De-noising Processing And Abnormal Extraction Of Geochemical Data Based On Wavelet Transform

Geochemical data processing occupies an important position in the process of geochemical prospecting, it based on the theory of mathematics and exploration geochemistry. The aims are to differentiate the geochemical background and anomaly, and extract the geochemical information related to the ore-forming from large amounts of data. In this process, the noise suppression of geochemical data is particularly important, the "noise" of the geochemical data comes from a series of superposition error in the process of data processing, such as sampling, processing, preprocessing and analysis. Therefore, thorough understanding the types and characteristics of geochemical data noise, the researchers can use correct method to remove noise effectively, and identify anomaly threshold reasonably, so as to extract the abnormal.Conventional method for determining geochemical anomaly threshold applies to geochemical data conforming to normal distribution or lognormal distribution. Content distribution of geochemical elements, in fact, is not confined to the two distribution pattern, and is highly nonlinear, such as respectively, the randomness and irregularity. These nonlinear characteristics often lead to the complexity of geochemical data processing, and the traditional geochemical data processing method is powerless in terms of fine depict the geochemical element space distribution. Therefore, the study of nonlinear method is crucial for the geochemical data processing. Wavelet analysis is a kind of multiscale analysis method of time-frequency localization and using the multi-resolution property of wavelet analysis, the geochemical data can be broken down into several different components, respectively corresponding to different levels of abnormal. Its local analysis ability is very strong, and can delineate abnormal range accurately.As a new time-frequency analysis tool, the theory and practical technology of wavelet analysis have been mature in terms of the signal and image processing, and wavelet analysis is relatively less in the geochemistry, but its algorithm thought conforms to the space distribution features of geochemical data. This paper, from de-noising processing of the geochemical data, based on the theory of wavelet analysis, researchse and developments an effective data processing method to remove noise, and keep the characteristics of the original data under the condition of any unknown noise information of raw data. This methods model will be used in the1:200000stream sediment survey data of Tonglushan Copper-iron (Cu-Fe) deposit. From the result of anomaly analysis, it is proved that the effect is good and the model is reliable.In this paper, two-dimensional wavelet transform is used in geochemical data processing, which in the process of wavelet threshold denoising, involve four key factors:the reasonable determination of the optimum wavelet base and decomposed layers, and the selection of threshold function and the quantization threshold. Calculation of general threshold is simple, but it needs to understand the data itself, and the real data in practical application is unknown. For this purpose, the paper studies emphatically geochemical data processing method about wavelet threshold, which based on generalized cross validation (GCV) criterion. This method needs not obtain any information about noise in advance, namely, it don’t need to estimate the variance of noise. Through searching and only depending on the input data, it can directly get the asymptotic optimal threshold, and denoise at the same time better keep the details features of original signal. It is a kind of method to evaluate smoothing degree of the parameters.In a word, this paper researches the de-noising processing and abnormal extraction of geochemical data based on wavelet analysis in order to delineate abnormal area effectively and provide a new technical method about geochemical data processing and abnormal analysis.