CN-122024890-A - Rock and mineral identification method and device based on spectrum multi-index feature coupling

CN122024890ACN 122024890 ACN122024890 ACN 122024890ACN-122024890-A

Abstract

The application provides a rock and mineral identification method and device based on spectrum multi-index feature coupling. The method comprises the steps of preprocessing a rock and mineral spectrum to be detected to obtain a target spectrum, carrying out global spectrum matching identification on the target spectrum by taking a preset reference spectrum as a reference to obtain a global spectrum factor based on form matching, carrying out local absorption characteristic identification on the target spectrum to obtain a local absorption characteristic vector matching factor, coupling the global spectrum factor and the local absorption characteristic vector matching factor based on preset weights to obtain comprehensive similarity of global and local coupling matching, and determining the rock and mineral corresponding to the reference spectrum with the highest comprehensive similarity as the target rock and mineral. By the method, the overall robustness and the local sensitivity can be achieved by carrying out overall spectrum recognition and local absorption feature recognition on the target spectrum and coupling the overall spectrum factor and the local absorption feature vector matching factor to achieve overall similarity of overall and local coupling matching, and then the accuracy of rock and ore recognition is improved.

Inventors

Bu Jiancai
Xiao Songchun
XIAO YUNYUN
ZHANG LIJUN
XU ZHIBIN
GAO YONG
CHEN YONG
LUO WENXIANG

Assignees

湖南省遥感地质调查监测所

Dates

Publication Date: 20260512
Application Date: 20251230

Claims (9)

1. The rock and mineral identification method based on the spectrum multi-index feature coupling is characterized by comprising the following steps of: Acquiring a rock ore spectrum to be detected, and preprocessing the rock ore spectrum to be detected to obtain a target spectrum; Performing global spectrum matching identification on the target spectrum by taking a preset reference spectrum as a reference to obtain a global spectrum factor based on form matching; Carrying out local absorption feature recognition on the target spectrum by taking a preset reference spectrum as a reference to obtain a local absorption feature vector matching factor; coupling the global spectrum factor based on morphological matching and the local absorption feature vector matching factor based on preset weight to obtain comprehensive similarity of global and local coupling matching; and determining the rock ore corresponding to the reference spectrum with the highest comprehensive similarity as a target rock ore.
2. The method for identifying rock and mineral according to claim 1, wherein the preprocessing includes at least a spectrum resampling process, a noise suppression process, a continuum removal process, a normalization process, and a spectrum effective interval screening process.
3. The rock and mineral identification method according to claim 1, wherein the performing global spectrum matching identification on the target spectrum with the preset reference spectrum as a reference to obtain a morphology matching-based global spectrum factor comprises: Acquiring a key characteristic wave band interval and a common wave band interval of the target spectrum; Respectively configuring weights for the key characteristic wave band interval and the common wave band interval, wherein the weight of the key characteristic wave band interval is larger than that of the common wave band interval; respectively carrying out weighting treatment on the total spectral reflectance of the target spectrum and the preset reference spectrum according to the weight of the key characteristic wave band interval and the weight of the common wave band interval to obtain the segmented weighted reflectance of the target spectrum and the preset reference spectrum; calculating a segment weighted spectrum matching angle based on the segment weighted reflectivities of the target spectrum and the preset reference spectrum; determining the segment weighted spectrum matching angle as the global spectrum factor based on morphology matching.
4. The rock and mineral identification method according to claim 1, wherein the performing local absorption feature identification on the target spectrum with the preset reference spectrum as a reference to obtain a local absorption feature vector matching factor includes: acquiring an effective band of the target spectrum, and extracting absorption characteristic parameters of the effective band; Constructing a first comprehensive absorption characteristic vector according to the absorption characteristic parameters of the effective band of the target spectrum; Constructing a second comprehensive absorption characteristic vector according to the absorption characteristic parameters of the effective band of the preset reference spectrum; and determining the local absorption feature vector matching factor according to the first comprehensive absorption feature vector and the second comprehensive absorption feature vector, wherein the local absorption feature vector matching factor comprises a basic Euclidean distance or an augmented Euclidean distance.
5. The method for identifying rock and mineral according to claim 1, wherein the coupling the morphology-based matching global spectral factor and the local absorption feature vector matching factor based on the preset weight to obtain the overall similarity of global and local coupling matching comprises: acquiring a preset global weight and a preset local weight; Determining global similarity according to the global spectrum factor based on morphology matching; Determining local similarity according to the local absorption feature vector matching factor; And coupling the global similarity with the local similarity according to the preset global weight and the preset local weight to obtain the comprehensive similarity of global and local coupling matching.
6. Rock and mineral identification device based on coupling of spectrum multi-index features, characterized in that, rock and mineral identification device includes: the acquisition module is used for acquiring a rock-ore spectrum to be detected, and preprocessing the rock-ore spectrum to be detected to obtain a target spectrum; The global identification module is used for carrying out global spectrum matching identification on the target spectrum by taking a preset reference spectrum as a reference to obtain a global spectrum factor based on form matching; the local recognition module is used for carrying out local absorption characteristic recognition on the target spectrum by taking a preset reference spectrum as a reference to obtain a local absorption characteristic vector matching factor; the coupling module is used for coupling the global spectrum factor based on the form matching and the local absorption characteristic vector matching factor based on the preset weight to obtain the comprehensive similarity of global and local coupling matching; and the determining module is used for determining the rock ore corresponding to the reference spectrum with the highest comprehensive similarity as the target rock ore.
7. An electronic device, comprising: A memory configured to store instructions; a processor configured to invoke the instructions from the memory and when executing the instructions is capable of implementing a method of rock mine identification based on spectral multi-index feature coupling according to any one of claims 1 to 5.
8. A machine-readable storage medium having instructions stored thereon for causing a machine to perform the method of rock mine identification based on spectral multi-index feature coupling of any one of claims 1 to 5.
9. A computer program product, characterized in that instructions in the computer program product, when executed by a processor of an electronic device, cause the electronic device to perform the method of rock-mine identification based on spectral multi-index feature coupling according to any one of claims 1 to 5.

Description

Rock and mineral identification method and device based on spectrum multi-index feature coupling Technical Field The invention relates to the technical field of rock and mineral identification, in particular to a rock and mineral identification method and device based on spectrum multi-index feature coupling. Background The hyperspectral remote sensing and ground rock and mineral spectrum testing technology is widely applied to the fields of geological disaster monitoring, resource investigation, mineral exploration and the like. Most minerals have a relatively stable crystal structure, and their characteristic absorption bands are well diagnosed in the visible-short wave infrared band. However, in the actual geological background, factors such as ore weathering, covering effect, particle size effect, multiple scattering noise, and mixed mineral composition can cause weakening of spectrum absorption band, deviation and even loss of local characteristics, so that the identification method based on single diagnostic absorption peak is unstable and poor in adaptability. On the other hand, spectrum Angle Matching (SAM) or related matching methods based on integral waveforms are robust, but have limited discrimination ability in the face of minerals with similar integral morphology but different local absorption details. The traditional method is not high in recognition accuracy under a complex geological background due to the fact that the traditional method depends on local characteristics and is easily affected by noise, or the traditional method depends on integral waveforms and is insensitive to weak differences. Disclosure of Invention The application aims to provide a rock and mineral identification method and device based on spectrum multi-index feature coupling so as to improve the accuracy of rock and mineral identification. In a first aspect, a method for identifying rock and mineral based on spectral multi-index feature coupling is provided, the method comprising: Acquiring a rock ore spectrum to be detected, and preprocessing the rock ore spectrum to be detected to obtain a target spectrum; Performing global spectrum matching identification on the target spectrum by taking a preset reference spectrum as a reference to obtain a global spectrum factor based on form matching; Carrying out local absorption feature recognition on the target spectrum by taking a preset reference spectrum as a reference to obtain a local absorption feature vector matching factor; coupling the global spectrum factor based on morphological matching and the local absorption feature vector matching factor based on preset weight to obtain comprehensive similarity of global and local coupling matching; and determining the rock ore corresponding to the reference spectrum with the highest comprehensive similarity as a target rock ore. Optionally, the preprocessing at least comprises spectrum resampling processing, noise suppression processing, continuum removal processing, normalization processing and spectrum effective interval screening processing. Optionally, the performing global spectrum matching recognition on the target spectrum with the preset reference spectrum as a reference to obtain a global spectrum factor based on morphology matching includes: Acquiring a key characteristic wave band interval and a common wave band interval of the target spectrum; Respectively configuring weights for the key characteristic wave band interval and the common wave band interval, wherein the weight of the key characteristic wave band interval is larger than that of the common wave band interval; respectively carrying out weighting treatment on the total spectral reflectance of the target spectrum and the preset reference spectrum according to the weight of the key characteristic wave band interval and the weight of the common wave band interval to obtain the segmented weighted reflectance of the target spectrum and the preset reference spectrum; calculating a segment weighted spectrum matching angle based on the segment weighted reflectivities of the target spectrum and the preset reference spectrum; determining the segment weighted spectrum matching angle as the global spectrum factor based on morphology matching. Optionally, the performing local absorption feature recognition on the target spectrum with the preset reference spectrum as a reference to obtain a local absorption feature vector matching factor includes: acquiring an effective band of the target spectrum, and extracting absorption characteristic parameters of the effective band; Constructing a first comprehensive absorption characteristic vector according to the absorption characteristic parameters of the effective band of the target spectrum; Constructing a second comprehensive absorption characteristic vector according to the absorption characteristic parameters of the effective band of the preset reference spectrum; and determining the local absorption feature vector mat