CN103942576A - Method for identifying coal and rock through airspace multiscale random characteristics - Google Patents
Method for identifying coal and rock through airspace multiscale random characteristics Download PDFInfo
- Publication number
- CN103942576A CN103942576A CN201410139075.8A CN201410139075A CN103942576A CN 103942576 A CN103942576 A CN 103942576A CN 201410139075 A CN201410139075 A CN 201410139075A CN 103942576 A CN103942576 A CN 103942576A
- Authority
- CN
- China
- Prior art keywords
- image
- coal
- rock
- primitive
- images
- 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.)
- Granted
Links
Abstract
The invention discloses a method for identifying coal and rock through airspace multiscale random characteristics. According to the method, a linear combination of a pixel gray value sum in randomly-generated image blocks of different numbers, sizes and positions is used for describing coal images and rock images; image characteristics of a coal training sample and a rock training sample are selected through the clustering algorithm to serve as an element dictionary, then, image characteristics of selected coal sample images and rock sample images are marked by the element dictionary according to a most-adjacent rule, an element frequency statistics regular column diagram of one sample image of the coal and the rock expresses one mode of the coal and the rock and the characteristics of the coal and the rock are expressed in multiple modes; in the process of identification, image characteristics are extracted and the column diagram is established according to images to be identified according to the method identical to that of the training images, the extracted characteristics and the column diagram are compared with those of the modes learned from the training stage, the distance of x<2> is used for measurement, and identification is conducted through the most-adjacent rule. The images of the different types of coal and rock at different light rays and different viewing points serve as the training samples, so that influence caused by illumination and the imaging viewpoint change is small, the method is not influenced by the change of the types of coal and rock, the identification rate is high, and stability is good.
Description
Technical field
The present invention relates to a kind of method with the multiple dimensioned random character identification in spatial domain coal petrography, belong to image recognition technology field.
Background technology
Coal and rock identify automatically identifies coal petrography stone object by a kind of method is coal or rock.In coal production process, coal and rock identify technology can be widely used in cylinder coal mining, driving, top coal caving, raw coal and select the production links such as spoil, for reducing getting working face operating personnel, alleviate labor strength, improve operating environment, to realize mine safety High-efficient Production significant.
Existing multiple coal and rock identify method, as natural Gamma ray probe method, radar detection method, stress pick method, infrared detecting method, active power monitoring method, shock detection method, sound detection method, dust detection method, memory cut method etc., but there is following problem in these methods: 1. need on existing equipment, install various kinds of sensors obtaining information additional, cause apparatus structure complicated, cost is high.2. violent, the serious wear of stressed complexity, vibration, dust are large in process of production for the equipment such as coal mining machine roller, development machine, and sensor is disposed more difficult, easily causes mechanical component, sensor and electric wiring to be damaged, and device reliability is poor.3. for dissimilar plant equipment, there is larger difference in the best type of sensor and the selection of picking up signal point, need to carry out personalized customization, the bad adaptability of system.
The existing coal and rock identify method based on image to image-forming condition as sensitivities such as illumination, viewpoints, if when the coal in coal to be identified or rock image imaging condition and when training or rock specimens image imaging condition are different, discrimination reduces greatly; In addition, if when coal to be identified, Rock Species change, the coal of need to resampling, rock specimens image are trained recognizer.
Need a kind of coal and rock identify method that solves or at least improve one or more problems intrinsic in prior art, to improve coal and rock identify rate and identification stability.
Summary of the invention
Therefore, the object of the present invention is to provide a kind of method with the multiple dimensioned random image feature identification in spatial domain coal petrography, this recognition methods be subject to illumination and imaging viewpoint variable effect little, the impact that not changed by coal, Rock Species, can in real time, automatically identify current coal, rock to liking coal or rock, for automated mining, automatic coal discharge, robotization, select the production runes such as cash reliable coal and rock identify information is provided.
1. according to a kind of embodiment form, coal and rock identify method of the present invention adopts following technical scheme to realize, and comprises the steps:
A. gathering respectively pixel size is the coal sample image set of w * h
with rock specimens image set
extract respectively every image I in two sample sets
cand I
rcharacteristics of image f
c∈ R
mand f
r∈ R
m, composing training collection
with
wherein, the characteristics of image f of described every image
c∈ R
mor f
r∈ R
mextraction principle is as follows:
(1) to every sample image, with one group of multi-scale filtering device { F of formula (1) definition
1,1..., F
w, hto its filtering
W, h is respectively width and the height of wave filter, by each filter response serial connection composing images proper vector P=(p
1, p
2... p
n)
t∈ R
n, n={wh}
2;
(2) P is carried out to f=Ψ P operation, wherein Ψ ∈ R
m * nfor random measurement matrix, its element r
ijfor:
ρ=n/10 α~n/6 α, α=constant, represents every a line nonzero element number in Ψ, f
i=∑
jr
ijp
j, f
i∈ R
1, f={f
1, f
2..., f
m}
t, m represents this characteristics of image dimension;
(3) f is normalized to unit length vector.
B. use K-means clustering method respectively to set of image characteristics
with
carry out cluster computing, obtain respectively Q cluster centre as primitive, be merged into the primitive dictionary TD of 2Q size;
C. respectively from coal sample image set
with rock specimens image set
in choose specific sample image M and open, by each characteristics of image f of each selected image
iprimitive mark with nearest with it in primitive dictionary TD, calculates the frequency that each primitive occurs, makes normalization histogram, is the primitive histogram of this image, and the M of coal opens the pattern of the primitive histogram formation coal of image
the M of rock opens the pattern of the primitive histogram formation rock of image
D. given unknown coal petrography object images Ix, uses the method identical with (3) with (1) in steps A, (2) to extract I
xcharacteristics of image f
x, use with method identical in step C and calculate I
xprimitive histogram h
x, use χ
2distance metric h
xwith coal and petromodel distance, if maxd is (h
x, h
c, m) < maxd (h
x, H
r, m), being coal, otherwise being rock, computing formula is as follows:
Accompanying drawing explanation
By following explanation, accompanying drawing embodiment becomes aobvious to be seen, its only with describe by reference to the accompanying drawings at least one preferably but the way of example of non-limiting example provide.
Fig. 1 is the basic procedure of coal and rock identify method of the present invention;
Embodiment
By the observation to coal, rock block sample, there is comparatively significantly difference in the contrast of the texture of coal and rock, is embodied in the degree of roughness of texture, sparse degree, the homogeneity of texture variations, the aspects such as the depth of rill.The realization that coal and the rock heterogeneite on texture is coal and rock identify provides condition precedent.Yet, image that imaging sensor becomes is not only relevant with article surface vein, reflectivity, imaging sensor itself, also relevant with illumination and imaging viewpoint, this makes on the one hand when illumination and imaging viewpoint change, same lump coal or image that rock becomes are very different, and under different image-forming conditions, to become image sometimes also to show closely similar for coal and rock on the other hand.Therefore, the present invention proposes a kind of method with the multiple dimensioned random image feature in spatial domain identification coal petrography, and object is also can effectively identify coal or rock when the light of become image and viewpoint change, to improve discrimination and to identify stability.
First the basic procedure of the method with the multiple dimensioned random image feature identification in spatial domain coal petrography is described.With reference to Fig. 1, concrete steps are as follows:
A. the sample image of taking common coal at different light, diverse location is as bituminous coal, stone coal etc., common roof and floor rock specimens image is as shale, sandstone etc., as each sample in the present invention is clapped 18, (camera position is constant, sample is placed on 3 diverse locations, variation illumination 3 times), from sample image, reduce the image that pixel size is w * h, each image after reducing is carried out to pre-service, as the present invention is processed into the gray level image of 0 average, unit deviation, make like this grey scale pixel value in image there is relative unchangeability to the variation of illumination; Coal after processing, rock specimens image form respectively coal sample image set
with rock specimens image set
extract respectively every image I in two sample sets
cand I
rcharacteristics of image f
c∈ R
mand f
r∈ R
m, composing training collection
with
wherein, the characteristics of image f of described every image
c∈ R
mor f
r∈ R
mextraction principle is as follows:
(1) to every sample image, with one group of multi-scale filtering device { F of formula (1) definition
1,1...
,f
w, hto its filtering
W, h is respectively width and the height of wave filter, by each filter response serial connection composing images proper vector P=(p
1, p
2... p
n)
t∈ R
n, n={wh}
2, wave filter size having been carried out to normalization here, normalized wave filter has been eliminated the impact of wave filter scale size, has also reflected topography's feature under different scale simultaneously.
As mentioned above, all filter responses are the proper vector of image, and characteristics of image dimension is very high like this, for reducing to calculate and storage burden, and don't lose image information, utilize compressed sensing principle to carry out dimensionality reduction to image feature space.
Element r in matrix
ijmeet the random measurement matrix Ψ ∈ R of formula (2)
m * n, m < < n has been proved to be the constraint isometry RIP that meets compressed sensing principle, and the random measurement matrix that meets RIP can catch the conspicuousness information in most original images.
(1) P is carried out to f=Ψ P operation, get ρ=L/10 α~L/6 α, α=constant, represents every a line nonzero element number in Ψ, f
i=∑
jr
ijp
jfor the weighted sum of filter response corresponding to nonzero element, weights are+1 or-1, filter response be grey scale pixel value in this filtering and, because wave filter is different scale, spatial distribution, so f
i∈ R
1, expressed the feature under a plurality of yardsticks that image space distributes, f={f
1, f
2..., f
m}
t, m has represented the characteristics of image dimension of this image.
In the present invention, get α=0.4, m=200;
Random measurement matrix Ψ is constant after generating, and acts on all training images and test pattern.
(2) by f
ibe normalized to unit length vector.
B. use K-means clustering method respectively to set of image characteristics
with
carry out cluster computing, obtain respectively Q cluster centre as primitive, as Q=20, be merged into the primitive dictionary TD of 2Q size, clustering criteria is:
B. respectively from coal sample image set
with rock specimens image set
the specific sample image M of middle extraction opens, and as extracted respectively 3 totally 12 of bituminous coal, stone coal, shale, sandstone, images, wherein the extraction of every 3 can be random.By the characteristics of image f of each selected image
iprimitive mark with nearest with it in primitive dictionary TD, calculates the frequency that each primitive occurs, makes normalization histogram, is the primitive histogram of this image, and the M of coal opens the pattern of the primitive histogram formation coal of image
the M of rock opens the pattern of the primitive histogram formation rock of image
A. given unknown coal petrography object images I
x, use the method identical with (3) with (1) in steps A, (2) to extract I
xcharacteristics of image f
x, use with method identical in step C and calculate I
xprimitive histogram h
x, use χ
2distance metric h
xwith coal and petromodel distance, if maxd is (h
x, h
c, m) < maxd (h
x, H
r, m), being coal, otherwise being rock, computing formula is as follows:
Claims (1)
1. by a method for the multiple dimensioned random character identification in spatial domain coal petrography, it is characterized in that, comprise the following steps:
A. gathering respectively pixel size is the coal sample image set of w * h
with rock specimens image set
extract respectively every image I in two sample sets
cand I
rcharacteristics of image f
c∈ R
mand f
r∈ R
m, composing training collection
with
wherein, the characteristics of image f of described every image
c∈ R
mor f
r∈ R
mextraction principle is as follows:
(1) to every sample image, with one group of multi-scale filtering device { F of formula (1) definition
1,1,..., F
w, hto its filtering
W, h is respectively width and the height of wave filter, by each filter response serial connection composing images proper vector P=(p
1,p
2... p
n)
t∈ R
n, n={wh}
2;
(2) P is carried out to f=Ψ P operation, wherein Ψ ∈ R
m * nfor random measurement matrix, its element r
ijfor:
ρ=n/10 α~n/6 α, α=constant, represents every a line nonzero element number in Ψ, f
i=∑
jr
ijp
j, f
i∈ R
1, f={f
1, f
2..., f
m}
t, m represents this characteristics of image dimension;
(3) f is normalized to unit length vector;
B. use K-means clustering method respectively to set of image characteristics
with
carry out cluster computing, obtain respectively Q cluster centre as primitive, be merged into the primitive dictionary TD of 2Q size;
C. respectively from coal sample image set
with rock specimens image set
in choose specific sample image M and open, by each characteristics of image f of each selected image
iprimitive mark with nearest with it in primitive dictionary TD, calculates the frequency that each primitive occurs, makes normalization histogram, is the primitive histogram of this image, and the M of coal opens the pattern of the primitive histogram formation coal of image
the M of rock opens the pattern of the primitive histogram formation rock of image
D. given unknown coal petrography object images Ix, uses the method identical with (3) with (1) in steps A, (2) to extract I
xcharacteristics of image f
x, use with method identical in step C and calculate I
xprimitive histogram h
x, use χ
2distance metric h
xwith coal and petromodel distance, if maxd is (h
x, h
cm) < maxd (h
x, H
rm), being coal, otherwise being rock, computing formula is as follows:
。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410139075.8A CN103942576B (en) | 2014-04-09 | 2014-04-09 | A kind of method that multiple dimensioned random character in use spatial domain recognizes coal petrography |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410139075.8A CN103942576B (en) | 2014-04-09 | 2014-04-09 | A kind of method that multiple dimensioned random character in use spatial domain recognizes coal petrography |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103942576A true CN103942576A (en) | 2014-07-23 |
CN103942576B CN103942576B (en) | 2017-06-16 |
Family
ID=51190239
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410139075.8A Active CN103942576B (en) | 2014-04-09 | 2014-04-09 | A kind of method that multiple dimensioned random character in use spatial domain recognizes coal petrography |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103942576B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104376313A (en) * | 2014-12-10 | 2015-02-25 | 中国矿业大学(北京) | Method for recognizing coal and rock by using local curve direction distribution of images |
CN105350963A (en) * | 2015-12-01 | 2016-02-24 | 中国矿业大学(北京) | Coal rock recognition method based on relativity measurement learning |
CN105447517A (en) * | 2015-11-20 | 2016-03-30 | 中国矿业大学(北京) | Airspace pyramid matching and identification coal rock method based on sparse coding |
CN106845509A (en) * | 2016-10-19 | 2017-06-13 | 中国矿业大学(北京) | A kind of Coal-rock identification method based on bent wave zone compressive features |
CN105973706B (en) * | 2016-06-07 | 2020-07-03 | 中国矿业大学(北京) | Coal rock mass multi-scale mechanical property analysis method based on industrial CT |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101515035A (en) * | 2009-03-16 | 2009-08-26 | 中国矿业大学(北京) | Top-coal caving law tracking instrument and a method for measuring the top-coal caving law |
CN102496004A (en) * | 2011-11-24 | 2012-06-13 | 中国矿业大学(北京) | Coal-rock interface identifying method and system based on image |
CN102521572A (en) * | 2011-12-09 | 2012-06-27 | 中国矿业大学 | Image recognition method of coal and gangue |
-
2014
- 2014-04-09 CN CN201410139075.8A patent/CN103942576B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101515035A (en) * | 2009-03-16 | 2009-08-26 | 中国矿业大学(北京) | Top-coal caving law tracking instrument and a method for measuring the top-coal caving law |
CN102496004A (en) * | 2011-11-24 | 2012-06-13 | 中国矿业大学(北京) | Coal-rock interface identifying method and system based on image |
CN102521572A (en) * | 2011-12-09 | 2012-06-27 | 中国矿业大学 | Image recognition method of coal and gangue |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104376313A (en) * | 2014-12-10 | 2015-02-25 | 中国矿业大学(北京) | Method for recognizing coal and rock by using local curve direction distribution of images |
CN104376313B (en) * | 2014-12-10 | 2017-05-17 | 中国矿业大学(北京) | Method for recognizing coal and rock by using local curve direction distribution of images |
CN105447517A (en) * | 2015-11-20 | 2016-03-30 | 中国矿业大学(北京) | Airspace pyramid matching and identification coal rock method based on sparse coding |
CN105350963A (en) * | 2015-12-01 | 2016-02-24 | 中国矿业大学(北京) | Coal rock recognition method based on relativity measurement learning |
CN105973706B (en) * | 2016-06-07 | 2020-07-03 | 中国矿业大学(北京) | Coal rock mass multi-scale mechanical property analysis method based on industrial CT |
CN106845509A (en) * | 2016-10-19 | 2017-06-13 | 中国矿业大学(北京) | A kind of Coal-rock identification method based on bent wave zone compressive features |
Also Published As
Publication number | Publication date |
---|---|
CN103942576B (en) | 2017-06-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103927514B (en) | A kind of Coal-rock identification method based on random local image characteristics | |
CN102509087B (en) | Coal-rock identification method based on image gray level co-occurrence matrixes | |
CN103927528B (en) | Coal and rock recognition method based on close neighborhood pixel gray level joint distribution characteristics | |
Tonini et al. | Rockfall detection from terrestrial LiDAR point clouds: A clustering approach using R | |
CN103942576A (en) | Method for identifying coal and rock through airspace multiscale random characteristics | |
CN103927553B (en) | Coal and rock recognition method based on multi-scale micro-lamination and contrast ratio joint distribution | |
Ersoz et al. | Crack identification for rigid pavements using unmanned aerial vehicles | |
CN102928435A (en) | Aircraft skin damage identification method and device based on image and ultrasound information fusion | |
CN102930253A (en) | Coal and rock identification method based on image discrete multi-wavelet transform | |
CN104751192B (en) | Coal-rock identification method based on image block symbiosis feature | |
US11670073B2 (en) | System and method for detection of carbonate core features from core images | |
CN105243401A (en) | Coal rock recognition method based on coal structure element study | |
CN104778461B (en) | Coal-rock identification method based on Similar measure study | |
CN113184483B (en) | Conveyer belt tearing prevention system and early warning method | |
CN105350963B (en) | A kind of Coal-rock identification method learnt based on relativity measurement | |
CN116561536B (en) | Landslide hidden danger identification method, terminal equipment and medium | |
CN104463098B (en) | With the structure tensor direction histogram feature recognition coal petrography of image | |
CN108035710B (en) | The method for dividing deep layer rock geology phase based on data mining | |
Mills et al. | Rock surface classification in a mine drift using multiscale geometric features | |
CN104751193A (en) | Distance constraint similarity based coal-rock interface recognition method | |
Lemy et al. | Image analysis of drill core | |
CN105354596B (en) | Coal-rock identification method based on structural sparse coding | |
Liu et al. | Intelligent recognition and identification of fracture types and parameters for borehole images based on developed convolutional neural networks and post-processing | |
CN114212483B (en) | CV algorithm-based scraper conveyor component fault identification method and device | |
Wang | Rock fracture image segmentation algorithms |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant |