海島海岸帶地物提取及動(dòng)態(tài)變化分析研究
發(fā)布時(shí)間：2018-07-31 09:08
【摘要】：海島海岸帶區(qū)域資源豐富、經(jīng)濟(jì)發(fā)達(dá),由于海域開發(fā)、城市擴(kuò)展嚴(yán)重導(dǎo)致海島海岸帶土地利用變化較快。因此,及時(shí)、準(zhǔn)確地掌握海島海岸帶資源分布,對(duì)其開發(fā)與利用、可持續(xù)發(fā)展具有重要意義。遙感技術(shù)由于其同步觀測(cè)區(qū)域廣、數(shù)據(jù)的客觀性和可比性、現(xiàn)勢(shì)性高、經(jīng)濟(jì)效益與社會(huì)效益較高的特性,已經(jīng)廣泛地應(yīng)用于海岸帶資源監(jiān)測(cè)領(lǐng)域。研究如何利用遙感影像高精度地提取海島海岸帶地物信息成為遙感工作者關(guān)注的熱點(diǎn)之一,也是本文的出發(fā)點(diǎn)。本文以青島市紅島的三期SPOT5衛(wèi)星影像為數(shù)據(jù)源,以SRTM3及黃島港潮高數(shù)據(jù)為輔助數(shù)據(jù),按照面向?qū)ο蟮幕谥R(shí)的分類方法高精度地提取地物信息并進(jìn)行動(dòng)態(tài)變化分析研究。主要的研究工作如下:(1)在1:3000比例尺下矢量化獲取三期遙感影像的道路信息,并將其作為專題信息參與多尺度分割,從而提高影像分割結(jié)果的實(shí)用性。(2)采用面向?qū)ο蟮亩喑叨确指罘椒ㄍ瓿蛇b感影像的最優(yōu)分割。本文對(duì)多尺度分割中各參數(shù)(形狀權(quán)重、緊致度、尺度參數(shù))進(jìn)行設(shè)置,得到了適宜于本次研究的參數(shù),即在形狀權(quán)重0.1、緊致度0.5,加入道路專題信息的前提下,建立三級(jí)分割網(wǎng):道路專題信息層、尺度500分割層、尺度120分割層,其中:于道路專題信息層提取道路,于尺度500的分割層提取海水,于尺度120的分割層提取植被、濕地、建筑物等地物。(3)基于知識(shí)的分類完成遙感影像地物分類。通過(guò)一系列的實(shí)驗(yàn)研究,建立各類地物的知識(shí):采用Density提取道路,GLCM Homogeneity、MNDWI提取海水,MNDVI、MNDWI提取濕地,NDVI提取植被,GLCM Dissimilarity、Mean SNIR提取建筑物。實(shí)驗(yàn)方法的地物分類精度高,分類結(jié)果與目視判讀相一致,三期影像分類圖的總體精度最高達(dá)97.80%,最低達(dá)到92.80%;卡帕系數(shù)最高達(dá)到0.970,最低達(dá)到0.900。(4)結(jié)合潮高數(shù)據(jù)從水邊線推算海岸線。從面向?qū)ο蟮幕谥R(shí)的分類圖獲取遙感影像的水邊線,其邊界連續(xù),能夠很好地劃分濕地與海水;通過(guò)查閱黃島港的潮高數(shù)據(jù),得出砂質(zhì)海岸的海岸線為與海水鄰接的養(yǎng)殖池塘的邊界,而基巖海岸的海岸線為其水邊線。(5)采用多種動(dòng)態(tài)變化分析方法,如總量分析、變化率分析、轉(zhuǎn)移矩陣分析,對(duì)海岸線及三期遙感影像分類圖進(jìn)行動(dòng)態(tài)變化分析,生成動(dòng)態(tài)變化分析圖表。結(jié)果表明海岸線總長(zhǎng)度增加,其中:2007年比2003年增加727.9m,2011年比2007年增加1957.79m,但是,從2003年至2011年砂質(zhì)海岸的岸線長(zhǎng)度在不斷減少而基巖海岸的岸線長(zhǎng)度在不斷增加;道路、在建區(qū)面積不斷上升而濕地面積不斷下降,濕地主要轉(zhuǎn)換為建筑物及其他地物,是人為因素造成的;土地綜合利用變化率高,其中2003至2007年土地綜合利用變化率達(dá)到1.44%,而2007至2011年土地綜合利用變化率達(dá)到1.95%。
[Abstract]:The coastal zone of the island is rich in resources and developed in economy. As a result of the development of the sea area, the urban expansion seriously causes the rapid change of land use in the coastal zone of the island. Therefore, it is of great significance to grasp the distribution of island coastal resources timely and accurately for its development and utilization and sustainable development. Remote sensing technology has been widely used in coastal resource monitoring field because of its wide synchronous observation area, objectivity and comparability of data, high present situation and high economic and social benefits. How to use the remote sensing image to extract the island coastal features information with high precision has become one of the hot topics of remote sensing researchers and is the starting point of this paper. In this paper, the third SPOT5 satellite image of Hongdao in Qingdao is taken as the data source, and the data of SRTM3 and the tidal height of Huangdao Port are taken as the auxiliary data. According to the object-oriented classification method based on knowledge, the ground object information is extracted with high precision and the dynamic change analysis is carried out. The main research work is as follows: (1) the road information of the third phase remote sensing image is vectorized at the scale of 1: 3000, and it is used as the thematic information to participate in multi-scale segmentation. In order to improve the practicability of image segmentation results. (2) the object oriented multi-scale segmentation method is used to complete the optimal segmentation of remote sensing images. In this paper, the parameters (shape weight, compactness, scale parameter) in multi-scale segmentation are set, and the parameters suitable for this study are obtained, that is, when the shape weight is 0.1, the compactness is 0.5, and the road thematic information is added, A three-level segmentation network was established: road thematic information layer, scale 500 segmentation layer, scale 120 segmentation layer, in which road was extracted from road thematic information layer, sea water was extracted from scale 500 partition layer, vegetation and wetland were extracted from scale 120 partition layer. (3) the classification of remote sensing images based on knowledge. Through a series of experimental studies, the knowledge of all kinds of ground objects was established: Density was used to extract GLCM homogenicity Density from sea water and wetland to extract vegetation. GLCM similarity and mean SNIR were used to extract buildings. The classification accuracy of the experimental method is high, and the classification results are consistent with visual interpretation. The overall accuracy of the third phase image classification map is 97.80, the lowest is 92.80, the kappa coefficient is 0.970 and the lowest is 0.900. (4) the shoreline is calculated from the water boundary with tidal height data. The waterlines of remote sensing images are obtained from object-oriented knowledge-based classification maps, whose boundaries are continuous and can well distinguish wetlands from sea water; by looking up the tidal height data of Huangdao Port, It is concluded that the shoreline of sandy coast is the boundary of culture pond adjacent to seawater, while the shoreline of bedrock coast is its water boundary. (5) A variety of dynamic change analysis methods, such as total quantity analysis, change rate analysis, transfer matrix analysis, are used. The dynamic change analysis of coastline and three stages remote sensing image classification map is carried out, and the dynamic change analysis chart is generated. The results show that the total length of shoreline has increased, including: 727.9 m more in 2007 than in 2003 and 1957.79 m in 2011 than in 2007. However, from 2003 to 2011, the length of shoreline of sandy coast is decreasing and the length of shoreline of bedrock coast is increasing. The area of the area under construction is rising and the area of the wetland is decreasing. The wetland is mainly converted into buildings and other features, which is caused by human factors, and the rate of change of comprehensive land use is high. From 2003 to 2007, the change rate of comprehensive land use reached 1.44%, while that of 2007 to 2011 reached 1.95%.
【學(xué)位授予單位】：中國(guó)石油大學(xué)(華東)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：P715.7
本文編號(hào)：2155103
[Abstract]:The coastal zone of the island is rich in resources and developed in economy. As a result of the development of the sea area, the urban expansion seriously causes the rapid change of land use in the coastal zone of the island. Therefore, it is of great significance to grasp the distribution of island coastal resources timely and accurately for its development and utilization and sustainable development. Remote sensing technology has been widely used in coastal resource monitoring field because of its wide synchronous observation area, objectivity and comparability of data, high present situation and high economic and social benefits. How to use the remote sensing image to extract the island coastal features information with high precision has become one of the hot topics of remote sensing researchers and is the starting point of this paper. In this paper, the third SPOT5 satellite image of Hongdao in Qingdao is taken as the data source, and the data of SRTM3 and the tidal height of Huangdao Port are taken as the auxiliary data. According to the object-oriented classification method based on knowledge, the ground object information is extracted with high precision and the dynamic change analysis is carried out. The main research work is as follows: (1) the road information of the third phase remote sensing image is vectorized at the scale of 1: 3000, and it is used as the thematic information to participate in multi-scale segmentation. In order to improve the practicability of image segmentation results. (2) the object oriented multi-scale segmentation method is used to complete the optimal segmentation of remote sensing images. In this paper, the parameters (shape weight, compactness, scale parameter) in multi-scale segmentation are set, and the parameters suitable for this study are obtained, that is, when the shape weight is 0.1, the compactness is 0.5, and the road thematic information is added, A three-level segmentation network was established: road thematic information layer, scale 500 segmentation layer, scale 120 segmentation layer, in which road was extracted from road thematic information layer, sea water was extracted from scale 500 partition layer, vegetation and wetland were extracted from scale 120 partition layer. (3) the classification of remote sensing images based on knowledge. Through a series of experimental studies, the knowledge of all kinds of ground objects was established: Density was used to extract GLCM homogenicity Density from sea water and wetland to extract vegetation. GLCM similarity and mean SNIR were used to extract buildings. The classification accuracy of the experimental method is high, and the classification results are consistent with visual interpretation. The overall accuracy of the third phase image classification map is 97.80, the lowest is 92.80, the kappa coefficient is 0.970 and the lowest is 0.900. (4) the shoreline is calculated from the water boundary with tidal height data. The waterlines of remote sensing images are obtained from object-oriented knowledge-based classification maps, whose boundaries are continuous and can well distinguish wetlands from sea water; by looking up the tidal height data of Huangdao Port, It is concluded that the shoreline of sandy coast is the boundary of culture pond adjacent to seawater, while the shoreline of bedrock coast is its water boundary. (5) A variety of dynamic change analysis methods, such as total quantity analysis, change rate analysis, transfer matrix analysis, are used. The dynamic change analysis of coastline and three stages remote sensing image classification map is carried out, and the dynamic change analysis chart is generated. The results show that the total length of shoreline has increased, including: 727.9 m more in 2007 than in 2003 and 1957.79 m in 2011 than in 2007. However, from 2003 to 2011, the length of shoreline of sandy coast is decreasing and the length of shoreline of bedrock coast is increasing. The area of the area under construction is rising and the area of the wetland is decreasing. The wetland is mainly converted into buildings and other features, which is caused by human factors, and the rate of change of comprehensive land use is high. From 2003 to 2007, the change rate of comprehensive land use reached 1.44%, while that of 2007 to 2011 reached 1.95%.
【學(xué)位授予單位】：中國(guó)石油大學(xué)(華東)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：P715.7
【相似文獻(xiàn)】
相關(guān)期刊論文 前9條
1 萬(wàn)峻;李子成;雷坤;;1954—2000年渤海灣典型海岸帶(天津段)景觀空間格局動(dòng)態(tài)變化分析[J];環(huán)境科學(xué)研究;2009年01期
2 梁治中;賈志軍;王林和;尚瑞朝;;桃林口水庫(kù)水源涵養(yǎng)區(qū)土地利用動(dòng)態(tài)變化分析[J];內(nèi)蒙古農(nóng)業(yè)科技;2010年03期
3 李思剛;蔣婷婷;王奕;;江蘇省森林資源二類調(diào)查消長(zhǎng)動(dòng)態(tài)變化分析[J];華東森林經(jīng)理;2011年04期
4 吳運(yùn)軍;張樹文;侯偉;張養(yǎng)貞;包春紅;;近50年來(lái)?yè)狭恿饔蚓用竦�、耕地和沼澤地�?dòng)態(tài)關(guān)系分析[J];資源科學(xué);2006年04期
5 賈松偉;;基于GIS和RS的小江流域景觀格局動(dòng)態(tài)變化分析[J];湖北農(nóng)業(yè)科學(xué);2012年18期
6 孟玲玲;賈世鳳;;疊疊溝林場(chǎng)森林資源動(dòng)態(tài)變化分析[J];現(xiàn)代農(nóng)業(yè)科技;2011年02期
7 何文莉;劉木生;郭秋忠;林聯(lián)盛;蔡海生;;基于遙感數(shù)據(jù)的江西贛州耕地時(shí)空分布與動(dòng)態(tài)變化分析[J];江西科學(xué);2012年06期
8 洪偉;楊武年;;基于遙感和GIS的南充市順慶區(qū)LUCC動(dòng)態(tài)變化分析[J];測(cè)繪;2010年03期
9 ;[J];;年期
相關(guān)會(huì)議論文 前1條
1 馬義娟;王尚義;;汾河上游流域1955-2004年耕地動(dòng)態(tài)變化分析[A];《自然地理學(xué)與生態(tài)建設(shè)》論文集[C];2006年
相關(guān)碩士學(xué)位論文 前2條
1 翟珊珊;海島海岸帶地物提取及動(dòng)態(tài)變化分析研究[D];中國(guó)石油大學(xué)(華東);2014年
2 李秀云;基于RS和GIS的城市動(dòng)態(tài)變化分析研究[D];內(nèi)蒙古師范大學(xué);2009年
，本文編號(hào)：2155103
本文鏈接：http://sikaile.net/kejilunwen/haiyang/2155103.html
最近更新
教材專著
