Sparse Representation Model And Its Application In Geochemical Data Denoising

In the process of collection and post-processing,soil geochemical data may be subject to human activities,processing equipment,and the data itself is relatively low in the soil,which causes the data to contain noise.If the data is directly processed on this basis,to reasonably distinguish the geochemical background and anomalies,there will be certain errors,which will affect the judgment of the relevant geological researchers.Therefore,it is especially important to reduce the noise of geochemical data.However,the useful information of soil geochemical data generally exists in the anomaly area.The general noise reduction algorithm tends to consider the data of the abnormal area as noise.For the meaningful noise reduction of soil geochemical data,it should combine the geological characteristics,namely spatial structure and locality,and ensure the completeness of its useful information.This has always been a difficult point in current research.The sparse representation method has been studied for a relatively long time in the field of signal and image.The theoretical and practical application techniques are mature,but they are applied less in geological data,especially in geochemical data.However,the idea of sparse representation method is very consistent with the spatial structural features and local features of geochemical data.Therefore,based on the sparse representation,we introduce a data processing method to aim to reduce noise of geochemical data.Which can effectively remove or attenuate noise,meanwhile,de-noising data can ensure the spatial structure characteristics of the original data and the local texture.At first,we introduce the most common models and the most classic algorithms about sparse representation.And then we attach importance to the methods and model suitable for the denoising of soil geochemical data.According to the characteristics of soil geochemical data,we established a denoising model conforming to the soil geochemical data.In the process of data fusion of data blocks segmented by the sliding window,we get a set of weights based on the SSIM value between the offsettarget recombination blocks and the non-offset target blocks.Finally,we apply the constructed model to the soil geochemical data denoising in a certain area of Tibet.At the same time,we use collected data as a reference group,the data denoised by the wavelet algorithm Comparison group.Via the same method to divide abnormal lower limitation,We obtain three the abnormal region maps corresponding to three sets of data.By comparing and analyzing the results of the three sets of experiments,we find that the abnormal region map obtained by these three methods are almost consistent,but the anomaly region using sparse-denoising data is more reasonable than the other two methods,especially the texture feature at the junction of the anomalous area and the background area.In summary,we first apply the sparse representation method as a new method to the denoising of soil geochemical data and successfully extended to the field of soil geochemistry for the sparse representation.