei Pan; Jingxin Chang; Xianjun Gao; Yuanwei Yang; Kang Zhong

doi:10.3788/AOS202242.1828005

Acta Optica Sinica, Volume. 42, Issue 18, 1828005(2022)

Building Contour Optimization Method Based on Main Direction

Meimei Pan¹, Jingxin Chang¹, Xianjun Gao^1、*, Yuanwei Yang^1,2,3, and Kang Zhong¹

Author Affiliations

¹School of Geosciences, Yangtze University, Wuhan 430100, Hubei, China

²Hunan Provincial Key Laboratory of Geo-Information Engineering in Surveying, Mapping and Remote Sensing, Hunan University of Science and Technology, Xiangtan 411201, Hunan, China

³Beijing Key Laboratory of Urban Spatial Information Engineering, Beijing 100045, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(19)

Fig. 1. Flow chart of building contour optimization

Download full size

Fig. 2. Comparison of results before and after pretreatment. (a)(d) Original remote sensing images; (b)(e) results of initial contour extraction; (c)(f) results after pretreatment

Download full size

Fig. 3. Flow chart of contour reconstruction regularization method

Download full size

Fig. 4. Flow chart of extracting best main direction

Download full size

Fig. 5. Schematic diagram of regularization algorithm. (a) Building contour; (b) adjusting angles of lines; (c) simplifying polylines and jointing segments; (d) reconstructed contour

Download full size

Fig. 6. Schematic diagram of depth optimization

Download full size

Fig. 7. Comparison of pretreatment result, regularization result and depth optimization result. (a) Original image; (b) reference data; (c) initial extraction result; (d) final optimization result; (e) contour after pretreatment; (f) regularized contour; (g) contour after homing; (h) optimization effect

Download full size

Fig. 8. Comparison of building contour optimization results of image S1. (a) Image S1; (b) reference data; (c) extraction result;(d) optimization result of method A; (e) optimization result of method B; (f) optimization result of proposed method

Download full size

Fig. 9. Experimental comparison of three optimization methods under different building images. (a) Building image 1; (b) building image 2; (c) building image 3; (d) building image 4; (e) building image 5; (f) building image 6

Download full size

Fig. 10. E_F1 of building optimization results of three methods

Download full size

Fig. 11. E_OA of building optimization results of three methods

Download full size

Fig. 12. Effect of threshold $υ$ on experimental results. (a) $υ$ =16; (b) $υ$ =26; (c) $υ$ =30; (d) $υ$ =39

Download full size

Fig. 13. Effect of threshold $ν$ on experimental results.(a) $ν$ =9; (b) $ν$ =20; (c) $ν$ =40; (d) $ν$ =60

Download full size

Fig. 14. Effect of threshold $t$ on experimental results.（a）t=19;（b）t=2;（c）t=9;（d）t=14

Download full size

Fig. 15. Optimization effect of non-rectangular buildings. (a) (e) Extraction results; (b) (f) optimization results; (c) (g) superposition of optimization result with original image; (d) (h) original images

Download full size

Table 1. Contour regularization reconstruction algorithm

View table

Table 1. Contour regularization reconstruction algorithm

input：an array

P = {p_{0}, p_{1}, \dots, p_{n}}

，which is group of building boundary points after preprocessing

output：a reconstructed building boundary

for $i$ ∈［0，n］do

if $0 ° \leq η < 45 °$ or $135 ° < η \leq 180 °$ then

Add（ $e_{i}$ ）

else

Add（ $v_{i}$ ）

end if

end for

$F$ is equal to simplifying this array（ $F$ ）

for $h$ in len（ $F$ ）do

if $F_{h}$ is vertical to $F_{h + 1}$

Insert（ $p_{c r o s s}$ ）

else

if $C$ is true

Merge $L$

else

Add（ $m_{l i n e}$ ）

end if

end for

return $F$

Table 2. Precision comparison of building contour optimization results of image S1
View table
Table 2. Precision comparison of building contour optimization results of image S1
Method E_CR E_CM E_F1 E_OA
Initial extraction 91.50 93.83 92.65 97.16
Method A 93.22 91.70 92.46 97.14
Method B 86.98 94.24 90.47 96.21
Proposed method 91.48 94.52 92.98 97.27

Table 3. Setting of each threshold

View table

Table 3. Setting of each threshold

Picture	$υ$	$ν$	$t$	Picture	$υ$	$ν$	$t$
Fig. 12（a）	16	9	7	Fig. 13（c）	16	40	7
Fig. 12（b）	26	9	7	Fig. 13（d）	16	60	7
Fig. 12（c）	30	9	7	Fig. 14（a）	11	10	19
Fig. 12（d）	39	9	7	Fig. 14（b）	11	10	2
Fig. 13（a）	16	9	7	Fig. 14（c）	11	10	9
Fig. 13（b）	16	20	7	Fig. 14（d）	11	10	14

Table 4. Time complexity
View table
Table 4. Time complexity
Stage Time complexity Sum
Step1 Step2 Step3 Step4
Total $Ο (n^{2})$
Pretreatment $Ο (n)$ $Ο (n^{2})$ $Ο (n^{2})$
Regularization $Ο (n)$ $Ο (n)$ $Ο (n)$ $Ο (n)$ $Ο (n)$
Depth optimization $Ο (n)$ $Ο (n)$

Tools

Get Citation

Copy Citation Text

Meimei Pan, Jingxin Chang, Xianjun Gao, Yuanwei Yang, Kang Zhong. Building Contour Optimization Method Based on Main Direction[J]. Acta Optica Sinica, 2022, 42(18): 1828005

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites