plane normal: is a deviation from the horizontal. Building roofs are mostly flat or gently sloped.
co-planarity: building has smooth planes, while vegetation leads to randomly distributed planes.
We used 0.6 and 0.4 as the weights of plane normal and co-planarity. Figure 2(f) depicts the segments classified into building. Their member points are assigned as building (Figure 2(h)).
(a)
(b)
(c)
(d)
3.2.1 Pylon detection: the line segments (Figure 3(a)) extracted in section 3.1.1 are converted into a binary image (Figure 3(b)). Each pixel (1m×1m) possesses line segments passing through the pixel. 2D lines in the pylon pixels are mostly perpendicular to ones in PL pixels as shown in figure 3(a). Based on the line orientation comparison, we investigate two cues to identify pylon:
maximum orientation variation (MOV): is a maximum variation of orientations analyzed from all pairs of lines in a pixel. It is approximately 90 degrees for pylon, while 0 degree for PL (Figure 3(c) is a normalized image from [0, 90] to [0, 1]).
orientation parallelism (OP): is a mean difference between the orientation of lines in current pixel and ones in neighbouring pixels. This feature is robust against PL lines having incorrect orientation thanks to smoothness effect. This feature shows similar distribution as maximum orientation variation (Figure 3(d)).
We employ Random Forests (RF) for the binary classification (pylon vs. non-pylon). RF combines a number of decision trees which are generated by learning different training sets (Breiman, 2001). Based on the two cues (MOV and OP), RF classifier is trained. Figure 3(e) presents RF classification results. Pylon segments classified could be split into two or more segments. To obtain connected pylon regions, morphology filters (dilation and erosion) are applied. Figure 3(f) shows a final pylon detection result.
(e)
(f)
(a)
(b)
(c)
(g) (h)
Figure 2. PL scene classification; (a) optical image, (b) linear feature extraction (red), (c) line segments, (d) PL (blue) vs. non-PL (black), (e) plane segments, (f) building (blue) vs. non-building (black), (g) ground truth, and (h) PL scene classification result.
3.2 Pylon localization
The pylon localization aims to identify linear segments associated with pylons and validate them by investigating contextual relations amongst extracted pylon regions. For instance, if two true pylon regions are connected each other, we should observe PLs between them and the orientation of the line connecting them equal to PL’s directions. We utilize these properties to progressively eliminate false pylon regions. This idea is implemented in a 2D graph analysis.
(d)
(e) (f)
Figure 3. Pylon segmentation; (a) line segments (blue boxes: pylons), (b) binary image from (a), (c) maximum orientation variation [0,1], (d) orientation parallelism [0,1], (e) pylon classification (RF), and (f) pylon segments.
3.2.2 Pylon validation: The pylon detection might produce false positive errors, which need to be pruned. There are ten pylons in our study area as shown in figure 4(a), however our pylon detection method over-detected them (Figure 4(b)). The pylon validation is designed for eliminating non-pylon segment (NPS), while keeping pylon segment (PS) by examining their contextual properties. First, we make all possible connections (edges) among all segments if the edge length is less than














a
英语 ISPRS journal of photogrammetry and remote sensing 国际摄影测量和遥感学会... Mat 英语 Annals of geophysics 地球物理学纪事 英语 英语 Minerals & ...
英语 ISPRS journal of photogrammetry and remote 国际摄影测量和遥感学会志英语 ...(英文版) Technology Beijing, Mineral, 英语 Metallurgy, Mat Annals of ...
英语 ISPRS journal of photogrammetry and remote sensing 国际摄影测量和遥感学会... Mat 英语 Annals of geophysics 地球物理学纪事 英语 英语 Minerals & ...
ISPRS Journal of Photogrammetry and Remote Sensing 64 (2009) 472–481 ...Annals IX Simposio Brasileiro de Sensoriamento Remoto. INPE. Santos, Brazil....
ANNALS OF MATHEMATICS ANNALS OF THE NEW YORK ...ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING ...167 168 169 JOURNAL OF PROTEOME RESEARCH JOURNAL ...
Isprs journal of photogrammetry and remote sensing ...annals of geophysics 0.413 pure and applied ...文档贡献者 kj99102hsl 贡献于2012-02-28 ...
2012年SCI杂志中科院分区表(完整版)_其它_高等教育...ISPRS J PHOTOGRAMM IEEE POWER ENERGY M CEMENT ...ANNALS OF BIOMED APPLIED SOFT INFORMATION & ...
ISPRS J PHOTOGRAMM ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING 1.6740...ANNALS OF GLACIOLOGY 地学 0260-3055 4 0.8020 2012 B SOC GEOL FR ...
ISPRS JOURNAL OF ISPRS J PHOTOGRAMMETRY AND REMOTE...ANNALS OF GLACIOLOGY ANN GLACIOL AUST J EARTH ...2012 0.9837 1546 430 挪威 400 荷兰 347 法国 ...
ISPRS J PHOTOGRAMM ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING 0924-...ANNALS OF THE NEW YORK ACADEMY OF SCIENCES 0077-8923 PHILOS PHILOSOPHICAL T...
热门文档