FAST-ICA Model And Application Of Weak-information Separation From Contaminative Soil

Soil pollution has been posed a major threat to china's food security and sustainable development. In order to effectively curb the increasingly serious environmental pollution, formulate appropriate strategies, the real-time, accurate, and effective monitoring data is necessary. Hyper-spectra spectrum could analysis crop changes in biochemical parameters at the stress of the pollution from the spectroscopy and it provides strong support for pollution prevention and control decision-making.In this paper, the method of Fast-ICA model has been used to analysis crop changes on biochemical parameters at the stress of soil pollution in order to separate and extract the remote sensing weak signal. Independent component generated by independent component analysis is a simulation of independent source signal. So, the components which have the most relevant relationship with the chlorophyll generated by the independent component analysis is bound to bring crop coercion information.To test the validity of the independent component for the separation of weak information, a model experiment of independent component information separation has been done with the high-spectral data collected on the Northeast Plain, North China Plain, the Yangtze River Plain which is the major grain-producing areas in three ways: for the same crops at different growth stages, the same growth region for different crops, the same crops in different regions of the weak model. In each experimental mode, the 425nm, 670nm as the center of the most significant absorption band of chlorophyll has been selected to do the independent component analysis. Respectively, in view of the disorder of the independent component generated by independent component, correlation analysis is implemented between samples generated by a number of independent components and chlorophyll content of field collection.The experimental results show that data in each group will have almost one or two independent components which is highly relevant to the chlorophyll content, and the component represents the contribution of the crop chlorophyll to the vegetation spectral characteristics. If the crop suffered contaminated stress, this component will bring the damage information of chlorophyll.To isolate the pollution weak information carried by the independent component, combined with red edge detection information, an independent component-red edge feature space has been established, for each sample, the red edge position is calculated, for each group of data, take the most concentrated red edge location as the characteristics red edge band, the calculation concentrate on the characteristics red edge of each sample, The ratio of independent component between 670nm band and 425nm band is the health factor, Combined with various kinds of soil heavy metals contents in laboratory analyzed, results show that the health factor of non-heavy metal stress is less than 1,and if the vegetation suffer heavy-metal contamination stress, the health factor is greater than 1, and the severity of pollution is directly proportional to the health factor. At the same time, independent component model of weak information extraction has relative stability for the changes of vegetation at the growth stage.