The Research Of Dynamic Change Detection Of Forest Cover Using High-resolution Data

Change detection means to extract features of changes over time by comparative analysis with multiple remote sensing images of the same area for a different time. Remote sensing image change detection has been paid more and more attention in many fields such as forest resources investigation. Meanwhile, China launched the GF-1 satellite in 2013, and its acquisition of multi spectral remote sensing image is suitable for forest change detection. Therefore, this paper through a comprehensive analysis and summarization of research achievements about change detection studied the information extraction of forest change information, and selected Bamiantong Forestry Bureau, Mudanjiang City, Heilongjiang Province as the research area, and collected two images of GF-1 in different times within the study region to change detection of the forest with object-oriented method. This paper carries out the following aspects of work:(1) Comprehensive comparative analysis of the advantages and disadvantages of the traditional pixel-based classification method and object-oriented classification method. Based on the inherent defects of the traditional classification methods, apply the objectoriented thinking into the extraction of target information, and elaborates some important parts such as image segmentation, feature extraction and image object classification in the object-oriented classification method. Experimental analysis and instructions are also conducted to solve scale effect problem in object-oriented information extraction.(2) Summarized the existing common change detection methods and analyzed the advantages and disadvantages of various methods.(3) Through the analysis of the experimental results by the traditional pixel based classification and based on the object-oriented classification methods, we can get the conclusion that the object-oriented classification accuracy is higher.(4) Detected the changes of the object-oriented classification results, and the detection result is relatively accurate by the analysis of visual interpretation.