Feature Recognition of Natural Gamma-Ray Spectral Data Based on K-L Transformation

The natural gamma-ray spectroscopy observation technique is an important nuclear geophysical detection method and is one of the main tools for solving problems related to Earth and environmental sciences. The traditional natural gamma-ray spectrum observation technique mainly focuses on the analysis and interpretation of a small number of parameters (U, Th, K) and basically does not require feature compression processing, but there are certain limitations. Generally natural gamma ray spectra are measured in a very broad range, for displaying physical properties, geological bodies, geotechnical and other ray spectra, the amount of data is very large, especially for low-energy spectral data, very complex, and the need for ray spectra to be identified by the characteristics of the ray spectra. In this paper, firstly, we obtain the raw data output from the radionuclide instrument, and secondly, we use matlab software to sample the raw data in the range of 0–3 MeV, the ray spectrum interval is more flexible, and the obtained data can be used in the form of ‘uniform sampling- mean output’, and then the output results are characterised by the K-L transform to identify the features of the output results, and extract those values with the largest eigenvalues to characterise its statistical features. The obtained data can be in the form of ‘uniform sampling-mean output’, and the ray spectrum is characterised by a number of output results obtained above, and finally, the K-L transform is used to identify the features of the output results, and the values with the largest eigenvalues are extracted to characterise its statistical features. Thus, it can analyse the difference between the rock and soil more accurately, and provide more accurate and faster methods and techniques for geological exploration, resource exploration, environmental monitoring and other fields.