Seamline determination based on semantic segmentation for aerial image mosaicking

Shunta Saito, Ryota Arai, Yoshimitsu Aoki

Research output: Contribution to journalArticle

9 Citations (Scopus)

Abstract

We propose a novel method for seamline determination based on semantic segmentation for aerial image mosaicking. First, we train a convolutional neural network (CNN) for pixel labeling to extract building regions. Using the trained CNN, we create a building probability map from an input aerial image with no pre-processing. We then use Dijkstra's algorithm to find the optimal seamline as a shortest path on the map. We evaluate the quality of the seamlines produced by our method on actual aerial images. Finally, we show that our seamlines never pass through any buildings and compare the effectiveness with the conventional mean-shift segmentation-based method.

Original languageEnglish
Article number7355281
Pages (from-to)2847-2856
Number of pages10
JournalIEEE Access
Volume3
DOIs
Publication statusPublished - 2015 Dec 17

Fingerprint

Semantics
Antennas
Neural networks
Labeling
Pixels
Processing

Keywords

  • Artificial neural networks
  • Computer vision
  • Image processing
  • Neural networks
  • Remote sensing

ASJC Scopus subject areas

  • Computer Science(all)
  • Engineering(all)
  • Materials Science(all)

Cite this

Seamline determination based on semantic segmentation for aerial image mosaicking. / Saito, Shunta; Arai, Ryota; Aoki, Yoshimitsu.

In: IEEE Access, Vol. 3, 7355281, 17.12.2015, p. 2847-2856.

Research output: Contribution to journalArticle

@article{91bb70bcd2f2453bbc48fd179c531da2,
title = "Seamline determination based on semantic segmentation for aerial image mosaicking",
abstract = "We propose a novel method for seamline determination based on semantic segmentation for aerial image mosaicking. First, we train a convolutional neural network (CNN) for pixel labeling to extract building regions. Using the trained CNN, we create a building probability map from an input aerial image with no pre-processing. We then use Dijkstra's algorithm to find the optimal seamline as a shortest path on the map. We evaluate the quality of the seamlines produced by our method on actual aerial images. Finally, we show that our seamlines never pass through any buildings and compare the effectiveness with the conventional mean-shift segmentation-based method.",
keywords = "Artificial neural networks, Computer vision, Image processing, Neural networks, Remote sensing",
author = "Shunta Saito and Ryota Arai and Yoshimitsu Aoki",
year = "2015",
month = "12",
day = "17",
doi = "10.1109/ACCESS.2015.2508921",
language = "English",
volume = "3",
pages = "2847--2856",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Seamline determination based on semantic segmentation for aerial image mosaicking

AU - Saito, Shunta

AU - Arai, Ryota

AU - Aoki, Yoshimitsu

PY - 2015/12/17

Y1 - 2015/12/17

N2 - We propose a novel method for seamline determination based on semantic segmentation for aerial image mosaicking. First, we train a convolutional neural network (CNN) for pixel labeling to extract building regions. Using the trained CNN, we create a building probability map from an input aerial image with no pre-processing. We then use Dijkstra's algorithm to find the optimal seamline as a shortest path on the map. We evaluate the quality of the seamlines produced by our method on actual aerial images. Finally, we show that our seamlines never pass through any buildings and compare the effectiveness with the conventional mean-shift segmentation-based method.

AB - We propose a novel method for seamline determination based on semantic segmentation for aerial image mosaicking. First, we train a convolutional neural network (CNN) for pixel labeling to extract building regions. Using the trained CNN, we create a building probability map from an input aerial image with no pre-processing. We then use Dijkstra's algorithm to find the optimal seamline as a shortest path on the map. We evaluate the quality of the seamlines produced by our method on actual aerial images. Finally, we show that our seamlines never pass through any buildings and compare the effectiveness with the conventional mean-shift segmentation-based method.

KW - Artificial neural networks

KW - Computer vision

KW - Image processing

KW - Neural networks

KW - Remote sensing

UR - http://www.scopus.com/inward/record.url?scp=84961081498&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84961081498&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2015.2508921

DO - 10.1109/ACCESS.2015.2508921

M3 - Article

AN - SCOPUS:84961081498

VL - 3

SP - 2847

EP - 2856

JO - IEEE Access

JF - IEEE Access

SN - 2169-3536

M1 - 7355281

ER -