CN115330664A - Image recognition-based surrounding rock weathering degree full-automatic recognition method and device - Google Patents
Image recognition-based surrounding rock weathering degree full-automatic recognition method and device Download PDFInfo
- Publication number
- CN115330664A CN115330664A CN202210161180.6A CN202210161180A CN115330664A CN 115330664 A CN115330664 A CN 115330664A CN 202210161180 A CN202210161180 A CN 202210161180A CN 115330664 A CN115330664 A CN 115330664A
- Authority
- CN
- China
- Prior art keywords
- surrounding rock
- image
- weathering
- index
- weathering degree
- 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.)
- Pending
Links
- 239000011435 rock Substances 0.000 title claims abstract description 131
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000012549 training Methods 0.000 claims abstract description 33
- 238000011158 quantitative evaluation Methods 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000003709 image segmentation Methods 0.000 claims abstract description 10
- 238000010606 normalization Methods 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000012417 linear regression Methods 0.000 claims abstract description 6
- 238000010586 diagram Methods 0.000 claims description 19
- 238000012937 correction Methods 0.000 claims description 15
- 238000005286 illumination Methods 0.000 claims description 15
- 238000012360 testing method Methods 0.000 claims description 14
- 238000001914 filtration Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 238000010219 correlation analysis Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 description 15
- 206010017076 Fracture Diseases 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 230000011218 segmentation Effects 0.000 description 5
- 208000010392 Bone Fractures Diseases 0.000 description 4
- 238000010276 construction Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000004927 fusion Effects 0.000 description 4
- 229910052500 inorganic mineral Inorganic materials 0.000 description 4
- 239000011707 mineral Substances 0.000 description 4
- 238000009412 basement excavation Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000006253 efflorescence Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 206010037844 rash Diseases 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000002845 discoloration Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 125000001475 halogen functional group Chemical group 0.000 description 1
- 238000003702 image correction Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000004660 morphological change Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/49—Analysis of texture based on structural texture description, e.g. using primitives or placement rules
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/32—Normalisation of the pattern dimensions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Geometry (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method and a device for fully automatically identifying the weathering degree of surrounding rock based on image identification, wherein the identification method comprises the following steps: acquiring a to-be-identified face image, and preprocessing and lighting correcting the to-be-identified face image to obtain a processed image; performing image segmentation based on the characteristic of the weathered surrounding rock on the processed image to obtain a segmented image; judging an index system based on a pre-constructed surrounding rock weathering degree, and acquiring corresponding index values from the processed image and the segmented image; obtaining a quantitative judgment result of the weathering degree of the surrounding rock through a corresponding index value after the training and the obtained quantitative evaluation model of the weathering degree of the surrounding rock and the normalization processing; the quantitative evaluation model of the weathering degree of the surrounding rock is a model constructed based on a multiple linear regression method and the weathering degree judgment index system of the surrounding rock. Compared with the prior art, the method has the advantages of high discrimination accuracy, capability of realizing quantitative determination and the like.
Description
Technical Field
The invention relates to the technical field of tunnel construction, in particular to a method and a device for fully automatically identifying the weathering degree of surrounding rock based on image identification.
Background
The weathering degree of the rock seriously affects the engineering characteristics of the surrounding rock, weathering coefficients (namely the ratio of the compressive strength of the weathered surrounding rock to the compressive strength of the non-weathered surrounding rock) are adopted in the highway tunnel design specification and the railway tunnel design specification as quantitative evaluation indexes for representing the weathering degree of the surrounding rock, as shown in table 1, but in actual engineering, drilling sampling and indoor tests are difficult to carry out on a tunnel face exposed after each excavation.
TABLE 1 surrounding rock weathering degree dividing table of Highway Tunnel design Specification
At present, in tunnel engineering, the weathering degree of the surrounding rock is divided by adopting a qualitative evaluation method, and is mainly analyzed and determined from comprehensive indexes such as rock color, secondary minerals, joint and weathering fracture development conditions, integrity degree, and physical, mechanical and hydraulic property changes of the rock, and the weathering effect and morphological change after weathering of the surrounding rock are greatly different due to the difference of the types of the surrounding rock and different hydrogeological environments, so that a uniform division standard is difficult to establish. The following problems exist in engineering practice:
(1) The method relying on visual inspection and identification of geological engineers is high in subjectivity and is easily influenced by on-site illumination and humidity, different geological engineers often have differences in results obtained by judging the surrounding rock conditions of the same face, and the subjective judgment method is not scientific.
(2) In order to solve the subjective errors, survey units and construction units often adopt a method for establishing a multi-person detection group, and multiple geologists judge and demonstrate the same face repeatedly to obtain a more accurate division result. However, the workload of the identification work is increased, the work efficiency is low, and huge personnel and fund waste is caused.
(3) Based on standardized quantitative determination, an indoor compressive strength test or a field wave velocity ratio test can be adopted, but the similar fresh rock is difficult to find as a control group in a field engineering environment. This quantitative evaluation method has little feasibility from the implementation point of view.
In the actual tunnel excavation engineering process, due to the fact that quantitative evaluation methods such as field sampling-indoor tests are low in efficiency and large in workload, the judgment of the weathering degree of the surrounding rock depends on the subjective judgment of a geological engineer based on the field characteristics of the surrounding rock. Due to the fact that surrounding rock conditions of the tunnel face are complex, engineers engaged in geological work for many years are difficult to accurately divide the weathering degree grade of the surrounding rock, and due to the fact that the judgment standard is fuzzy and the subjectivity is strong, different geological engineers have differences in judgment results of the weathering degree of the same tunnel face.
Aiming at the problems of multiple indexes, fuzzy judgment standards and inaccurate judgment results in the traditional judgment method, researchers try to analyze by adopting a deep learning means, such as patent application No. CN201910768866.X, but the following obvious defects exist:
(1) The training speed is slow. One problem of the neural network is that the training speed has a problem, and in order to obtain a better fitting effect, network levels are often required to be increased and the number of neurons in each layer is increased, however, the number of parameters to be optimized in the network is greatly increased, so that the calculated amount is too large, the training takes a long time, and the model iteration speed is seriously influenced; on the other hand, under the influence of hardware performance, a simpler neural network training is required, which virtually reduces the accuracy of the model; finally, the neural network often needs to collect a large number of effective samples to train a model with strong generalization ability, otherwise, overfitting of the model is easily caused.
(2) The interpretability is poor. For the constructed multilayer neural network, the input and the output of the intermediate neurons are irregular, the characteristic knowledge extracted by each layer of network is difficult to understand, the multilayer neural network can be only used as a black box, and the training results are usually that the training results are not known; on one hand, further adjustment and optimization are difficult to perform according to the training model, and on the other hand, the mutual relation among the surrounding rock indexes cannot be researched, so that the indexes are further quantized; finally, the black box system trained based on neural networks is difficult to understand and apply by field personnel.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides a high-accuracy image-recognition-based full-automatic surrounding rock weathering degree recognition method and device, so that quantitative judgment of the surrounding rock weathering degree is realized.
The purpose of the invention can be realized by the following technical scheme:
a full-automatic surrounding rock weathering degree identification method based on image identification comprises the following steps:
acquiring a to-be-identified face image, and preprocessing and lighting correcting the to-be-identified face image to obtain a processed image;
performing image segmentation based on the characteristic of the weathered surrounding rock on the processed image to obtain a segmented image;
judging an index system based on a pre-constructed surrounding rock weathering degree, and acquiring corresponding index values from the processed image and the segmented image;
obtaining a quantitative judgment result of the weathering degree of the surrounding rock through a corresponding index value after the training and the obtained quantitative evaluation model of the weathering degree of the surrounding rock and the normalization processing;
the quantitative evaluation model of the weathering degree of the surrounding rock is a model constructed based on a multiple linear regression method and the weathering degree judgment index system of the surrounding rock.
Further, the surrounding rock weathering degree judgment index system is constructed based on parameter correlation analysis and comprises a first-level index and a second-level index, the first-level index comprises color, surrounding rock integrity and palm surface texture, the second-level index of the color comprises first, second and third color moments of a component a and b components in a Lab color space, the second-level index of the surrounding rock integrity comprises the number and length of joint crack strips in a unit area, and the second-level index of the palm surface texture comprises contrast, correlation, entropy, stability and second order moments of the texture.
Further, the quantitative evaluation model for the weathering degree of the surrounding rock is obtained by training through the following steps:
constructing a working face image database, wherein working face image samples in the database are provided with image marks, the image marks comprise lithology and weathering degree, and the working face image samples are images obtained after preprocessing and illumination correction are carried out on collected images;
performing image segmentation based on weathering surrounding rock characteristics on the tunnel face image sample;
acquiring a corresponding index value of each sample based on the surrounding rock weathering degree judgment index system, and acquiring a model input and output data set of each sample after normalization processing;
and randomly forming a training set and a testing set by the samples, and training to obtain the quantitative evaluation model of the weathering degree of the surrounding rock.
Further, the preprocessing comprises image denoising processing and image size unification.
Further, the image denoising process includes gaussian low-pass filtering and median filtering.
Further, the illumination correction specifically includes:
converting the preprocessed image from an RGB color space into a Lab color space;
separating Lab color space into 3 independent channel components to form an L component diagram, an a component diagram and a b component diagram, wherein L is a brightness component, and a and b are color components and are unrelated to illumination brightness, and performing brightness correction processing on the L component diagram;
extracting texture features from the original RGB image to form a feature gray level image;
and fusing the L component diagram, the a component diagram, the b component diagram and the characteristic gray level diagram.
Further, the palm face images in the database cover multiple weathering degrees in non-weathering, slightly weathering, medium weathering, strong weathering and full weathering.
Further, when the quantitative evaluation model of the weathering degree of the surrounding rock is trained, an SVM classifier is built by adopting various kernel functions, the training set and the test set are applied to carry out classification accuracy comparison, and an optimal kernel function is selected; and searching global optimal hyper-parameters by adopting a grid division method, and constructing an SVM classifier which accords with the optimal degree of the surrounding rock weathering of the working face.
Further, the quantitative evaluation model of the weathering degree of the surrounding rock is expressed as follows:
wherein:
β 1 ＝A 1 +A 2 +A 3 +A 4 +A 5 +A 6
β 2 ＝A 7 +A 8
β 3 ＝A 9 +A 10 +A 11 +A 12 +A 13
in the formula: c, color index, which is obtained by combining 6 secondary indexes of the subordinate plants;
w is a surrounding rock integrity index which is obtained by combining 2 secondary indexes;
t is a palm surface texture index which is obtained by combining 5 secondary indexes;
β 1 -the color index corresponds to a weight;
β 2 -the surrounding rock integrity indicator corresponds to a weight;
β 3 -the palm face texture index corresponds to a weight;
n i -index corresponding to index, i =1,2, \ 8230, 13;
x i -normalized index value.
The present invention also provides an electronic identification device comprising:
one or more processors;
a memory; and
one or more programs stored in the memory, the one or more programs including instructions for performing the fully automatic identification method as described above.
Compared with the prior art, the method has the advantages that the judgment index which has clear physical significance and can represent the weathering degree of the surrounding rock is constructed, the corresponding relation between the characteristics of the surrounding rock and the weathering degree and the quantitative index value are constructed on the basis of the judgment index, and finally the quantitative judgment of the weathering degree of the surrounding rock is realized, so that the method has the following beneficial effects:
(1) According to the method, the collected tunnel face image material is subjected to image denoising and illumination intensity correction, so that the influence of a complex environment of a construction site on the judgment accuracy is eliminated to the maximum extent.
(2) According to the method, a large number of indexes are listed by extracting the color, integrity and texture characteristics of the palm surface image, on the basis of performing parameter correlation analysis on each index, the indexes which do not influence the result are removed, the indexes with strong correlation are screened, and the judgment index capable of representing the weathering degree of the surrounding rock is finally determined, so that the problem of fuzzy judgment index in the conventional judgment process is solved, and meanwhile, the model training time can be greatly reduced.
(3) According to the method, related characteristic values are extracted based on a large number of face materials, and a multivariate linear fitting means is adopted, so that an expression of multi-characteristic index combination and each index weight are obtained under the condition of meeting the requirement of precision, and quantitative description of the weathering index of the surrounding rock is accurately realized.
(4) The method adopts the traditional machine learning means, the support vector machine and the polynomial fitting method have good interpretability, each algorithm gives an optimization target, and geometric representation can be adopted; specific relation expression between input quantity and output quantity can be constructed on the basis of a large number of case learning, the value range of the index can be summarized and quantified, and quantitative determination of the weathering degree of the surrounding rock can be rapidly realized.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a diagram of image fusion effect;
FIG. 3 is a schematic diagram of the model training process of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Interpretation of terms
A palm surface: the tunnel face is also called as the sub face of \30979. I.e. a working face where the excavation of a tunnel (in coal mining, mining or tunnelling) is constantly propelled forward.
Degree of weathering of surrounding rock: the damage degree of the weathering effect on the rock mass comprises the disintegration and change degree of the rock mass and the weathering depth; generally, the method comprises the following steps: the weathering degree is five, namely non-weathering, slightly weathering, middle weathering, strong weathering and total weathering.
Example 1
The invention provides a full-automatic surrounding rock weathering degree identification method based on image identification, which is characterized in that mobile phone photographing collection is carried out on the latest excavated and exposed face by site constructors, a feature index system of a surrounding rock weathering degree digital image is established based on surrounding rock weathering degree judging indexes in relevant specifications, image preprocessing, image segmentation and image relevant feature extraction are carried out on the face collected image, comprehensive judgment is carried out on the surrounding rock weathering degree based on extracted features, each index is quantized on the basis of a large amount of data, an evaluation model is established, and finally, automatic judgment of the face surrounding rock weathering degree is realized.
As shown in FIG. 1, the full-automatic identification method of the invention comprises the following steps:
the method comprises the following steps of S1, acquiring a to-be-identified face image, and carrying out pretreatment and illumination correction on the to-be-identified face image to obtain a processed image;
s2, performing image segmentation based on weathered surrounding rock characteristics on the processed image to obtain a segmented image;
s3, judging an index system based on the pre-constructed surrounding rock weathering degree, and acquiring corresponding index values from the processed image and the segmented image;
and S4, obtaining a quantitative determination result of the weathering degree of the surrounding rock through a training obtained quantitative evaluation model of the weathering degree of the surrounding rock and the corresponding index value after normalization processing, wherein the quantitative evaluation model of the weathering degree of the surrounding rock is a model constructed based on a multiple linear regression method and the weathering degree determination index system of the surrounding rock.
(1) Parametric index correlation analysis and key index determination
In engineering, the weathering degree of the surrounding rock is mainly analyzed from multiple dimensions such as color, secondary minerals, joints, weathering fissure development, integrity, hardness degree, water physical property, freshness and surface smoothness.
However, the indexes have correlation, the color change of the surrounding rock mainly comes from mineral alteration and new secondary minerals are generated, and the two indexes can be summarized into a single color index from the perspective of machine vision; similarly, the joint, weathering fracture development and integrity are summarized as the integrity index of the surrounding rock; rock freshness, surface smoothness or roughness, etc. are summarized as the face texture features. So far, three first-level indexes capable of reacting to weather the surrounding rock are obtained: "color, surrounding rock integrity, tunnel face texture".
Based on machine vision technology:
1) For color feature description, 9 secondary indexes including first moment, second moment and third moment of L, a and b components can be adopted for representation;
2) For the description of the integrity characteristics of the surrounding rock, the description is mainly from the perspective of joints and weathered fractures, and the description can be represented by 2 secondary indexes including the number of joints in a unit area and the total length of the joints in the unit area;
3) For the description of the texture features of the palm surface, 5 secondary indexes of contrast, correlation, entropy, smoothness and secondary moment of the texture can be adopted for representation.
By adopting a method for calculating a correlation index (Pearson correlation coefficient), the correlation strength among the indexes and between the indexes and the result is researched, and the calculation formula is as follows:
in the formula: x is the number of i ，y i -group i data measured simultaneously for x and y quantities;
n is the total number of data statistics;
s (x), S (y) -the standard deviation of the data corresponding to the x and y quantities.
According to the correlation research result, three secondary indexes of a first moment, a second moment and a third moment of the image brightness component L which have no influence on the judgment result are removed, and finally the remaining 13 secondary indexes are reserved as the judgment indexes of the surrounding rock weathering degree, as shown in the table 2.
TABLE 2 Weathering degree determination index for surrounding rock
(2) Quantitative evaluation model for weathering degree of surrounding rock
The method adopts a quantitative evaluation model of the weathering degree of the surrounding rock, which is a model constructed based on a multiple linear regression method and the weathering degree judgment index system of the surrounding rock and is obtained by training through the following steps:
11 Constructing a palm surface image database, wherein a palm surface image sample in the database is provided with an image mark, the image mark comprises lithology and weathering degree, and the palm surface image sample is an image obtained after preprocessing and illumination correction are carried out on an acquired image.
In this embodiment, the collected image is obtained by shooting the image by the engineering technician at a certain distance in front of the face with a mobile phone, and the light supplement lamp is turned on during the shooting process to ensure that the image of the face is not covered by a shadow. The shot image covers various weathering degrees of 'non-weathering, slightly weathering, middle weathering, strong weathering and full weathering' as much as possible, and each photo is marked with 'lithology + weathering degree'.
The image preprocessing mainly comprises the following steps: and image denoising processing and image size unification. The image noise can be divided into gaussian noise and impulse noise aiming at dust and light source interference generated by a severe construction environment in a tunnel and self interference of photographing equipment. Firstly, filtering and eliminating Gaussian noise by adopting a Gaussian low-pass filtering method, and then filtering and eliminating pulse noise by adopting a median filtering method. And finally, performing pixel transformation processing on the palm surface image subjected to filtering processing, and performing image size unification and format standardization.
In this embodiment, the illumination correction specifically includes:
(101) Color space transformation
The RGB color data of the pre-processed image, acquired directly by the CMOS camera, is transformed by a two-step matrix, first converting the RGB color space to the CIEL XYZ color space:
the CIEL XYZ color space is then converted to the LAB color space:
(102) Illumination correction based on single-scale Retinex enhancement algorithm
By adopting the Retinex algorithm idea, the image is defined as the product of the reflected light image and the incident light image, and can be expressed as follows by the formula:
S(x,y)＝R(x,y)·L(x,y) (5)
wherein S (x, y) is an original image, R (x, y) is a reflected light image including essential attributes of the image, and L (x, y) is an incident image including a dynamic range of the image. Based on the theory, the enhancement and correction of the image illumination intensity can be realized by means of adjusting or eliminating the incident light image of the image. Expressed in logarithmic form as:
log(R(x,y))＝log(S(x,y))-log(L(x,y)) (6)
the incident light image can be obtained by performing convolution calculation on the original image and the Gaussian function, and the calculation method comprises the following steps:
L(x,y)＝F(x,y)*S(x,y) (7)
where F (x, y) is a Gaussian surround function defined as:
where c denotes the gaussian surround scale and λ is a constant such that the integral of F (x, y) is 1, substituting equation (7) into equation (6) yields an image correction algorithm:
r(x,y)＝log(R(x,y))＝log(S(x,y))-log[F(x,y)*S(x,y)] (9)
(103) Feature extraction and image fusion
And extracting image texture information in the original RGB image by adopting a canny edge detection operator to generate a gray image. In order to avoid halo, color distortion and detail loss caused by the traditional algorithm processing, a fusion idea is adopted to perform fusion operation on the corrected L component image, the separated a and b component images and the texture characteristic gray level image to generate an illumination intensity correction image with details and colors reserved, as shown in fig. 2.
After the processing, the pre-processed image is subjected to illumination brightness correction in a Lab color space and is fused with the texture characteristics of the image, so that the brightness of the image is uniform under the conditions of detail retention and color distortion prevention of the image, and the image is more suitable for human eyes.
12 Performing image segmentation based on weathering surrounding rock characteristics on the working face image sample.
The traditional color image segmentation is based on color gradient change for segmentation, and has poor applicability to weathered surrounding rocks; in order to accurately segment joint and fracture areas, extract the number of joint fractures and count the length, and aim at the characteristics (discoloration and breakage) of the surrounding rock, a traditional color image segmentation method is adopted in combination with a surrounding rock joint characteristic comprehensive segmentation mode. The traditional color image region segmentation method based on color difference segmentation is combined with tunnel face joint, crack and texture features for correction to obtain image region segmentation more in line with characteristics of weathered surrounding rocks, and subsequent features can be conveniently extracted.
13 Obtaining a corresponding index value of each sample based on the surrounding rock weathering degree judgment index system, and obtaining a model input and output data set of each sample after normalization processing.
Aiming at the data with multiple indexes, various types of data are processed to dimensionless numerical values in a unified interval by adopting normalization processing, and the normalization processing is adopted to map as follows:
in the formula: x is the number of min -the minimum value in the raw data set min (x);
x max -the maximum value in the original dataset max (x);
x-raw data;
y-normalized data.
Marking the surrounding rock sample according to the weathering degree, wherein: the "non-efflorescing" is marked "1", "slightly efflorescing" is marked "2", "medium efflorescence" is marked "3", "strongly efflorescing" is marked "4" and "fully efflorescing" is marked "5". The input and output data set form can be expressed as:
{x 1 ,x 2 ,x 3 ,...,x 13 ,y}
the following steps: x is the number of 1 -a normalized a-component first moment calculation value;
x 2 -normalized a-component second moment calculation values;
x 3 -a normalized a-component third moment calculation;
x 4 -a normalized b-component first moment calculation value;
x 5 -normalized b-component second moment calculation values;
x 6 -a normalized b-component third moment calculation;
x 7 -calculating the number of joint fractures per normalized unit area;
x 8 calculating the total length of the joint fracture in the unit area after normalization;
x 9 -calculating a normalized texture contrast value;
x 10 -normalized texture correlation calculation;
x 11 -calculating a normalized texture entropy value;
x 12 -calculating a normalized texture stationarity value;
x 13 -normalized texture second moment calculation values;
y is an image mark value, namely a training model output value, and the values are 1,2, 3, 4 and 5.
14 And) randomly combining the samples into a training set and a testing set, and training to obtain the quantitative evaluation model of the weathering degree of the surrounding rock.
And dividing the image sample into a training set and a testing set according to the ratio of 7 to 3, and taking the 13 characteristic parameters and the label parameters capable of representing the weathering degree of the face surrounding rock as input parameters. Constructing an SVM classifier by using several commonly used kernel functions, comparing classification accuracy by using the training set and the test set, and selecting an optimal kernel function; and searching the values of the global optimal hyperparameters C and gamma by adopting a grid division method, and finally constructing the SVM classifier which accords with the optimal tunnel face surrounding rock weathering degree. Based on a large number of image characteristic parameters, a polynomial fitting algorithm is adopted, the relation between the characteristic quantity and the output quantity is fitted under the condition of certain precision, the combination relation between each first-level index and the subordinate second-level index is trained, the weight of each first-level index is weighted, and a quantitative calculation expression of the weathering degree of the surrounding rock is obtained. Meanwhile, according to the calculation method and the corresponding weight of the three indexes of color, surrounding rock integrity and tunnel face texture determined by the polynomial expression, the parameter value range is summarized and divided. And finally, establishing a quantitative evaluation model of the weathering degree of the surrounding rock, and realizing the quantitative evaluation of the weathering degree of the surrounding rock.
In this embodiment, a multiple linear regression method is used to establish a model, the highest degree n in the polynomial and the training precision threshold epsilon are set, and when the deviation between the output result of the model and the tag value is within the allowable precision epsilon range, the model training is completed. And verifying the model by adopting the test set, and if the prediction precision of the trained model on the test set is not high, readjusting the highest degree n of the polynomial until training the model which has better performance on the training set and the test set. The model training process is shown in fig. 3.
In this embodiment, the model finally constructed is as follows:
wherein:
β 1 ＝A 1 +A 2 +A 3 +A 4 +A 5 +A 6
β 2 ＝A 7 +A 8
β 3 ＝A 9 +A 10 +A 11 +A 12 +A 13
in the formula: c, color index parameter, which is obtained by combining 6 secondary indexes;
w is a surrounding rock integrity index which is obtained by combining 2 secondary indexes;
t is the palm surface texture index, which is obtained by combining the following 5 secondary indexes.
β 1 -the color index corresponds to a weight;
β 2 -the surrounding rock integrity indicator corresponds to a weight;
β 3 -the palm face texture index corresponds to a weight;
n i -an index of the corresponding index.
Based on the existing tunnel face sample, the method is adopted for processing, the model fitting precision is set to be 50%, and each index is 1 (n) i = 1), a corresponding determination model of the index of the weathering degree of the surrounding rock can be established, and the determination model is shown in table 3.
Wherein:
(1) The color index represents the alteration change area and the change degree of the surrounding rock on the tunnel face and has actual physical significance; the larger the index value is, the larger the weathering altered area of the palm surface is, and the deeper the weathering degree is.
(2) The integrity index of the surrounding rock represents the development condition of the joint fissure of the tunnel face and has actual physical significance; the larger the index value is, the more broken the surrounding rock of the tunnel face is, and the deeper the weathering degree is.
(3) The tunnel face texture index represents the roughness of the tunnel face surrounding rock surface and has practical physical significance; the larger the index value, the rougher the palm face and the deeper the degree of efflorescence.
(4) The value range of each judgment index in table 3 is obtained by training and summarizing the current sample, and the value range can be corrected when the sample amount is increased until the sample amount reaches a certain degree, and the value range tends to be stable.
TABLE 3 Weathering degree index judgment model for surrounding rock
Note: in order to simplify the calculation amount, the following processing measures are adopted in the table:
(1) The model precision is set to be 50%, and only the weathering degree index F is continuous in value;
(2) Setting each secondary index n i 1, the index combination becomes a linear combination;
(3) The color index, the surrounding rock integrity index and the tunnel face texture index are processed in a way of corresponding to the weight, so that the sum of the weights is 1;
(4) Under the conditions of the above processing, a calculation result with a value interval length of 1 is generated.
The above method, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Example 2
The present embodiments provide an electronic identification device comprising one or more processors, memory, and one or more programs stored in the memory, the one or more programs including instructions for performing the fully automatic identification method of embodiment 1.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations can be devised by those skilled in the art in light of the above teachings. Therefore, the technical solutions that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the prior art according to the concepts of the present invention should be within the scope of protection determined by the claims.
Claims (10)
1. A full-automatic surrounding rock weathering degree identification method based on image identification is characterized by comprising the following steps:
acquiring a to-be-identified face image, and preprocessing and lighting correcting the to-be-identified face image to obtain a processed image;
performing image segmentation based on the characteristic of the weathered surrounding rock on the processed image to obtain a segmented image;
judging an index system based on the pre-constructed surrounding rock weathering degree, and acquiring corresponding index values from the processed image and the segmented image;
obtaining a quantitative judgment result of the weathering degree of the surrounding rock through a corresponding index value after the training and the obtained quantitative evaluation model of the weathering degree of the surrounding rock and the normalization processing;
the quantitative evaluation model of the weathering degree of the surrounding rock is a model constructed based on a multiple linear regression method and the weathering degree judgment index system of the surrounding rock.
2. The method for fully automatically identifying the weathering degree of the surrounding rock based on the image identification is characterized in that the surrounding rock weathering degree judgment index system is constructed based on parameter correlation analysis and comprises a first-level index and a second-level index, the first-level index comprises color, surrounding rock integrity and palm surface texture, the second-level index of the color comprises first moment, second moment and third moment of color of a component and b component in Lab color space, the second-level index of the surrounding rock integrity comprises the number and length of joint crack strips in unit area, and the second-level index of the palm surface texture comprises contrast, correlation, entropy, smoothness and second moment of the texture.
3. The method for fully automatically identifying the weathering degree of the surrounding rock based on the image identification is characterized in that the surrounding rock weathering degree quantitative evaluation model is obtained by training through the following steps:
constructing a working face image database, wherein working face image samples in the database are provided with image marks, the image marks comprise lithology and weathering degree, and the working face image samples are images obtained after preprocessing and illumination correction are carried out on collected images;
performing image segmentation based on weathered surrounding rock characteristics on the tunnel face image sample;
acquiring a corresponding index value of each sample based on the surrounding rock weathering degree judgment index system, and acquiring a model input and output data set of each sample after normalization processing;
and randomly forming a training set and a testing set by the samples, and training to obtain the quantitative evaluation model of the weathering degree of the surrounding rock.
4. The method for fully automatically identifying the weathering degree of the surrounding rock based on the image identification as claimed in claim 1 or 3, wherein the preprocessing comprises image denoising processing and image size unification.
5. The method for fully automatically identifying the weathering degree of the surrounding rock based on the image identification is characterized in that the image denoising process comprises Gaussian low-pass filtering and median filtering.
6. The method for fully automatically identifying the weathering degree of the surrounding rock based on image identification as claimed in claim 1 or 3, wherein the illumination correction specifically comprises:
converting the preprocessed image from an RGB color space into a Lab color space;
separating Lab color space into 3 independent channel components to form an L component diagram, an a component diagram and a b component diagram, wherein L is a brightness component, and a and b are color components and are irrelevant to illumination brightness, and performing brightness correction processing on the L component diagram;
extracting texture features from the original RGB image to form a feature gray image;
and fusing the L component diagram, the a component diagram, the b component diagram and the characteristic gray level diagram.
7. The method for fully automatically identifying the weathering degree of the surrounding rock based on the image identification is characterized in that the image of the working face in the database covers multiple weathering degrees in non-weathering, slightly weathering, medium weathering, strong weathering and full weathering.
8. The method for full-automatic identification of the weathering degree of the surrounding rock based on the image identification is characterized in that when the quantitative evaluation model of the weathering degree of the surrounding rock is trained, an SVM classifier is built by adopting various kernel functions, the training set and the test set are used for carrying out classification accuracy comparison, and an optimal kernel function is selected; and searching global optimal hyper-parameters by adopting a grid division method, and constructing an SVM classifier which accords with the optimal degree of the surrounding rock weathering of the working face.
9. The method for fully automatically identifying the weathering degree of the surrounding rock based on image identification as claimed in claim 2, wherein the quantitative evaluation model of the weathering degree of the surrounding rock is expressed as:
wherein:
β 1 ＝A 1 +A 2 +A 3 +A 4 +A 5 +A 6
β 2 ＝A 7 +A 8
β 3 ＝A 9 +A 10 +A 11 +A 12 +A 13
in the formula: c, color index, which is obtained by combining 6 secondary indexes of the subordinate plants;
w is a surrounding rock integrity index which is obtained by combining 2 secondary indexes of the subordinate;
t is a palm surface texture index which is obtained by combining 5 secondary indexes;
β 1 -the color index corresponds to a weight;
β 2 -the surrounding rock integrity indicator corresponds to a weight;
β 3 -the palm face texture index corresponds to a weight;
n i -index corresponding to index, i =1,2, \ 8230, 13;
x i -normalized index value.
10. An electronic identification device, comprising:
one or more processors;
a memory; and
one or more programs stored in the memory, the one or more programs including instructions for performing the fully automatic identification method of any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210161180.6A CN115330664A (en) | 2022-02-22 | 2022-02-22 | Image recognition-based surrounding rock weathering degree full-automatic recognition method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210161180.6A CN115330664A (en) | 2022-02-22 | 2022-02-22 | Image recognition-based surrounding rock weathering degree full-automatic recognition method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115330664A true CN115330664A (en) | 2022-11-11 |
Family
ID=83915647
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210161180.6A Pending CN115330664A (en) | 2022-02-22 | 2022-02-22 | Image recognition-based surrounding rock weathering degree full-automatic recognition method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115330664A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116913416A (en) * | 2023-09-12 | 2023-10-20 | 中国地质大学（北京） | Quantitative evaluation and prediction method and device for weathering degree of sea rock |
| CN116952712A (en) * | 2023-04-13 | 2023-10-27 | 成都理工大学 | Quantitative evaluation method for rock brittleness of unconventional oil and gas reservoir |
| CN117333679A (en) * | 2023-10-09 | 2024-01-02 | 中国地质环境监测院 | Rock mass structure identification system based on image identification |
-
2022
- 2022-02-22 CN CN202210161180.6A patent/CN115330664A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116952712A (en) * | 2023-04-13 | 2023-10-27 | 成都理工大学 | Quantitative evaluation method for rock brittleness of unconventional oil and gas reservoir |
| CN116952712B (en) * | 2023-04-13 | 2024-03-12 | 成都理工大学 | Quantitative evaluation method for rock brittleness of unconventional oil and gas reservoir |
| CN116913416A (en) * | 2023-09-12 | 2023-10-20 | 中国地质大学（北京） | Quantitative evaluation and prediction method and device for weathering degree of sea rock |
| CN116913416B (en) * | 2023-09-12 | 2023-12-26 | 中国地质大学（北京） | Quantitative evaluation and prediction method and device for weathering degree of sea rock |
| CN117333679A (en) * | 2023-10-09 | 2024-01-02 | 中国地质环境监测院 | Rock mass structure identification system based on image identification |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115330664A (en) | Image recognition-based surrounding rock weathering degree full-automatic recognition method and device | |
| CN112149512A (en) | Helmet wearing identification method based on two-stage deep learning | |
| Abascal et al. | Identifying degrees of deprivation from space using deep learning and morphological spatial analysis of deprived urban areas | |
| CN115019171B (en) | Non-contact surrounding rock rapid partition grading method | |
| CN114581764B (en) | Underground structure crack disease discriminating method based on deep learning algorithm | |
| CN105243387A (en) | Open-pit mine typical ground object classification method based on UAV image | |
| CN114612789A (en) | Method for extracting evergreen forest stand changes through long-time satellite remote sensing | |
| CN110889834A (en) | Road tunnel surrounding rock grading method based on cloud computing | |
| Guo et al. | Surface defect detection of civil structures using images: Review from data perspective | |
| CN116403093B (en) | Multi-source remote sensing interpretation method and device for potential landslide along railway | |
| CN109269987A (en) | A kind of qualitative estimating and measuring method of salt lake water body salinity based on SPOT5 remotely-sensed data | |
| CN117522149A (en) | Tunnel security risk identification method and device and security management platform | |
| CN115761038B (en) | Tunnel face geological sketch method and system based on image spectrum technology | |
| CN111815625A (en) | Early lung cancer prediction method based on deep learning | |
| CN115497006B (en) | Urban remote sensing image change depth monitoring method and system based on dynamic mixing strategy | |
| CN112906528B (en) | Urban building material classification method fusing multi-source satellite remote sensing data | |
| Wu et al. | A novel coal-gangue recognition method in underground coal mine based on image processing | |
| CN112767365A (en) | Flaw detection method | |
| CN117876381B (en) | AI visual detection method and system for identifying and analyzing concrete structure cracks | |
| CN117854061B (en) | Tunnel face trace crack extraction method and system based on three-dimensional point cloud | |
| Wei et al. | Application of crack detection algorithm using convolutional neural network in concrete pavement construction | |
| CN117828525A (en) | Tunnel surrounding rock stress monitoring method and system based on multisource information fusion analysis | |
| CN117688640A (en) | Intelligent generation method for digital twin-driven tunnel surrounding rock joint network | |
| Bera | Enhancing the Classification and Identification of Natural Rocks using Swin-Transformer Architecture | |
| CN116343042A (en) | Co-vibration landslide extraction method, system, electronic equipment and storage medium |
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 |