| With the coming of the21th century, China has been experiencing the urbanization transformation, which is characterized by information technology application to industrialization and metropolitanization. The rapid urban expansion has constantly transferred the plough, forests, and water bodies into built up area, which has not only created huge social wealth, but also brought a series ecological and environmental problems meanwhile. It's thus significant to study the sustainable urban development in terms of the process, the mechanism, and the influences of the urban land use changes on human social economy and environment. In the past decade, researchers in different disciplines have carried out a lot of studies, focusing on where, when, how, and why the LUCC happens. The birth and development of Land Change Science has been greatly promoted by the International Human Dimensions Programme (IHDP), International Geosphere-Biosphere Programme (IGBP), and Global Land Project (GLP).Remote Sensing has been acknowledged as a promising and feasible method in LUCC research. Dynamic monitoring on land use has become a major part in remote sensing applications since the launching of the first Landsat satellite. Due to the wide usage of remote sensing technology, the methods of detection of land use change have achieved great improvements both in accuracy and efficiency.The main contents of this paper are as follows:1. The basic theories and methods of remote sensing in monitoring of LUCC at home and abroad were firstly reviewed, which determined largely the research framework. The detection methods such as pixel-based image, the object-oriented detection were summarized, followed by the exploration of the structure-based extraction technology, and object-oriented extraction method.2. The key technologies of urban land use dynamic monitoring and information extraction based on high-resolution remote sensing images were discussed, which led to the method choices and perspective solutions. These key technologies include object-oriented remote sensing image preprocessing, object-oriented character extraction, object creating method, object-oriented image segmentation, classification, and matching. Image change detecting methods were investigated, such as algebra operation, changing comparison, and classifying accuracy comparison. Furthermore, the multi-scale, spectral signature, shape feature, and textural feature in image extraction were analyzed.3. A new remote sensing image change detection method integrating spectrum, textural, and structure was proposed and realized. It was verified by a case study of multiple geo-object change detection which showed the effectiveness of the proposed algorithm.4. Single time and multi-temporal remote sensing change extraction methods were expounded, respectively. According to the current land use classes, the object-oriented multi-level features of land use, spatial texture, and structure feature were analyzed. Based on these features, the paper segmented the high resolution remote sensing images at different scales, and proposed the optimal segment scales and classifying rules for typical spatial features. A case study of the suburban areas of Wuhan indicated that the object-oriented extraction method can achieve a higher accuracy.5. The changing detection experiments in various areas, such as suburban area of Wuhan, scenic-resident mixed areas, high density areas, and water-mountain mixed areas were carried out. The experimental data sources include natural color aerial image with0.6m resolution,4bands QuickBird image with0.6m resolution,4bands IKONOS image with lm resolution,3bands combined SPOTS image with2.5m resolution. The result of multiple geo-object changing detection verified the effectiveness of the object-oriented remote sensing image changing detection considering the factors of mixed spectrum, spatial texture, and structure feature.6. Taking high density built-up area and suburban area of Wuhan city as study cases, we extracted the urban land in object level and compared the result with that of pixel-oriented image extraction method. The result indicated the high accuracy of the object-oriented method proposed in this paper.The results and innovations of this paper were as follows: 1. The object-oriented extraction in land use change information, the basic idea and algorithm of multi-scale segmentation of remote sensing images were insightfully studied. A new remote sensing image change detection method integrating spectrum, textural, and structure was raised and implemented, which was used in typical spatial feature extraction based on high resolution remote sensing image and adoption of the multiple-image segment method.2. We segmented the high resolution remote sensing images at different scales, and proposed the optimal segment scales and classifying rules for typical spatial features, which have proved successful to solve the problem in image extraction for typical urban land.3. Taking Wuhan as an example, the change detection and urban land extraction at object level of remote sensing images from various data sources were carried out, adapting object-oriented change detection method of remote sensing image considering the factors of mixed spectrum, spatial texture, and structure feature. And the results of experiments verified the correctness and validity of the algorithm propose in this paper. |
PDF Full Text Request