CN114529827B - Mineral product boundary detection method and system based on artificial intelligence - Google Patents
Mineral product boundary detection method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN114529827B CN114529827B CN202210436413.9A CN202210436413A CN114529827B CN 114529827 B CN114529827 B CN 114529827B CN 202210436413 A CN202210436413 A CN 202210436413A CN 114529827 B CN114529827 B CN 114529827B
- Authority
- CN
- China
- Prior art keywords
- pixel point
- complexity
- scale
- cluster
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the field of mineral boundary detection, in particular to a mineral boundary detection method and system based on artificial intelligence. The method comprises the following steps: the method comprises the steps of obtaining an RGB image of a region to be detected, obtaining a geological complexity vector according to texture complexity and color complexity of each pixel point, obtaining a multi-scale space image according to the RGB image, obtaining cluster clusters corresponding to each scale space image according to the geological complexity vector, constructing a graph structure corresponding to each cluster according to the geological complexity vector of each pixel point in each cluster, obtaining an embedded vector corresponding to each cluster according to the graph structure, constructing a multi-scale undirected graph according to the embedded vector corresponding to each cluster, and inputting the multi-scale undirected graph into a trained neural network to obtain a mineral product region boundary of the region to be detected. The method constructs the multi-scale undirected graph to process the area to be detected, has better anti-fuzzy capability and simultaneously improves the safety of the mineral boundary detection process.
Description
Technical Field
The invention relates to the field of mineral boundary detection, in particular to a mineral boundary detection method and system based on artificial intelligence.
Background
Mineral resources are an important material foundation for human society to live on, and are important guarantees for national safety and economic development. The method comprises the following steps that (1) an opencast mining seam is required to define the peripheral boundary of an ore body and the bottom boundary of the ore body in an opencast stope and is an important basis for determining the boundary of the opencast stope; for the underground mining ore bed, the end part, the upper boundary and the lower boundary of the trend of the ore body and the depth increasing condition of the ore body are required to be controlled. These are the necessary geological data bases for determining the position of the mining wellhead, developing roadways and stope arrangements, etc. The traditional mineral boundary acquisition scheme collects sensor feedback data based on drilling and blasting, and determines the boundary of the mineral seam according to data analysis. Such methods are costly and have low safety.
Disclosure of Invention
In order to solve the problem of low safety of the traditional method for detecting the mineral boundary, the invention aims to provide a mineral boundary detection method and a system based on artificial intelligence, and the adopted technical scheme is as follows:
in a first aspect, the invention provides a mineral product boundary detection method based on artificial intelligence, which comprises the following steps:
acquiring an RGB image of a region to be detected;
obtaining a geological complexity vector of each pixel point according to the texture complexity and the color complexity of each pixel point in the RGB image;
obtaining a multi-scale space image according to the RGB image, and recording pixel points with geological complexity larger than a set complexity threshold as key points according to geological complexity vectors of the pixel points in the multi-scale space image; clustering key points corresponding to the spatial images of all scales to obtain clustering clusters corresponding to the spatial images of all scales;
according to the geological complexity vector of each key point in each cluster, constructing a graph structure corresponding to each cluster, according to the graph structure, obtaining an embedded vector corresponding to each cluster, and according to the embedded vector corresponding to each cluster, constructing a multi-scale undirected graph;
and inputting the multi-scale undirected graph into a trained neural network to obtain the mineral area boundary of the area to be detected, wherein the neural network is used for detecting the boundary of the mineral area.
In a second aspect, the present invention provides an artificial intelligence-based mineral boundary detection system, which includes a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the artificial intelligence-based mineral boundary detection method.
Preferably, the method for calculating the texture complexity and the color complexity of each pixel point in the RGB image includes:
processing each pixel point by adopting LTP ternary coding to obtain an LTP value of each pixel point;
obtaining the texture complexity of each pixel point according to the LTP value of each pixel point;
and acquiring three channel values of the Lab space corresponding to each pixel point, and acquiring the color complexity corresponding to each pixel point according to the three channel values of the Lab space corresponding to each pixel point.
Preferably, the method for obtaining the color complexity corresponding to each pixel point according to the three channel values of the Lab space corresponding to each pixel point comprises:
constructing a color vector of each pixel point according to three channel values of the Lab space corresponding to each pixel point;
calculating a color difference vector of each pixel point and a corresponding neighborhood pixel point according to the color vector of each pixel point;
and summing the color difference vectors of each pixel point and the corresponding neighborhood pixel point to obtain the color variation of each pixel point, and taking the color variation of each pixel point as the color complexity of the corresponding pixel point.
Preferably, the method for obtaining the cluster corresponding to each scale space image according to the geological complexity vector of each pixel point in each scale space image comprises the following steps:
obtaining corresponding key points in the space image of each scale by adopting an ORB algorithm according to the geological complexity feature vector of each pixel;
and clustering corresponding key points in each hierarchy image by adopting a DBSCAN clustering method to obtain a cluster corresponding to each hierarchy image in the multi-scale space.
Preferably, the method for constructing the graph structure corresponding to each cluster according to the geological complexity vector of each key point in each cluster comprises:
obtaining geological complexity vectors of all key points in all the clustering clusters, and calculating cosine similarity between the key points corresponding to all the scale space images;
constructing a graph structure corresponding to each cluster: and taking the key points corresponding to the spatial images of all scales as nodes of the graph structure, taking geological complexity vectors of the key points corresponding to the spatial images of all scales as characteristic values of corresponding nodes in the graph structure, and taking the cosine similarity as an edge weight value between the corresponding nodes in the graph structure.
Preferably, the training method of the neural network includes:
collecting a plurality of geological images as a training set of a neural network, and marking the geological images with corresponding labels, wherein the labels are boundaries of mineral areas;
and training the neural network according to the training set and the labels corresponding to the images in the training set.
The invention has the following beneficial effects: the invention obtains the texture complexity and the color complexity of each pixel point in the RGB image of the area to be detected, constructs the geological complexity vector of each pixel point according to the texture complexity and the color complexity, has higher characteristic accuracy compared with single gray information or color information representation of each pixel point, and has better anti-fuzzy capability compared with single gray information or color information because the monitoring camera has the problem of fuzzy image due to focusing delay in the moving monitoring process, the invention obtains the multi-scale space image of the area to be detected, clusters corresponding to each scale space image are obtained by clustering key points in each scale space image, each cluster in each scale space is converted into an embedded vector, a multi-scale undirected graph is constructed according to the embedded vector corresponding to each cluster, and the multi-scale undirected graph is input into a trained neural network, the boundaries of the mineral area are obtained. The invention constructs the geological complexity vector by combining the texture complexity and the color complexity, can comprehensively characterize the characteristics of the mineral layer, considers the cross correlation among key points in each scale image and the cross correlation among key points in each cluster in a multi-scale space, extracts the characteristics of the multi-scale undirected graph with scale invariance, and improves the accuracy of the mineral boundary of the region to be detected.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a method for detecting a mineral boundary based on artificial intelligence according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an original image of an area to be detected according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of key points of an original image of an area to be detected;
FIG. 4 is a schematic diagram of a key point image obtained by downsampling an original image at intervals;
FIG. 5 is a schematic diagram of an image of the keypoints obtained by downsampling the alternate points of FIG. 4;
fig. 6 is a schematic diagram of a key point image obtained by downsampling the interval points of fig. 5.
Detailed Description
To further illustrate the technical means and effects of the present invention for achieving the predetermined objects, the following detailed description of the method and system for detecting mineral boundaries based on artificial intelligence according to the present invention is provided with the accompanying drawings and the preferred embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the mineral boundary detection method and system based on artificial intelligence in detail with reference to the accompanying drawings.
Mineral product boundary detection method embodiment based on artificial intelligence
In the prior art, data fed back by the sensors are collected based on drilling and blasting, and the data are analyzed to obtain the boundary of the mineral area, so that the safety is low. In order to solve the above problem, this embodiment proposes an artificial intelligence based mineral boundary detection method, as shown in fig. 1, the artificial intelligence based mineral boundary detection method of this embodiment includes the following steps:
step S1, an RGB image of the region to be detected is acquired.
In this embodiment, a monitoring camera is used to perform mobile acquisition on an area to be detected, and a monitoring camera acquires an image of the area to be detected for example to perform subsequent processing, where an original image of the area to be detected is shown in fig. 2, and an area with a lighter color in the image is a mineral resource. And enhancing the edge and texture characteristics of the image of the mineral area by using a self-adaptive edge interpolation algorithm to obtain the video frame of the RGB format of the enhanced area to be detected. Edge adaptive interpolation is a well-known technique and will not be described in detail here.
And step S2, obtaining the geological complexity vector of each pixel point according to the texture complexity and the color complexity of each pixel point in the RGB image.
In this embodiment, an LTP three-value encoding mode is adopted to process each pixel point in an RGB image of a mineral area, so as to obtain an LTP value corresponding to each pixel point. LTP ternary coding modes are well known in the art and will not be described further herein.
For the (x, y) pixel point, calculating the difference between the LTP value of eight pixel points in its neighborhood and the LTP value, in this embodiment, the extension line connecting the pixel point and the pixel point in its neighborhood is used as the coordinate axis, the direction in which the pixel point points to its neighborhood pixel point is specified as the positive direction of the coordinate axis, the absolute value of the difference between the LTP value of the pixel point and the LTP value of its neighborhood pixel point is used as the modular length of the difference vector, if the LTP value of its neighborhood pixel point is greater than the LTP value of the pixel point, the direction of the vector is the same as the positive direction of the coordinate axis; conversely, the direction of the vector is opposite to the coordinate axis forward direction. Finally, eight difference vectors are obtained and are taken as a set and recorded as
And the eight difference vectors reflect the texture variation of the pixel point in eight directions of the neighborhood. Its texture complexity is expressed as:
wherein the content of the first and second substances,
is the sum of the difference vectors of the (x, y) pixel points in the neighborhood direction,
the (x, y) pixel point is at
And (5) a difference vector in the direction of each neighborhood pixel point.
Taking the sum of the LTP value of each pixel point and the difference vector of each pixel point in eight directions of the neighborhood as the texture complexity of the corresponding pixel point, specifically expressed as:
wherein the content of the first and second substances,
the texture complexity of the (x, y) pixel,
LTP value of (x, y) pixel.
In this embodiment, the color change direction is described by using pixel-level information, specifically, a color vector of each pixel is constructed according to three channel values of each pixel in the Lab space, and the color vector of each pixel is represented as:
wherein the content of the first and second substances,
is the color vector of the (x, y) pixel,
is the brightness of the (x, y) pixel,
the (x, y) pixel point is a component from green to red,
the (x, y) pixel component is blue to yellow.
Calculating the difference between the color vector and the color vector of eight adjacent pixel points corresponding to each pixel point to obtain the difference vector of the color vector, wherein the difference vector reflects the Lab color variation of the pixel point in the direction of the corresponding adjacent pixel point, and the Lab color variation of each pixel point in the eight adjacent pixel points is summed to obtain the color complexity of each pixel point, and for the (x, y) pixel point, the color complexity is expressed as
。
The texture complexity comprises geological facies and texture change directions, the facies represent geological texture features, and the texture change directions represent trend of mineral seam textures and information of geological structures; the color complexity characterizes the type of rock formation, i.e., mineral information and lithology. In the embodiment, the texture complexity and the color complexity corresponding to each pixel point are used as geological complexity vectors corresponding to the pixel points, and compared with the characteristics of representing each pixel point by single gray scale information or color information, the content of the geological complexity vectors is richer, and the higher classification precision can be ensured when subsequent classification tasks are performed based on the indexes. The geological complexity vector may be expressed as:
wherein the content of the first and second substances,
and the geological complexity vector is the (x, y) pixel point.
Step S3, obtaining a multi-scale space image according to the RGB image, and marking the pixel points with the geological complexity larger than a set complexity threshold as key points according to the geological complexity vector of each pixel point in each scale space image; and clustering the key points corresponding to the spatial images of each scale to obtain a cluster corresponding to the spatial images of each scale.
Focusing delay may occur in the process of moving monitoring of the monitoring camera, and further image blurring may occur, so that a multi-scale space is constructed in the embodiment to simulate out-of-focus conditions which may occur in video frames. The features extracted in the multi-scale space have scale invariance after fusion, so that compared with the features of a single scale, the multi-scale features have good anti-fuzzy capability, namely anti-noise performance.
Specifically, for each frame of image, a multi-scale space is constructed by using point-spaced downsampling, in this embodiment, the downsampling interval is one pixel point, four scale images of the multi-scale space are obtained, and each scale image has different resolutions so as to simulate the near-large-far-small characteristics of physiological vision.
In this embodiment, an ORB algorithm is adopted, key points corresponding to each scale image are extracted based on a geological complexity vector corresponding to each pixel, the key points represent pixel points with higher geological complexity change of each scale image, that is, pixel points with geological complexity greater than a set complexity threshold are marked as key points, fig. 3 is a schematic diagram of original image key points in a region to be detected, fig. 4 is a schematic diagram of key point images obtained by downsampling original image partition points, fig. 5 is a schematic diagram of key points obtained by downsampling the original image partition points 4, fig. 6 is a schematic diagram of key point images obtained by downsampling the original image partition points 5, and whiter points in the images are key points. The set of key points of each scale image reflects the place of the corresponding scale imageMass complexity information. Clustering corresponding key points in each scale image by adopting a DBSCAN clustering method to obtain cluster clusters corresponding to each level image, wherein the neighborhood radius eps =4 of DBSCAB and the minimum density threshold are set in the embodiment
In specific application, the setting can be carried out according to actual needs. The ORB algorithm and DBSCAN clustering are well known techniques and will not be described herein.
And step S4, constructing a graph structure corresponding to each cluster according to the geological complexity vector of each key point in each cluster, obtaining an embedded vector corresponding to each cluster according to the graph structure, and constructing a multi-scale undirected graph according to the embedded vector corresponding to each cluster.
The discrete key points do not consider the correlation among the key points, the description of the overall geological complexity of the image is not comprehensive, and the topological structure of the undirected graph can effectively represent the cross correlation among all nodes in the undirected graph, so that the graph structure is constructed for each cluster in each scale image to make up for the defects of the discrete point set characteristics.
Specifically, the feature values of all the key points in each cluster in each scale image are taken as a set, i.e. the feature values of all the key points in each cluster in each scale image are taken as a set
Wherein the content of the first and second substances,
is a set formed by the characteristic values of all key points in the jth cluster in the ith scale image,
and the geological complexity vector of the Nth key point in the jth cluster in the ith scale image is obtained.
Using the key points in each cluster in each scale image as nodes of the graph structure, using the cosine similarity between the key points as the edge weight values between corresponding nodes, calculating the cosine similarity between the key points in each cluster in each scale image, and using the edge weight values between all the nodes in each cluster in each scale image as a set, namely
Wherein the content of the first and second substances,
the weight value of the ith cluster in the ith scale image is collected;
for the jth cluster in the ith scale image
The edge weight between each key point and the Nth node.
For each cluster in each scale image, a graph structure is constructed to characterize the cross-correlation between key points in the cluster, i.e.
Wherein the content of the first and second substances,
a graph structure formed for the jth cluster in the ith scale image,
and forming a set for edge weights among all key points in the jth cluster in the ith scale image.
In the subsequent steps, a graph structure formed by each cluster in a multi-scale space is required to be used as a node, and then a multi-scale undirected graph structure is constructed, while a multi-layer nested graph structure can cause dimension disasters and is extremely complex in calculation, so that graph embedding is utilized before the construction of the multi-scale undirected graphThe method comprises the step of converting the graph structure corresponding to each cluster in each scale image into a low-dimensional vector. In this embodiment, graph embedding is performed by using a graph2vec method, and an embedding vector of a graph structure corresponding to each cluster is output, that is, the input is
Output is as
，
The embedding vector of the jth cluster in the ith scale image.
And constructing a multi-scale undirected graph structure for all embedded vectors in the multi-scale space. Specifically, each embedded vector is used as a node of an undirected graph, data contained in the node is the embedded vector, and the edge weight between the nodes is the cosine similarity between the embedded vectors. The formula for calculating cosine similarity between embedded vectors is as follows:
wherein, the first and the second end of the pipe are connected with each other,
for the embedded vector corresponding to the b-th graph structure in the a-th scale image,
represents
And
cosine similarity between them.
And S5, inputting the multi-scale undirected graph into a trained neural network to obtain the mineral area boundary of the area to be detected, wherein the neural network is used for detecting the boundary of the mineral area.
In order to improve the accuracy of the neural network, the geological images collected by the embodiment include mineral images, boundary images and stratum images, and the neural network is trained by using the training set.
And inputting the multi-scale undirected graph of the region to be detected into the trained neural network, wherein the neural network is a GNN network, the activation function uses a ReLU function, the loss function uses a cross entropy function, and the output is the mineral region boundary of the region to be detected.
The embodiment acquires the texture complexity and the color complexity of each pixel point in the RGB image of the area to be detected, constructs the geological complexity vector of each pixel point according to the texture complexity and the color complexity, has higher feature accuracy compared with single gray information or color information representation of each pixel point, and has better anti-fuzzy capability due to focusing delay in the mobile monitoring process of a monitoring camera, acquires the multi-scale space image of the area to be detected, clusters key points in each scale space image to obtain cluster clusters corresponding to each scale space image, converts each cluster in each scale space into an embedded vector, constructs a multi-scale undirected graph according to the embedded vector corresponding to each cluster, and inputs the multi-scale undirected graph into a trained neural network, the boundaries of the mineral area are obtained. The geological complexity vector is constructed in a mode of combining texture complexity and color complexity, the characteristics of a mineral layer can be comprehensively characterized, the multi-scale undirected graph constructed by the method considers the cross correlation among key points in each scale image and the cross correlation among key points in each cluster in a multi-scale space, the extracted multi-scale undirected graph characteristics have scale invariance, the accuracy of the mineral boundary of the region to be detected is improved, the method provided by the embodiment does not need to collect data fed back by a sensor based on drilling and blasting, the boundary of the mineral region is further obtained, and the safety of the mineral boundary detection process is improved.
Mineral product boundary detection system embodiment based on artificial intelligence
The system for detecting the mineral boundary based on the artificial intelligence comprises a memory and a processor, wherein the processor executes a computer program stored in the memory to realize the method for detecting the mineral boundary based on the artificial intelligence.
Because the method for detecting the mineral boundary based on the artificial intelligence has been described in the embodiment of the method for detecting the mineral boundary based on the artificial intelligence, the embodiment does not give any further details to the method for detecting the mineral boundary based on the artificial intelligence.
It should be noted that: the above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (5)
1. A mineral product boundary detection method based on artificial intelligence is characterized by comprising the following steps:
acquiring an RGB image of a region to be detected;
obtaining a geological complexity vector of each pixel point according to the texture complexity and the color complexity of each pixel point in the RGB image;
obtaining a multi-scale space image according to the RGB image, and recording pixel points with geological complexity larger than a set complexity threshold as key points according to geological complexity vectors of the pixel points in the multi-scale space image; clustering key points corresponding to the spatial images of all scales to obtain clustering clusters corresponding to the spatial images of all scales;
according to the geological complexity vector of each key point in each cluster, constructing a graph structure corresponding to each cluster, according to the graph structure, obtaining an embedded vector corresponding to each cluster, and according to the embedded vector corresponding to each cluster, constructing a multi-scale undirected graph;
inputting the multi-scale undirected graph into a trained neural network to obtain the boundary of the mineral area of the area to be detected, wherein the neural network is used for detecting the boundary of the mineral area;
according to the geological complexity vector of each pixel point in each scale space image, marking the pixel points with the geological complexity larger than a set complexity threshold as key points; the method for clustering the key points corresponding to the spatial images of each scale to obtain the cluster corresponding to the spatial images of each scale comprises the following steps:
obtaining corresponding key points in the space image of each scale by adopting an ORB algorithm according to the geological complexity feature vector of each pixel;
clustering corresponding key points in each hierarchy image by adopting a DBSCAN clustering method to obtain a cluster corresponding to each hierarchy image in the multi-scale space;
the method for constructing the graph structure corresponding to each cluster according to the geological complexity vector of each key point in each cluster comprises the following steps:
obtaining geological complexity vectors of all key points in all the clustering clusters, and calculating cosine similarity between the key points corresponding to all the scale space images;
constructing a graph structure corresponding to each cluster: and taking the key points corresponding to the spatial images of all scales as nodes of the graph structure, taking geological complexity vectors of the key points corresponding to the spatial images of all scales as characteristic values of corresponding nodes in the graph structure, and taking the cosine similarity as an edge weight value between the corresponding nodes in the graph structure.
2. The method for detecting mineral boundaries based on artificial intelligence of claim 1, wherein the method for calculating the texture complexity and the color complexity of each pixel point in the RGB image comprises:
processing each pixel point by adopting LTP ternary coding to obtain an LTP value of each pixel point;
obtaining the texture complexity of each pixel point according to the LTP value of each pixel point;
and acquiring three channel values of the Lab space corresponding to each pixel point, and acquiring the color complexity corresponding to each pixel point according to the three channel values of the Lab space corresponding to each pixel point.
3. The method for detecting mineral product boundaries based on artificial intelligence of claim 2, wherein the method for obtaining the color complexity corresponding to each pixel point according to the three channel values of the Lab space corresponding to each pixel point comprises:
constructing a color vector of each pixel point according to three channel values of the Lab space corresponding to each pixel point;
calculating a color difference vector of each pixel point and a corresponding neighborhood pixel point according to the color vector of each pixel point;
summing the color difference vectors of each pixel point and the corresponding neighborhood pixel point to obtain the color variation of each pixel point, and taking the color variation of each pixel point as the color complexity of the corresponding pixel point.
4. The artificial intelligence-based mineral product boundary detection method according to claim 1, wherein the training method of the neural network comprises:
collecting a plurality of geological images as a training set of a neural network, and marking the geological images with corresponding labels, wherein the labels are boundaries of mineral areas;
and training the neural network according to the training set and the labels corresponding to the images in the training set.
5. An artificial intelligence based mineral boundary detection system comprising a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the artificial intelligence based mineral boundary detection method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210436413.9A CN114529827B (en) | 2022-04-25 | 2022-04-25 | Mineral product boundary detection method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210436413.9A CN114529827B (en) | 2022-04-25 | 2022-04-25 | Mineral product boundary detection method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114529827A CN114529827A (en) | 2022-05-24 |
| CN114529827B true CN114529827B (en) | 2022-07-01 |
Family
ID=81627857
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210436413.9A Active CN114529827B (en) | 2022-04-25 | 2022-04-25 | Mineral product boundary detection method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114529827B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116433887B (en) * | 2023-06-12 | 2023-08-15 | 山东鼎一建设有限公司 | Building rapid positioning method based on artificial intelligence |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108254802A (en) * | 2018-01-11 | 2018-07-06 | 武志明 | A kind of technique for delineating on metalliferous deposit boundary influenced by secondary fragmentation construction |
-
2022
- 2022-04-25 CN CN202210436413.9A patent/CN114529827B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108254802A (en) * | 2018-01-11 | 2018-07-06 | 武志明 | A kind of technique for delineating on metalliferous deposit boundary influenced by secondary fragmentation construction |
Non-Patent Citations (1)
| Title |
|---|
| 一种基于多信息融合的模糊边界检测算法;高梁等;《仪器仪表学报》;20111115(第11期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114529827A (en) | 2022-05-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kim et al. | Data-driven scene parsing method for recognizing construction site objects in the whole image | |
| JP2000030057A (en) | Automatic video interpretation system | |
| Qiao et al. | Deep learning-based pixel-level rock fragment recognition during tunnel excavation using instance segmentation model | |
| CN112651406B (en) | Depth perception and multi-mode automatic fusion RGB-D significance target detection method | |
| CN105654505B (en) | A kind of collaboration track algorithm and system based on super-pixel | |
| CN114529827B (en) | Mineral product boundary detection method and system based on artificial intelligence | |
| CN113988147B (en) | Multi-label classification method and device for remote sensing image scene based on graph network, and multi-label retrieval method and device | |
| CN114943324B (en) | Neural network training method, human motion recognition method and device, and storage medium | |
| CN116797787B (en) | Remote sensing image semantic segmentation method based on cross-modal fusion and graph neural network | |
| CN109711245B (en) | Closed loop detection method based on image candidate region | |
| CN111160389A (en) | Lithology identification method based on fusion of VGG | |
| CN115471467A (en) | High-resolution optical remote sensing image building change detection method | |
| CN116665067B (en) | Ore finding target area optimization system and method based on graph neural network | |
| CN115527269B (en) | Intelligent human body posture image recognition method and system | |
| Hu et al. | RGB-D image multi-target detection method based on 3D DSF R-CNN | |
| Jemilda et al. | Moving object detection and tracking using genetic algorithm enabled extreme learning machine | |
| CN116258953A (en) | Remote sensing image target detection method | |
| CN106251375A (en) | A kind of degree of depth study stacking-type automatic coding of general steganalysis | |
| CN115830449A (en) | Remote sensing target detection method with explicit contour guidance and spatial variation context enhancement | |
| CN116934796A (en) | Visual target tracking method based on twinning residual error attention aggregation network | |
| CN113283393B (en) | Deepfake video detection method based on image group and two-stream network | |
| CN115984093A (en) | Depth estimation method based on infrared image, electronic device and storage medium | |
| CN116012702A (en) | Remote sensing image scene level change detection method | |
| CN116109813A (en) | Anchor hole drilling identification method, system, electronic equipment and medium | |
| CN114758285A (en) | Video interaction action detection method based on anchor freedom and long-term attention perception |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230327 Address after: 226000 No. 3888 Century Avenue, Qidong Economic Development Zone, Nantong City, Jiangsu Province Patentee after: GUODONG TECHNOLOGY CO.,LTD. Address before: 226000 No. 10, Shuntian Road, Qidong Development Zone, Nantong City, Jiangsu Province Patentee before: Qidong Yaheng Electromechanical Equipment Technology Co.,Ltd. |
|
| TR01 | Transfer of patent right |