Automate Green-Coverage Measure Based On Remote Sensing Images And Digital Images Of Ali In Tibet

Green coverage is an important indicator of health status of rangeland, and it's meaningful for improving the efficiency and accuracy of green coverage measurement. Currently, the measuring methods can be classified into two categories including remote sensing monitoring method and ground test method. Digital image analysis technique becomes the most economic-efficient method; however, there have been too much artificial intervention and passivity in automatic measurement methods. In this thesis, we automate rangeland green coverage measure, a novel threshold optimization algorithm based on mean gradient magnitude of edge pixels, which can be more objective and faster to measure green coverage.The main contents of this thesis are as followings:(1)In this thesis samples were taken and images were captured by digital camera, and proper vegetation index was chosen (Excess Green in this thesis) according to vegetation spectral features in digital images. Then green coverage were measured, using continuum removal algorithm and the upper-inflexion-point thresholding algorithm based on the M-Et curve (mean gradient magnitude of edge pixels vs Excess Green threshold).(2)Respectively, green-coverage was measured by automating green-coverage measurement method, Maximum likelihood method, Visual interpreting method and color-saturation method. The result from linear regressive analysis of the four sets of data (the data from Visual interpreting method is assumed to be true value) showed that the accuracy of automating green-coverage measurement method could be compared to that of maximum likelihood method, and it had no problem of underestimating to high green coverage with few artificial intervention, more accurate results and stronger applicability. However, its effect was not nice on plants without obvious green feature (glaucous plants as an example).(3)Green-coverge inversion model was derived by linear inversion between NDVI and green-coverage of quadrates in study area. After that, regional green-converge grade figure was made.