detecting invasive plants using hyperspectral and high resolution satellite images | 量子荧光|微型光纤光谱仪-ag贵宾会

the rapid spread of nonnative plant species have caused considerable negative impact to the biodiver sity and ecosystems in taiwan. to better under s tand the status and to suppor t researcher s and decision makers to develop strategies and remedies for this problem, it is necessary to obtain accurate spatial information and the progres sion about the invasions of foreign species into native ecocommu ni ty. the availability of hyper spect ral and high resolution satellite data provides researcher s an oppor tuni ty to pursue more complex analysis and have a great potential to achieve better performance and results in an invasive plants investigation. high resolution images provide detail spatial informa tion about the target areas but are often limited to single or few spectral bands. on the other hand, hyperspect ral data consist of tens to hundreds of contiguous bands but lack of spatial details. therefore, a combination of both types of data is likely to be an optimal approach to the mapping of alien plants. however, with the large data volume and high data dimensionality, the major challenge of using hyperspect ral and high resolution data together is to extract useful information effectively and efficiently. this paper present s a work in progres s of developing a systematic method to use hyperspect ral and high resolution satellite images to identify an invasive plant (horse tamarind, leucaena leucocephala ) that is spreading in an alarming rate in southern taiwan. the developed method first locates "areas of interes t" where target species is likely to populate most densely. then a two- level analysis procedure is implemented using hyperspect ral and high resolution satellite images to identify and map the distribution of target species. the first phase of the procedure is to analyze hyper spect ral images with selected (helpful) features to obtain a preliminary result. the second phase is to isolate the areas where discrimination of target plant species is not satisfactory and to improve the accuracy of discrimination with the analysis of canopy structure s in high resolution satellite images. verification with ground truth samples indicates that the developed method of combining high resolution and hyperspect ral images analysis is an effective and efficient approach to detect invasive plant s in a large area.