CN112613531B - Soil and rock classification method and device based on image processing - Google Patents
Soil and rock classification method and device based on image processing Download PDFInfo
- Publication number
- CN112613531B CN112613531B CN202011336766.9A CN202011336766A CN112613531B CN 112613531 B CN112613531 B CN 112613531B CN 202011336766 A CN202011336766 A CN 202011336766A CN 112613531 B CN112613531 B CN 112613531B
- Authority
- CN
- China
- Prior art keywords
- image
- gradient
- soil
- rock
- circuit
- 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
- 239000002689 soil Substances 0.000 title claims abstract description 109
- 239000011435 rock Substances 0.000 title claims abstract description 91
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000012545 processing Methods 0.000 title claims abstract description 17
- 238000001514 detection method Methods 0.000 claims abstract description 44
- 238000005070 sampling Methods 0.000 claims abstract description 39
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 31
- 238000004364 calculation method Methods 0.000 claims description 26
- 238000001914 filtration Methods 0.000 claims description 24
- 238000012937 correction Methods 0.000 claims description 15
- 238000007781 pre-processing Methods 0.000 claims description 14
- 238000004891 communication Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 7
- 238000002203 pretreatment Methods 0.000 claims 1
- 239000004927 clay Substances 0.000 abstract description 3
- 238000013461 design Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 13
- 239000000523 sample Substances 0.000 description 13
- 238000000605 extraction Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 239000002245 particle Substances 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000010355 oscillation Effects 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 239000003990 capacitor Substances 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000000919 ceramic Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000006056 electrooxidation reaction Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
- G01N27/02—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating impedance
- G01N27/04—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating impedance by investigating resistance
- G01N27/048—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating impedance by investigating resistance for determining moisture content of the material
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Chemical & Material Sciences (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Multimedia (AREA)
- Electrochemistry (AREA)
- Health & Medical Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a soil and rock classification method and device based on image processing, and belongs to the technical field of image processing. In the design of the automatic classification method of soil and rock, the invention carries out unsupervised clustering two-class classification on the images based on the gradient sampling duty ratio of the images, and completes one-class classification. In soil subdivision, a soil moisture content detection circuit is combined, and the soil moisture content is utilized to realize the distinction of the powdery clay and the plain fill. In rock subdivision, a vector profile feature set is provided to represent texture complexity of rock images, and unsupervised clustering secondary classification is carried out through the texture complexity, so that secondary classification of the rock is realized; and further, a profile estimation feature set is provided to represent the approximation degree of the shape and the circle of the object, and the unsupervised clustering secondary classification is carried out to realize the tertiary classification of the rock. According to the invention, the fine classification and identification of soil and rock can be automatically completed according to the sampling image, the classification is quick and efficient, and the requirement on hardware is low.
Description
Technical Field
The invention relates to an image processing technology, in particular to a scheme for classifying soil and rock by using the image processing technology.
Background
For a construction scene, different construction modes exist for different geological features, so that detection and evaluation of the construction environment are required.
The detection and evaluation are originally distinguished by manual excavation sampling, and the categories of the environmental soil and the rock are identified, but the method is very time-consuming and labor-consuming, and has low accuracy.
With the continuous update of image processing technology, the soil or rock of the sample can be accurately identified and classified by means of image processing. However, the existing classification solutions mostly complete the identification of soil or rock by means of machine learning/deep learning, for example, CN110261330a discloses a rock classification deep learning model rock classification constraint inheritance loss method. Such a method needs to collect a huge sample set, and needs to accurately define the category of the sample, otherwise, the classifier is affected, and in addition, the calculation amount of the method is larger. Moreover, the existing methods can only realize the identification and classification of one of the soil or the rock.
Disclosure of Invention
The invention aims at: in order to solve the problems, a soil and rock classification method based on image processing is provided to provide a solution capable of automatically classifying soil/rock in an image.
The technical scheme adopted by the invention is as follows:
a method for classifying soil and rock, comprising: A. the following processing is performed on each of the acquired original images: performing first preprocessing on the acquired original image to obtain a corresponding first binarized image; calculating the transverse gradient and the longitudinal gradient of the first binarized image, and calculating the gradient sampling duty ratio corresponding to the original image according to the transverse gradient and the longitudinal gradient; B. performing unsupervised clustering second classification on gradient sampling duty ratios corresponding to all original images, and classifying environments corresponding to all original images into soil types and rock types at one time; for the soil class, executing the step C to perform secondary classification, and for the rock class, executing the step D to perform secondary classification; C. classifying the soil corresponding to the original image according to the water content of the soil corresponding to the original image; D. performing unsupervised clustering secondary classification on the texture complexity of the original image, and secondarily classifying the rock corresponding to the original image into a first rock type and a second rock type; for the first rock, executing the step E to classify for three times; E. performing second preprocessing on the original image to obtain a corresponding second binarized image, and performing unsupervised clustering second classification on rocks corresponding to the original image by taking the connected domain duty ratio and the mass center boundary distance confusion degree of the second binarized image as characteristics.
Further, the method for calculating the gradient sampling duty ratio comprises the following steps: equidistant sampling is carried out on the rows/columns of the transverse/longitudinal gradient; calculating the ratio of the gradient information points of the sampling row/column to the information points in each row/column respectively; and performing unsupervised clustering second classification on the ratio of the sampling row/column, and taking the class with more clusters as the gradient sampling duty ratio corresponding to the original image.
Further, the step C is characterized in that the soil water content detection circuit is used for detecting the water content of soil, the soil water content detection circuit comprises a square wave signal generating circuit, a sensing circuit, a rectifying circuit, a filtering circuit, a correcting circuit, an amplifying circuit and a display circuit, the square wave signal generating circuit, the sensing circuit, the rectifying circuit and the filtering circuit are sequentially connected, the input end of the correcting circuit is connected with a power supply of the soil water content detection circuit, the filtering circuit and the output end of the correcting circuit are respectively connected with two input ends of the amplifying circuit, and the output end of the amplifying circuit is connected with the display circuit.
Further, the texture complexity comprises three characteristics of contour complexity, transverse curvature coefficient and longitudinal curvature coefficient; the calculation method of the contour complexity comprises the following steps: combining the transverse gradient and the longitudinal gradient of the original image into a gradient map, traversing the gradient map, counting the total number of points with non-zero pixel values, and representing the complexity of the contour by the proportion of the total number of points with non-zero pixel values to the total pixel points of the gradient map; the transverse curvature coefficient is obtained by calculating the curvature of the transverse gradient of the original image; the longitudinal curvature coefficient is obtained by calculating the curvature of the longitudinal gradient of the original image.
Further, the calculating the curvature of the lateral gradient of the original image includes: binarizing the transverse gradient of the original image to obtain a transverse gradient image; filtering the transverse gradient image to filter out a communication region with an area smaller than a first threshold value; calculating the curvature of the filtered transverse gradient image to obtain a transverse curvature map; and taking the variance value of the point with the pixel point not being zero in the transverse curvature map as a transverse curvature coefficient.
Further, the calculating the curvature of the longitudinal gradient of the original image includes: binarizing the longitudinal gradient of the original image to obtain a longitudinal gradient image; filtering the longitudinal gradient image to filter out a communication region with an area smaller than a second threshold value; calculating the curvature of the filtered longitudinal gradient image to obtain a longitudinal curvature map; and taking the variance value of the point with the pixel point not being zero in the longitudinal curvature map as a longitudinal curvature coefficient.
Further, the performing a second preprocessing on the original image to obtain a corresponding second binarized image includes: performing first preprocessing on the original image to obtain a corresponding first binarized image; and calculating a connected region in the first binarized image, filtering the connected region with the area smaller than a third threshold value in the first binarized image, and performing opening operation on the rest part to obtain a second binarized image.
Further, the connected domain duty ratio is an area ratio of an area of each connected domain to a corresponding minimum frame selected rectangle in the second binarized image.
Further, the calculating method of the centroid boundary distance confusion degree comprises the following steps: extracting the outline of the connected domain in each minimum frame selection rectangle in the second binarization image; and calculating Euclidean distances from each point on each connected domain contour to the centroid, and taking the variance value of each Euclidean distance as the boundary distance confusion of the centroid.
The invention also provides a soil and rock classification device based on image processing, which comprises a processor, a computer readable storage medium and a soil water content detection circuit, wherein the processor is respectively connected with the computer readable storage medium and the soil water content detection circuit; the soil water content detection circuit is used for detecting the water content of soil, and the computer readable storage medium stores a computer program; the soil water content detection circuit transmits the detection result to the processor, and the processor runs the computer program to execute the soil and rock classification method.
In summary, due to the adoption of the technical scheme, the beneficial effects of the invention are as follows:
1. according to the invention, the fine classification and identification of soil and rock can be automatically completed according to the sampling image.
2. The scheme of the invention completely accords with the natural properties of soil and rock, and has high classification accuracy.
3. The invention has the advantages of no need of early sampling and training, direct use, simpler configuration, small calculated amount, quick classification and low requirement on hardware.
Drawings
The invention will now be described by way of example and with reference to the accompanying drawings in which:
FIG. 1 is one embodiment of a process of a soil, rock classification method.
Fig. 2 is a process of first preprocessing an image.
Fig. 3 is a graph of the lateral/longitudinal gradient of a soil-like picture and a rock-like picture.
Fig. 4 is a graph of a horizontal/vertical gradient sample of a soil-like picture and a rock-like picture.
Fig. 5 is an embodiment of a soil moisture content detection circuit.
Fig. 6 is an embodiment of a soil pressure detection circuit.
Fig. 7 is a block diagram of the soil environment detection system.
Fig. 8 is a diagram of one embodiment of a display circuit.
Fig. 9 is an overall gradient map of a rock class picture.
Fig. 10 is a lateral curvature map of a rock-like picture.
Fig. 11 is a longitudinal curvature map of a rock-like picture.
Fig. 12 is a process of second preprocessing an image.
Fig. 13 is a connected information extraction diagram.
Fig. 14 is a schematic view of 6 soils/rocks.
Detailed Description
All of the features disclosed in this specification, or all of the steps in a method or process disclosed, may be combined in any combination, except for mutually exclusive features and/or steps.
Any feature disclosed in this specification (including any accompanying claims, abstract) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. That is, each feature is one example only of a generic series of equivalent or similar features, unless expressly stated otherwise.
The drawings illustrate:
in fig. 2, (1) an original color image; (2) a gray scale image; (3) a first binary image.
In fig. 3, (1) a lateral gradient map of a soil-like picture; (2) a longitudinal gradient map of the soil type picture; (3) a lateral gradient map of the rock class picture; (4) longitudinal gradient map of rock class image.
In fig. 4, (1) a lateral gradient sampling plot of a soil class picture; (2) a lateral gradient sampling map of a rock-like picture; (3) a transverse gradient sampling graph of the soil type picture; (4) longitudinal gradient sampling graph of rock class image.
In fig. 9, (1) a gradient map of a pebble image; (2) a gradient map of the oo-boulder image; (3) a gradient map of the mud egg gravel image; (4) gradient map of sandstone sandwich rock image.
In fig. 10, (1) a lateral curvature map of a pebble image; (2) a lateral curvature map of the oo-boulder image; (3) a transverse curvature map of the mud egg gravel image; (4) a transverse curvature map of the sandstone sandwich rock image.
In fig. 11, (1) a longitudinal curvature map of a pebble image; (2) a longitudinal curvature map of the oo-boulder image; (3) a longitudinal curvature map of the mud egg gravel image; (4) longitudinal curvature map of sandstone sandwich rock image.
In fig. 12, (1) a raw color map of a pebble image; (2) a pebble image binary image; (3) a pebble image area threshold post-filter map; (4) a pebble image post-operation chart.
In fig. 13, (1) a pebble image connected information extraction map; (2) The oval-boulder image communication information extraction diagram is characterized in that blue asterisks are centroids of all communication areas, and red dotted rectangular boxes are minimum box selection rectangles.
The invention takes two major classes of pebbles, gravel and sandstone splints as two major classes of secondary classification of rock, pebbles and pebbles as two minor classes of tertiary classification of rock, and silty clay and plain fill as major classes of secondary classification of soil as examples, and discloses a soil and rock classification method based on image processing, wherein the process is shown in figure 1, and the classification method comprises the following steps:
A. the following processing is performed on each of the acquired original images:
A1. and carrying out first preprocessing on the acquired original image to obtain a corresponding first binarized image.
Generally, for an image shot in the environment, which is a color image (black-and-white image loses information), as shown in fig. 2, a first preprocessing process converts an acquired color original image into a gray image, then finds a threshold value for maximizing a gray value difference between two parts of the gray image according to an Otsu global threshold value determination method, and then divides the gray image into two binary images by applying the threshold value, wherein the binary images are the first binary images.
A2. And calculating the transverse gradient and the longitudinal gradient of the first binarized image, and calculating the gradient sampling duty ratio corresponding to the original image according to the transverse gradient and the longitudinal gradient.
The horizontal and vertical gradients of the image represent the difference of the two-dimensional matrix of the image in the horizontal and vertical directions respectively, and can reflect the density degree of the texture of the object in the image. The image transverse and longitudinal gradient calculation formulas are shown as formulas (1) and (2).
Wherein G is x 、G y Is the calculated image horizontal and vertical gradient, and F (x, y) represents the pixel value at the position of the image (x, y). Fig. 3 shows a horizontal-vertical gradient diagram of the soil type picture and the rock type picture.
The horizontal and vertical gradients are randomly sampled and observed, and the horizontal (vertical) gradients of the soil type picture and the rock type picture are respectively collected and compared with the middle row (column), and the result is shown in fig. 4. As can be seen from fig. 4, for the soil type picture, the overall density is higher, the gradient calculated edge information is more, and the gradient is larger than the background (the point with the gradient of 0); the overall density of the rock pictures is low, the edge information calculated by gradients is less, and compared with the background, the gradient sampling duty ratio is adopted as a characteristic parameter to distinguish soil images from rock images.
The method for calculating the gradient sampling duty ratio comprises the following steps:
equidistant sampling is performed on the rows/columns of the lateral gradient. The sampling distance may be set according to the size of the gradient sampling map, and in some embodiments, the sampling distance is set to 5.
Calculating the ratio of the gradient information points of the sampling row/column to the information points in each row/column respectively: the ratio of the gradient information point (the point with the gradient of 1) to the total point number of each row (column), namely the duty ratio, is calculated for the sampling rows (columns), and the calculation formulas are shown in formulas (3) and (4).
Wherein F1 i 、F2 j Representing the duty cycle values of the sampling rows and columns, F x 、F y Respectively representing the transverse and longitudinal images of the image, and m and n respectively representing the rows and columns of the gradient image.
Performing unsupervised clustering second classification on the ratio of the sampling row/column, and taking the class with more clusters as the gradient sampling duty ratio corresponding to the original image: and performing unsupervised clustering of class 2 on the row (column) duty ratio values obtained by sampling, removing the classes with smaller quantity, so as to reduce interference items caused by image noise or other factors, and taking the classes with larger quantity as the final gradient sampling duty ratio of the image.
B. Performing unsupervised clustering second classification on gradient sampling duty ratios corresponding to all original images, and classifying environments corresponding to all original images into soil types and rock types at one time; and (3) performing the step C for secondary classification of the soil class, and performing the step D for secondary classification of the rock class.
The gradient sampling duty cycle of each image is used as a characteristic of classification. And randomly selecting gradient sampling duty cycle characteristics of 1 image from each of the soil class and the rock class as an initial clustering center, then calculating the distance between the characteristic of each image and each clustering center, and distributing the characteristic participating in the current calculation to the closest clustering center. The cluster centers and the features assigned to them represent a cluster. Every time a sample (feature) is allocated, the clustering center of the cluster is recalculated according to the existing objects in the cluster until the clustering center is not changed any more and the square sum of errors is minimum locally, so that the clustering is completed once. The pictures participating in the calculation are finally gathered into two types (soil type/rock type), so that the two classifications of the soil/rock are realized.
C. And classifying the soil corresponding to the original image according to the water content of the soil corresponding to the original image.
The water content of the soil is detected by a soil water content detection circuit.
As shown in fig. 7, the soil water content detection circuit comprises a square wave signal generation circuit, a sensing circuit, a rectifying circuit, a filtering circuit, a correction circuit, an amplifying circuit and a display circuit, wherein the square wave signal generation circuit, the sensing circuit, the rectifying circuit and the filtering circuit are sequentially connected, the input end of the correction circuit is connected with the power supply of the soil water content detection circuit, the output ends of the filtering circuit and the correction circuit are respectively connected with the two input ends of the amplifying circuit, and the output end of the amplifying circuit is connected with the display circuit.
The principle of the detection circuit is as follows: the soil moisture content and the soil resistivity form a linear relation, an oscillation signal is applied to the sensing circuit, and the sensing circuit outputs a corresponding voltage signal (namely, a detection voltage which is reduced along with the increase of the soil moisture content) due to the action of the soil resistivity, and the voltage signal is subjected to the action of a subsequent circuit to display a divided corresponding value on the display circuit. The square wave signal generating circuit of the detection circuit applies an oscillation signal to the sensing circuit, and the sensing circuit acts on the soil to be detected. Under the action of an oscillation signal, the output detection voltage of the sensing circuit sequentially passes through the rectifying circuit and the filter circuit and then is input into the amplifying circuit, the correction circuit sends the separated voltage to the amplifying circuit, the amplifying circuit usually adopts an operational amplifier, the filter circuit is connected to the inverting terminal of the operational amplifier, the correction circuit is connected to the non-inverting terminal of the operational amplifier, and the amplifying circuit outputs the amplified signal to the display circuit for numerical display. The display circuit is formed by connecting a digital voltmeter and a nixie tube circuit.
In one embodiment, in order to prevent the electrochemical corrosion phenomenon of the probes and reduce the influence of the distributed capacitance between the two probes, the soil moisture content is mainly in proportional relation with the soil resistivity, and the working voltage of the soil moisture content detection circuit adopts an oscillating signal with medium and low frequencies. As shown in FIG. 5, the design is made of U1 and R 2 、R 3 、D 1 、D 2 、C 2 、C 3 A square wave signal generator with the frequency of 1kHz is formed by the components and is used as the working voltage of the detection circuit. Square wave signal passes through coupling capacitor C 4 And an adjustable resistor R 4 Applied to the detection probe. Current limiting resistor R 5 And diode D 3 、D 4 The rectifier circuit is configured to reduce the output voltage as the soil moisture increases. The output voltage is R 6 And C 5 After the filter circuit is formed, the filter circuit is connected to the inverting terminal of U2C through a voltage follower U2A. Resistor R 8 Potentiometer R 9 And the voltage follower U2B forms a correction circuit, the amplification circuit is zeroed and corrected, the output of the correction circuit is connected to the non-inverting terminal of the U2C, and when the soil probe is not connected, the voltage of the non-inverting input terminal of the U2B is adjusted, so that the output voltage of the U2C is zero, and the zeroing function is realized. U2C is a differential amplifier with gain defined by resistor R 12 /R 7 And (5) determining. And finally, the output signal of the measuring circuit is transmitted to a digital voltmeter through a voltage follower U2D to display the final output voltage value, so that the numerical value of the soil water content can be directly read. Display electricThe digital voltmeter is connected with the nixie tube circuit, as shown in fig. 8, the digital voltmeter collects analog voltage by adopting ICL7107, the ICL7107 is a double-integration analog-to-digital converter, the ICL7107 is a very wide integrated circuit applied, and the digital voltmeter comprises a 3 1/2 bit digital-to-analog converter, can directly drive an LED nixie tube, and does not need a driving circuit. The ICL7107 is internally provided with functions of reference voltage, independent analog switch, logic switch, display drive, automatic zero setting and the like. The digital voltmeter can realize the measurement of 0-20V direct current voltage. The nixie tube circuit adopts a MAN6710 nixie tube, the eight-section nixie tube MAN6710 adopts a common anode connection method, the nixie tube can display 19.9V voltage at maximum, and automatic polarity display can be realized by inputting negative voltage.
As shown in fig. 7, this embodiment also discloses a soil environment detection system, including soil moisture content detection circuit, soil pressure detection circuit and display circuit, soil moisture content detection circuit includes square wave signal generating circuit, sensing circuit, rectifier circuit, filter circuit, correction circuit and amplifier circuit, and square wave signal generating circuit, sensing circuit, rectifier circuit and filter circuit connect gradually, the power of soil moisture content detection circuit is connected to the input of correction circuit, filter circuit with the output of correction circuit is connected respectively the two inputs of amplifier circuit, and the output of amplifier circuit is connected display circuit. The soil pressure detection circuit comprises a bridge circuit and an operational amplifier, wherein two arms of the bridge circuit are respectively connected to two ends of the operational amplifier, and the output end of the operational amplifier is connected with the display circuit.
As shown in FIG. 5, the soil moisture content detection circuit is designed by U1 and R 2 、R 3 、D 1 、D 2 、C 2 、C 3 A square wave signal generator with the frequency of 1kHz is formed by the components and is used as the working voltage of the detection circuit. Square wave signal passes through coupling capacitor C 4 And an adjustable resistor R 4 Applied to the detection probe. Current limiting resistor R 5 And diode D 3 、D 4 The rectifier circuit is configured to reduce the output voltage as the soil moisture increases. The output voltage is R 6 And C 5 After the filter circuit is formed, the filter circuit is connected to the inverting terminal of U2C through a voltage follower U2A. Resistor R 8 Potentiometer R 9 And the voltage follower U2B forms a correction circuit, the amplification circuit is zeroed and corrected, the output of the correction circuit is connected to the non-inverting terminal of the U2C, and when the soil probe is not connected, the voltage of the non-inverting input terminal of the U2B is adjusted, so that the output voltage of the U2C is zero, and the zeroing function is realized. U2C is a differential amplifier with gain defined by resistor R 12 /R 7 And (5) determining. And finally, the output signal of the measuring circuit is transmitted to the display circuit through a voltage follower U2D. As shown in fig. 6, the soil pressure detecting circuit is a bridge circuit composed of 4 ceramic piezoresistors, two arms of the bridge circuit are respectively connected to two input ends of an operational amplifier LM2904, and an output end of the operational amplifier is connected to a display circuit. The display circuit is formed by connecting a digital voltmeter and a nixie tube circuit, the digital voltmeter is used for acquiring analog voltage by adopting ICL7107, the ICL7107 is a double-integration analog-to-digital converter, the ICL7107 is a very wide integrated circuit applied to one block, and the display circuit comprises a 3 1/2-bit digital-to-analog converter, can directly drive an LED nixie tube and does not need a driving circuit. The ICL7107 is internally provided with functions of reference voltage, independent analog switch, logic switch, display drive, automatic zero setting and the like. The digital voltmeter can realize the measurement of 0-20V direct current voltage. The nixie tube circuit adopts a MAN6710 nixie tube, the eight-section nixie tube MAN6710 adopts a common anode connection method, the nixie tube can display 19.9V voltage at maximum, and automatic polarity display can be realized by inputting negative voltage.
D. Performing unsupervised clustering secondary classification on the texture complexity of the original image, and secondarily classifying the rock corresponding to the original image into a first rock type and a second rock type; for the first type of rock, step E is performed for three classifications.
The present example takes pebbles, and mud cobbles, sandstone splinters as the two classification subjects. The contour edge textures of pebbles and pebbles are smooth and regular, and the complexity is low; the edge textures of the profiles of the mud gravel and the sandstone splint rock are disordered, and the complexity is high, so that the rock can be divided into pebbles, pebbles and boulders and the sandstone splint rock according to the complexity of the textures.
In the embodiment, a profile complexity, a transverse curvature coefficient and a longitudinal curvature coefficient form a quantity to a profile feature set to represent texture complexity of a rock image, and the texture complexity is taken as a feature of object classification. And (3) calculating the gradients in the transverse and longitudinal directions of the image according to the formulas (1) and (2) to obtain the edge texture information of the original image, and representing the contour complexity of the image by using the proportion of edge information points (points with non-zero pixel values) to the total pixel points.
The calculation method of the contour complexity comprises the following steps:
D1. the lateral and longitudinal gradients of the original image are combined into a gradient map.
Obtaining a gradient G in the transverse and longitudinal directions of the picture from the formulas (1) and (2) X And G y Combining the two to obtain a global gradient map G, wherein a calculation formula is shown as a formula (5):
G＝G x +G y (5)
wherein G is the overall gradient map of the image, G x 、G y The lateral and longitudinal gradients of the images obtained by the formulas (1) and (2), respectively. An overall gradient map of the rock class picture is shown in fig. 9.
D2. Traversing the gradient map, counting the total number of points with non-zero pixel values.
Traversing the gradient graph G, finding out the point (edge information point) with the pixel value not being 0, and counting the total number N.
D3. The contour complexity is characterized by the proportion of the total number of points with non-zero pixel points to the total pixel points of the gradient map.
Calculating the proportion of the edge information points to the total pixel points, wherein a calculation formula is shown in a formula (6):
R＝N/(m*n) (6)
wherein R is the ratio of the edge information points to the total pixel points, N is the total number of the edge information points, and m and N are the row and column numbers of the gradient image G respectively.
The transverse curvature coefficient is obtained by calculating the curvature of the transverse gradient of the original image. The longitudinal texture difference degree of the image is reflected, and the calculation process comprises the following steps:
to be of original imageThe lateral gradient binarization obtains a lateral gradient image: obtaining the lateral gradient G of the image according to formula (1) x And binarizes it.
And filtering the transverse gradient image to filter out a connected region with the area smaller than a first threshold value so as to reduce the influence of the connected region of the non-rock part in the background on image processing. Comprising the following steps:
a. traversing the image and determining a connected region in the image.
b. The area of each communication area is calculated.
c. And setting an area threshold (namely a first threshold) through multiple experiments, filtering out a communication area with an area smaller than the threshold, and realizing a filtering effect.
And calculating the curvature of the filtered transverse gradient image to obtain a transverse curvature map. The calculation formulas are shown in formulas (7) - (15):
G down ＝(G x1 2 +G x2 2 ) 1.5 (13)
G up ＝G x2 2 ·G x1x -2G x1 ·G x2x ·G x2 +G x1 2 ·G x2y (14)
K 1 ＝G up /G down (15)
wherein G is x 、G y The gradients in the transverse and longitudinal directions of the pictures obtained according to the formulas (1) and (2); k (K) 1 The lateral curvature of the image is calculated. The transverse curvature diagram is shown in fig. 10.
And taking the variance value of the point with the pixel point not being zero in the transverse curvature map as a transverse curvature coefficient. The invention uses the variance value of the effective information point (point which is not 0) in the transverse curvature as the transverse curvature coefficient to represent the longitudinal edge texture difference degree of the image. The calculation process comprises the following steps:
a. traversing the transverse curvature graph, finding out effective information points in the transverse curvature graph, and counting the number N of the effective information points 1 。
b. The mean value of the effective information points is calculated, and the calculation formula is shown as a formula (16):
wherein K is mean To calculate the effective point mean value, K 1 For the transverse curvature of the image, N 1 The number of the effective points.
c. And (3) circularly traversing the effective information points of the transverse curvature map, and calculating the variance of the effective information points, wherein a variance calculation formula is shown as a formula (17):
K s ＝∑(K 1x -K mean ) 2 /N 1 (17)
wherein K is s To calculate the effective information point variance, K 1x To traverse K 1 All available information points, K mean N is the mean value of effective information points 1 The number of the effective points.
The longitudinal curvature coefficient is obtained by calculating the curvature of the longitudinal gradient of the original image. Reflecting the imageThe calculation process is basically the same as the extraction of the transverse curvature coefficient, and only G in the formulas (7) - (17) x Change to G y Thus obtaining the product. Fig. 11 is a longitudinal curvature map of a rock-like picture.
Based on the above, the texture complexity of the original image is subjected to unsupervised clustering two-classification, 1 image vector profile feature set is randomly selected from pebbles, pebble, mud, pebble and sandstone splitted rock as an initial clustering center, then the distance between the feature set of each image and each clustering center is calculated, and the feature set participating in the current calculation is distributed to the closest clustering center. The cluster centers and the feature sets assigned to them represent a cluster. Every time a sample (feature set) is allocated, the clustering center of the cluster is recalculated according to the existing objects in the cluster until the clustering center is not changed any more and the square sum of errors is minimum locally, so that the clustering is completed once. The pictures participating in the calculation are finally gathered into two types (pebbles, mud, gravel and sandstone splitted rock), so that the secondary classification of the rock is realized.
E. Performing second preprocessing on the original image to obtain a corresponding second binarized image, and performing unsupervised clustering second classification on rocks corresponding to the original image by taking the connected domain duty ratio and the mass center boundary distance confusion degree of the second binarized image as characteristics.
After the classifying step, the rock images are further subdivided into two major categories of pebbles, pebbles and pebbles, and sandstone splint rocks, wherein the pebbles and the pebbles are large in texture difference and easy to distinguish, and the pebbles are small in texture difference and difficult to distinguish, so that an automatic classifying method of the pebbles and the pebbles is further designed aiming at the difficulty.
According to definition, pebbles refer to natural oval particles without edges and corners, wherein the natural oval particles are formed by long-term transportation of weathered rock through water flow, and the particle size of the natural oval particles is 60-200 mm; the oo-boulder is rock with particle content of more than 200mm and more than 50% of total weight and mainly round and sub-round particles; in addition, by comparing pebble and pebble images to be easily obtained, pebble shapes are closer to circles, and the edges and corners of the pebbles are more obvious, so that the difference between pebbles and pebbles on the images is mainly represented as the shapes, accordingly, a profile estimation feature set is provided herein to represent the similarity of the shapes of objects and the circles and is used as the feature for image classification.
The process of performing the second preprocessing on the original image to obtain the corresponding second binarized image includes: performing first preprocessing on the original image to obtain a corresponding first binarized image; and calculating a connected region in the first binarized image, filtering the connected region with the area smaller than a third threshold value in the first binarized image, and performing opening operation on the rest part to obtain a second binarized image.
Specifically, the read color image is converted into a gray image, a threshold value which maximizes the gray value difference between two parts of the gray image is found out according to the Otsu global threshold value determination method, and the gray image is divided into two-value images by applying the threshold value. The binary image is subjected to image filtering, and the filtering method comprises the following steps:
a. traversing the image and determining a connected region in the image.
b. The area of each communication area is calculated.
c. And setting an area threshold (namely a third threshold) through multiple experiments, and filtering out the connected areas with the area smaller than the threshold.
d. And (3) performing an opening operation on the image, further eliminating noise points, and enabling the boundary of each communication area to be clearer.
The process of image filtering is shown in fig. 12. As can be seen from fig. 12, after the first area threshold filtering, some non-pebble interference areas are filtered, but some noise points and some partial areas are still adhered, after the opening operation, the noise points are further filtered, and meanwhile, the boundaries of the connected areas are clearer. As can be seen from fig. 12, each communication area of the pebble image is approximately circular or elliptical, has high fitting degree with the rectangular frame, and has a larger duty ratio in the rectangular frame; and each communication area of the oo-boulder image has a corner sense, the fitting degree with the rectangular frame is lower, and the occupation ratio in the rectangular frame is also smaller. Therefore, it is possible to use the connected domain duty ratio as a classification element.
In some embodiments, the connected domain duty ratio is an area ratio of an area of each connected domain to a corresponding minimum frame rectangle in the second binarized image. The calculation process comprises the following steps:
a. and (3) extracting connectivity information: traversing the filtered image, determining each connected region in the image, and combining with a regiolprops function in a MATLAB image processing toolbox to obtain the mass center C and the area S of each connected region A And the minimum box selects the position information of the rectangular boundingbox. The result of the connectivity information extraction is shown in fig. 13. In fig. 13, the centroid of each connected region is marked in the form of a blue asterisk, and the minimum box selection rectangle is marked in the form of a red dotted rectangular box.
b. Extracting the duty ratio of the connected domain: calculating the length and width of the binding box by using the position information of the binding box obtained during the extraction of the communication information, and further calculating to obtain the area S b . The calculation formula of the duty ratio of the connected domain is shown in formula (18).
S R ＝S A /S b (18)
Wherein S is R Calculating the calculated duty ratio of the connected domain; s is S A For the area of the connected region obtained during connected information extraction, S b And selecting a rectangular area for the corresponding minimum frame.
Because the more circular the contour of the object is, the closer the distances from each point of the contour to the centroid, i.e. the smaller the variance, based on this principle, in some embodiments, the method for calculating the distance confusion of the centroid boundary includes:
and extracting the outline of the connected domain in each minimum frame selected rectangle in the second binarized image.
And calculating Euclidean distances from each point on each connected domain contour to the centroid, and taking the variance value of each Euclidean distance as the boundary distance confusion of the centroid.
Specifically, the method for calculating the distance confusion of the centroid boundary comprises the following steps:
a. and extracting the connected domain outline in each binding box by using a Sobel edge extraction method, wherein the calculation formula is as follows:
G sx ＝[F b (x+1，y-1)+2F b (x+1，y)+F b (x+1，y+1)]-[F b (x-1，y-1+2Fbx-1，y+Fbx-1，y+1] (19)
G sy ＝[F b (x-1，y-1)+2F b (x，y-1)+F b (x+1，y-1)]-[F b (x-1，y+1+2Fb(x，y+1)+Fb(x+1，y+1)] (20)
θ s ＝arctan(G sx /G sy ) (22)
wherein F is b Corresponding image areas for each bounding box (the pixel value of the connected area is 1, and the rest are 0); g s A binding box image gradient; θ s Is the gradient direction thereof; when gradient G s Beyond a certain threshold, the corresponding point will be set as an edge point.
b. The Euclidean distance from each point of the connected domain outline to the mass center is calculated, and the calculation formula is as follows:
wherein D (i) represents the Euclidean distance from the ith contour point to the centroid, (x) i ，y i ) Represents the ith contour point coordinate, (C) x ，C y ) Representing centroid coordinates.
c. Calculating the variance of each distance
Wherein D is v For calculating the variance value of the distance between each contour point and the mass center, N edge Is the number of contour points.
The extracted two feature coefficients of the connected domain duty ratio and the centroid boundary distance confusion form a profile estimation feature set, and the profile estimation feature set is used as a classified feature to perform unsupervised clustering on rocks corresponding to the secondarily classified original images: and randomly selecting profile estimation feature sets of 1 image from pebbles and pebbles as initial clustering centers, calculating the distance between the feature set of each image and each clustering center, and distributing the feature set participating in the current calculation to the closest clustering center. The cluster centers and the feature sets assigned to them represent a cluster. Every time a sample (feature set) is allocated, the clustering center of the cluster is recalculated according to the existing objects in the cluster until the clustering center is not changed any more and the square sum of errors is minimum locally, so that the clustering is completed once. The pictures involved in the calculation are finally gathered into two classes (pebbles and pebbles), which enable further subdivision of the rock class.
The embodiment of the invention also provides a computer readable storage medium, which stores a computer program and runs the computer program to execute the soil and rock classification method.
The embodiment of the invention also provides a soil and rock classification device based on image processing, which comprises the computer readable storage medium, a processor and a soil water content detection circuit, wherein the processor is respectively connected with the computer readable storage medium and the soil water content detection circuit, the soil water content detection circuit detects the soil water content and transmits the detected result to the processor, and the processor runs a computer program in the computer readable storage medium to execute the soil and rock classification method.
The scheme in the embodiment of the invention can be applied to classifying various common soil and rocks, as shown in fig. 14, wherein (1) to (6) are sequentially silty clay, plain filled soil, pebble, oo-boulder, mud-pebble and sandstone sandwich rock, and the scheme (method or device) can automatically classify the enumerated soil or rocks. The invention can also be applied to classification of other types of soil and rock, the principle is the same, and the classification is not listed here.
The invention is not limited to the specific embodiments described above. The invention extends to any novel one, or any novel combination, of the features disclosed in this specification, as well as to any novel one, or any novel combination, of the steps of the method or process disclosed.
Claims (5)
1. A method of classifying soil and rock, comprising:
A. the following processing is performed on each of the acquired original images:
performing first preprocessing on the original image to obtain a corresponding first binarized image;
calculating the transverse gradient and the longitudinal gradient of the first binarized image, and calculating the gradient sampling duty ratio corresponding to the original image according to the transverse gradient and the longitudinal gradient;
the calculation formulas of the transverse gradient and the longitudinal gradient of the image are shown as (1) and (2);
（1）
（2）
wherein the method comprises the steps of、/>Is the calculated image transversal gradient, longitudinal gradient, < >>Representation of image->Pixel values at the locations;
the gradient sampling duty ratio calculation method comprises the following steps:
wherein F1 i 、F2 j Duty cycle values representing the sampling rows and columns, respectively;
performing unsupervised clustering second classification on the duty ratio values of the sampling rows/columns, and taking the class with more clusters as the gradient sampling duty ratio corresponding to the original image;
B. performing unsupervised clustering second classification on gradient sampling duty ratios corresponding to all original images, and classifying environments corresponding to all original images into soil types and rock types at one time; for the soil class, executing the step C to perform secondary classification, and for the rock class, executing the step D to perform secondary classification;
C. classifying the soil corresponding to the original image according to the water content of the soil corresponding to the original image;
D. performing unsupervised clustering secondary classification on the texture complexity of the original image, and secondarily classifying the rock corresponding to the original image into a first rock type and a second rock type; for the first rock, executing the step E to classify for three times;
the texture complexity comprises three characteristics of contour complexity, a transverse curvature coefficient and a longitudinal curvature coefficient;
the calculation method of the contour complexity comprises the following steps:
combining the transverse gradient and the longitudinal gradient of the original image into a gradient map,
traversing the gradient map, counting the total number of points with non-zero pixel values,
characterizing the complexity of the contour by the proportion of the total number of points with non-zero pixel points to the total pixel points of the gradient map;
the transverse curvature coefficient is obtained by calculating the curvature of the transverse gradient of the original image; the calculating of curvature of the lateral gradient of the original image includes:
binarizing the transverse gradient of the original image to obtain a transverse gradient image;
filtering the transverse gradient image to filter out a communication region with an area smaller than a first threshold value;
calculating the curvature of the filtered transverse gradient image to obtain a transverse curvature map;
taking the variance value of the point with the pixel point not being zero in the transverse curvature map as a transverse curvature coefficient;
the longitudinal curvature coefficient is obtained by calculating the curvature of the longitudinal gradient of the original image; the calculating curvature of the longitudinal gradient of the original image includes:
binarizing the longitudinal gradient of the original image to obtain a longitudinal gradient image;
filtering the longitudinal gradient image to filter out a communication region with an area smaller than a second threshold value;
calculating the curvature of the filtered longitudinal gradient image to obtain a longitudinal curvature map;
taking the variance value of the point with the pixel point not being zero in the longitudinal curvature map as a longitudinal curvature coefficient;
E. performing second preprocessing on the original image to obtain a corresponding second binarized image, and performing unsupervised clustering second classification on rocks corresponding to the original image by taking the connected domain occupation ratio and the mass center boundary distance confusion degree of the second binarized image as characteristics; the calculating method of the centroid boundary distance confusion degree comprises the following steps:
extracting the outline of the connected domain in each minimum frame selection rectangle in the second binarization image;
and calculating Euclidean distances from each point on each connected domain contour to the centroid, and taking the variance value of each Euclidean distance as the boundary distance confusion of the centroid.
2. The method for classifying soil and rock according to claim 1, wherein the step C detects the water content of the soil by using a soil water content detection circuit, the soil water content detection circuit comprises a square wave signal generating circuit, a sensing circuit, a rectifying circuit, a filtering circuit, a correction circuit, an amplifying circuit and a display circuit, the square wave signal generating circuit, the sensing circuit, the rectifying circuit and the filtering circuit are sequentially connected, an input end of the correction circuit is connected with a power supply of the soil water content detection circuit, an output end of the filtering circuit and an output end of the correction circuit are respectively connected with two input ends of the amplifying circuit, and an output end of the amplifying circuit is connected with the display circuit.
3. The method of classifying soil and rock according to claim 1, wherein said performing a second pre-treatment on the original image to obtain a corresponding second binarized image comprises:
performing first preprocessing on the original image to obtain a corresponding first binarized image;
and calculating a connected region in the first binarized image, filtering the connected region with the area smaller than a third threshold value in the first binarized image, and performing opening operation on the rest part to obtain a second binarized image.
4. The method of classifying soil and rock according to claim 1, wherein the connected domain ratio is an area ratio of an area of each connected domain to a corresponding minimum frame selected rectangle in the second binarized image.
5. A soil and rock classification device comprising a processor, a computer readable storage medium and a soil water content detection circuit, said processor being connected to said computer readable storage medium and said soil water content detection circuit, respectively; wherein the soil water content detection circuit is used for detecting the water content of soil, and the computer readable storage medium stores a computer program; the soil water content detection circuit transmits the results of the detection to the processor, which executes the computer program to perform the soil and rock classification method according to any one of claims 1, 3, 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011336766.9A CN112613531B (en) | 2020-11-25 | 2020-11-25 | Soil and rock classification method and device based on image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011336766.9A CN112613531B (en) | 2020-11-25 | 2020-11-25 | Soil and rock classification method and device based on image processing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112613531A CN112613531A (en) | 2021-04-06 |
| CN112613531B true CN112613531B (en) | 2023-12-12 |
Family
ID=75225076
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011336766.9A Active CN112613531B (en) | 2020-11-25 | 2020-11-25 | Soil and rock classification method and device based on image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112613531B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116540316B (en) * | 2023-07-06 | 2023-09-01 | 华设检测科技有限公司 | Geological Soil Layer Testing Method Based on SVM Classification Algorithm and Clustering Algorithm |
| CN117823232A (en) * | 2024-03-04 | 2024-04-05 | 邹城市宇光煤矿机械有限公司 | Coal mine production safety monitoring method and system and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106780485A (en) * | 2017-01-12 | 2017-05-31 | 西安电子科技大学 | SAR image change detection based on super-pixel segmentation and feature learning |
| CN110264459A (en) * | 2019-06-24 | 2019-09-20 | 江苏开放大学（江苏城市职业学院） | A kind of interstices of soil characteristics information extraction method |
| CN110427979A (en) * | 2019-07-10 | 2019-11-08 | 广东工业大学 | Road puddle recognition methods based on K-Means clustering algorithm |
| CN110866912A (en) * | 2019-11-15 | 2020-03-06 | 成都理工大学 | Shale streak layer heterogeneity data processing method based on imaging logging image texture |
| CN113838011A (en) * | 2021-09-13 | 2021-12-24 | 中南大学 | Rock block degree and/or distribution rule obtaining method, system, terminal and readable storage medium based on digital image color gradient |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9946931B2 (en) * | 2015-04-20 | 2018-04-17 | Los Alamos National Security, Llc | Change detection and change monitoring of natural and man-made features in multispectral and hyperspectral satellite imagery |
-
2020
- 2020-11-25 CN CN202011336766.9A patent/CN112613531B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106780485A (en) * | 2017-01-12 | 2017-05-31 | 西安电子科技大学 | SAR image change detection based on super-pixel segmentation and feature learning |
| CN110264459A (en) * | 2019-06-24 | 2019-09-20 | 江苏开放大学（江苏城市职业学院） | A kind of interstices of soil characteristics information extraction method |
| CN110427979A (en) * | 2019-07-10 | 2019-11-08 | 广东工业大学 | Road puddle recognition methods based on K-Means clustering algorithm |
| CN110866912A (en) * | 2019-11-15 | 2020-03-06 | 成都理工大学 | Shale streak layer heterogeneity data processing method based on imaging logging image texture |
| CN113838011A (en) * | 2021-09-13 | 2021-12-24 | 中南大学 | Rock block degree and/or distribution rule obtaining method, system, terminal and readable storage medium based on digital image color gradient |
Non-Patent Citations (3)
| Title |
|---|
| 基于遥感图像的山地冰川识别方法对比;范慧颖;董武;康宝生;张大奇;;水利水电技术(第05期);全文 * |
| 岩石颗粒图像中的斜长石识别;周磊;滕奇志;何海波;;现代计算机(专业版)(第03期);全文 * |
| 范慧颖 ; 董武 ; 康宝生 ; 张大奇 ; .基于遥感图像的山地冰川识别方法对比.水利水电技术.(第05期),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112613531A (en) | 2021-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lu et al. | Land cover classification in a complex urban-rural landscape with QuickBird imagery | |
| CN106910186B (en) | Bridge crack detection and positioning method based on CNN deep learning | |
| Detert et al. | Automatic object detection to analyze the geometry of gravel grains–a free stand-alone tool | |
| Guan et al. | Iterative tensor voting for pavement crack extraction using mobile laser scanning data | |
| CN108197583B (en) | Building change detection method based on graph cut optimization and image structure characteristics | |
| CN106557764B (en) | A kind of water level recognition methods based on binary-coded character water gauge and image procossing | |
| CN112613531B (en) | Soil and rock classification method and device based on image processing | |
| CN101539629B (en) | Remote sensing image change detection method based on multi-feature evidence integration and structure similarity | |
| CN110119438B (en) | Airborne LiDAR point cloud filtering method based on active learning | |
| CN109919944B (en) | Combined superpixel graph-cut optimization method for complex scene building change detection | |
| CN103971126A (en) | Method and device for identifying traffic signs | |
| CN106407924A (en) | Binocular road identifying and detecting method based on pavement characteristics | |
| CN103048329A (en) | Pavement crack detecting method based on active contour model | |
| CN109829423B (en) | Infrared imaging detection method for frozen lake | |
| Gao et al. | A robust pointer meter reading recognition method for substation inspection robot | |
| Yuan et al. | Learning to count buildings in diverse aerial scenes | |
| CN107392929A (en) | A kind of intelligent target detection and dimension measurement method based on human vision model | |
| CN110176020A (en) | A kind of bird's nest impurity method for sorting merging 2D and 3D rendering | |
| JP4747122B2 (en) | Specific area automatic extraction system, specific area automatic extraction method, and program | |
| CN115861409A (en) | Soybean leaf area measuring and calculating method, system, computer equipment and storage medium | |
| Mu et al. | McGET: A rapid image-based method to determine the morphological characteristics of gravels on the Gobi desert surface | |
| CN106815851A (en) | A kind of grid circle oil level indicator automatic reading method of view-based access control model measurement | |
| CN105488798B (en) | SAR image method for measuring similarity based on point set contrast | |
| Yan et al. | Quality analysis on ransac-based roof facets extraction from airborne lidar data | |
| CN117197661A (en) | Method for identifying loess water falling hole by utilizing laser radar point cloud data |
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 |