Study On The Optical Spectral Model Of Dense Object And Component Inversion

Remote sensing has become one of the most primary techniques for human to study earth and spatial exploring. The spectrum of certain object can be obtained by remote sensing technique. The property of an object can be recognized by studying its spectrum which is used in many applications including agriculture and other business. To understand the object based on its spectrum, there are two key problems need to be studied: (1) what are the object components and (2) what are the contents of the components.Dense object is a wide range of existence on earth such as soil, deserts, all kinds of granular minerals, and so on. In addition, it is also the major components of the planet's surface. Therefore, it is necessary to study the spectrum and inverse the content of the components. Presently, the study of scattering spectrum is divided into two aspects. One is spectrum of object with continuous surface; in this case, the light is regarded only shine on the surface. The second case is to study the spectrum of discrete media, According to radiative transfer theory, Hapke proposed the hapke model which can be applied to dense granular surface; while the impact of surface shape was ignored in this model. Pure object virtually don't existent in practice, and a linear combination model is often used, but this model only used in ignoring the multiple scattering and effects of mixed land cover of the border situation.Focusing on the two problems: two problems are researched in this paper. Firstly, based on the rough surface scattering theory and Hapke model, a mixture model considering the surface shape of the dense granular was proposed. The fractal is used to simulate rough surface. The fractal shadowing function is acquired by fractal shielding data and particle swarm optimization in the case of vertical observation..Secondly, considering that the problem that obtaining the content of the components only by a single spectral curve is an ill-posed problems, this paper presented a Knowledge-Based iterative model. First, the possible components were obtained by the corresponding prior knowledge of the object; then the model of the spectral mixture problem was linearized; the components and its corresponding contents were obtained by a inversion technique based on PSO.several experimental results show that the excellent performance of the proposed method.