Paddy Of Dry Season Use Information Extraction By Object-oriented Technology Based On Remote Sensing Image

Over the years, With the rapid development of satellite remote sensing, the spatial resolution of remote sensing image is also rising, so large numbers of high-resolution remote sensing images to make remote sensing applied to more areas. The internal spectral differences of remote sensing image are also increasing, with the spatial resolution of the remote sensing image continues to improve, and how to quickly and accurately extraction high-resolution remote sensing image information has become a major research direction of remote sensing image information processing. Traditional remote sensing image extraction information are mostly pixel-based and it only ues the spectral information to extracted information, high-resolution remote sensing images have rich infomation in addition to the spectral information but also have a lot of other information, so we has been unable to meet the high-resolution remote sensing image information extraction requirements if only using the Spectral characteristics of image. Object-oriented information extraction technology under great concern for high-resolution remote sensing image information extraction and it provides a unique value for high-resolution remote sensing image information extraction. The technology is based on the image object and the image bject is the smallest unit of information extraction, not only using the spectral information of the image, but also make full use of spatial information and texture information of the image, etc. We can also use visual interpretation as a basis for information extraction, So it greatly increased the accuracy and efficiency of remote sensing image information extraction.In this paper, base on the learning and research object-oriented image analysis technology, use the first object-oriented software eCogniton as extraction software platform, applied object-oriented analysis and multi-scale segmentation techniques to high resolution remote sensing images extract paddy fields information during the dry season. The main contents and results are as follows:(1) Studied the processing method of the object-oriented high-resolution remote sensing image and presented the extraction processe of paddy use information during the dry season, the problems and difficulties were studied in extraction process.(2) Discussed the multiscale image segmentation methods, and analyzes the impact when the segmentation process parameters on the scale and segmentation results, based on the principle of optimal segmentation scale proposed a scale parameter to calculate the optimal partition optimal segmentation of each land type and scale parameters to the study area.(3) Classification rules were established for each study area to class. After dividing the image into the image combining visual interpretation of the analysis, take advantage of high-resolution remote sensing images rich spatial and spectral characters, select the optimal combination of features in each class using fuzzy classification and classification methods to establish thresholds classification rules for each category of the study area in information extraction.(4) Compared The object-oriented classification method and the traditional pixel-based methods. Article use envi software, using the maximum likelihood method extraction the study area information and comparing the results based on overall classification accuracy,the results show that the classification of object-oriented classification method is based on the object, and the area is continuous, relatively close to the real surface features, and to solve this problem in a pixel-based classification method "salt and pepper phenomenon".(5) The utilization of the dry season paddy fields were analyzed. use the results of object-oriented classification, the study area during the dry season paddy pattern and each table in the class were analyzed.Innovation thesis is:(1) In the segmentation scale choice, based on the "homogeneity of the object image internal is largest and among object image the heterogeneity biggest" principle proposed an optimal segmentation scale parameter to calculate the optimal segmentation scale and parameters to each class.(2) The use of object-oriented methods to extract the vector polygon data, greatly improving the integration of remote sensing and GIS, extraction results are also very convenient for data storage.