Study On Extraction Method Of Vegetation Coverage Information

Vegetation coverage is an important information in many global and regional changes monitoring model. In ecosystems it plays a very important role. Obtaining vegetation coverage information timely and rapidly is one of tasks of regional soil conservation and ecological environment construction projects study. To search an appropriate method, this paper develops on the two parts of researches which are digital photos and QuickBird image respectively.(1) Takes Wuqi county for object, gathers digital photos of meadow plot in the field, preprocessing these photos, based on dimidiate pixels model, develops a procedure to recognize spectrum information and discovers the spectrum law of photo pixels, then extracts vegetation coverage information and gets the vegetation coverage. Be compared to visual interpretation one, this result is lower, so it concludes that the result of this method applying to this research area is credible. It can validate the remote sensing result very conveniently and rapidly. The research will provide new way of the base data gathering for soil and water conservation general benefit evaluating in reducing cultivated land to plant forest and grass in Wuqi, and has remarkable impulse on the project implement of reducing cultivated land to plant forest and grass and rehabition of vegetation such forest and meadow in Wuqi.It suggested that the procedure function module of this method can be improved more, so it can deal with more photo process demand. On the other aspect, the vegetation type of this research only refers to meadow, not to arbor and shrubbery and other. In addition, this study is based on Wuqi county of reducing cultivated land to plant forest and grass in loess plateau, the spectrum information law of vegetation and non-vegetation is gained in this area, it needs more sample data for validating and completing whether it's suit for other area or not. So this method requires being perfected and then can imply on extended area.(2) It chooses a part of area in Caijiachuan watershed for study, taking QuickBird image for original data, using object-oriented information extraction software.By lots of trial, it selects some appropriate multi-resolution segmentation parameters to segment the image time after time and layer by layer. It selects many feature using for classification and extraction and sets rational feature value, and so extracts vegetation coverage information of the research area, and discusses the calculate of vegetation fraction, the percentage of shrubbery and arbor forest closure. It indicates that it's feasible using this method to extract vegetation coverage information, which can suffice the precision require in vegetation gathering. This method can provide a reference for vegetation coverage information extracting based on high resolution image, and bring a new vegetation data gathering approach for soil and water conservation in Caijiachuan watershed. It also does a favor on promoting vegetation comeback and environment construction, boosting significant forest construction project, such as the desertification actuality and control impact monitoring and evaluating. It also has a good assist on the work which high resolution image technology plays a science and technology role in national forest produce and practice, and environment construction.Talk as a whole, the national application research of high resolution image isn't mature enough, in addition, vegetation remote sensing is influenced by terrain and environment much and there are plenty of uncertain factor, all of these cause bring great difficulty to high resolution image vegetation coverage information extraction in area of complex terrain and vegetation structure. Therefore, it needs more trial and study whether the method of this study is suit vegetation coverage information extraction in other area of complex terrain and vegetation structure, whether it can use to replace the vegetation resources investigating work in filed.