Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques
碩士 === 國立臺灣海洋大學 === 通訊與導航工程學系 === 101 === In recent years, oil pollutions in the sea area around Taiwan, are getting more serious. The main reasons of oil spills come from the ships’ operation or accident. To protect the marine environment and resources in Taiwan, it is essential to detect and monit...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2013
|
Online Access: | http://ndltd.ncl.edu.tw/handle/48086837872209308952 |
id |
ndltd-TW-101NTOU5300039 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-101NTOU53000392015-10-13T23:28:48Z http://ndltd.ncl.edu.tw/handle/48086837872209308952 Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques 應用影像切割技術於SAR影像中油汙染及船舶之自動偵測 Po-Yuan Chen 陳柏源 碩士 國立臺灣海洋大學 通訊與導航工程學系 101 In recent years, oil pollutions in the sea area around Taiwan, are getting more serious. The main reasons of oil spills come from the ships’ operation or accident. To protect the marine environment and resources in Taiwan, it is essential to detect and monitor the oil spills. The operational difficulty in sampling oil spills and the effectiveness which may be driven by ocean current, these factors subjoin the difficulty in monitoring oil spills. Synthetic Aperture Radar (SAR) images have many adavantages including high resolution and wide coverage areas even under unfavorable weather conditions, day and night. SAR image has been shown to be very useful in oil slick monitoring. The thesis provide a region-based automatic detection of oil spills and ships in SAR images by image segmentation technique and detection theory. In SAR image, the signal reflection influence cause multiplicative speckle noise which makes the slick detection more difficult. At first, we relieve the speckle noise by Refined Lee Filter. Then, we apply Moment-Preserving Principle to partition the images into some proper regions according to the statistics and local terrain characteristics of image. Based on the segmentation results which contain ship regions and oil pollution regions, we build the data model for the ships, oil and sea background, respectively. Finally, according to Bayes Decision Rule and Multiple Hypotheses Theory, we derived automatic decision rules for oil and ship identification. Computer simulation results verify the proposed method achieves over 90% detection probability in oil spills and ships even if untrained images. Lena Chang 張麗娜 2013 學位論文 ; thesis 67 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立臺灣海洋大學 === 通訊與導航工程學系 === 101 === In recent years, oil pollutions in the sea area around Taiwan, are getting more serious. The main reasons of oil spills come from the ships’ operation or accident. To protect the marine environment and resources in Taiwan, it is essential to detect and monitor the oil spills. The operational difficulty in sampling oil spills and the effectiveness which may be driven by ocean current, these factors subjoin the difficulty in monitoring oil spills.
Synthetic Aperture Radar (SAR) images have many adavantages including high resolution and wide coverage areas even under unfavorable weather conditions, day and night. SAR image has been shown to be very useful in oil slick monitoring. The thesis provide a region-based automatic detection of oil spills and ships in SAR images by image segmentation technique and detection theory.
In SAR image, the signal reflection influence cause multiplicative speckle noise which makes the slick detection more difficult. At first, we relieve the speckle noise by Refined Lee Filter. Then, we apply Moment-Preserving Principle to partition the images into some proper regions according to the statistics and local terrain characteristics of image. Based on the segmentation results which contain ship regions and oil pollution regions, we build the data model for the ships, oil and sea background, respectively. Finally, according to Bayes Decision Rule and Multiple Hypotheses Theory, we derived automatic decision rules for oil and ship identification. Computer simulation results verify the proposed method achieves over 90% detection probability in oil spills
and ships even if untrained images.
|
author2 |
Lena Chang |
author_facet |
Lena Chang Po-Yuan Chen 陳柏源 |
author |
Po-Yuan Chen 陳柏源 |
spellingShingle |
Po-Yuan Chen 陳柏源 Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques |
author_sort |
Po-Yuan Chen |
title |
Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques |
title_short |
Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques |
title_full |
Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques |
title_fullStr |
Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques |
title_full_unstemmed |
Automatic Detection of Oil Spills and Ships in SAR Images by Segmentation Techniques |
title_sort |
automatic detection of oil spills and ships in sar images by segmentation techniques |
publishDate |
2013 |
url |
http://ndltd.ncl.edu.tw/handle/48086837872209308952 |
work_keys_str_mv |
AT poyuanchen automaticdetectionofoilspillsandshipsinsarimagesbysegmentationtechniques AT chénbǎiyuán automaticdetectionofoilspillsandshipsinsarimagesbysegmentationtechniques AT poyuanchen yīngyòngyǐngxiàngqiègējìshùyúsaryǐngxiàngzhōngyóuwūrǎnjíchuánbózhīzìdòngzhēncè AT chénbǎiyuán yīngyòngyǐngxiàngqiègējìshùyúsaryǐngxiàngzhōngyóuwūrǎnjíchuánbózhīzìdòngzhēncè |
_version_ |
1718085597079273472 |