CN108830831A - One kind is based on the improvement matched zinc flotation froth nature velocity characteristic extracting method of SURF - Google Patents
One kind is based on the improvement matched zinc flotation froth nature velocity characteristic extracting method of SURF Download PDFInfo
- Publication number
- CN108830831A CN108830831A CN201810446667.2A CN201810446667A CN108830831A CN 108830831 A CN108830831 A CN 108830831A CN 201810446667 A CN201810446667 A CN 201810446667A CN 108830831 A CN108830831 A CN 108830831A
- Authority
- CN
- China
- Prior art keywords
- point
- froth
- velocity
- matching
- image
- 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
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
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention proposes one kind based on the improvement matched zinc flotation froth nature velocity characteristic extracting method of SURF, choosing has scale invariability, rotational invariance and the stronger SURF feature operator of anti-noise ability carry out feature extraction to image, by the similitude and its specific motion profile of analyzing foam shape, it proposes to update matching process according to foam motion conditions real-time iterative, motion match is carried out using matching condition is improved, obtain zinc flotation froth initial velocity, matching result calculates the cycle movement for considering scraper plate in the initial separatory cell under actual conditions, selection underflow speed obtains flotation froth nature velocity characteristic.The present invention overcomes conventional foam VELOCITY EXTRACTION method matching errors it is big, speed is slow the problems such as, application speed improves 10 times or more, updates SURF operator matching process by the iteration of proposition and greatly improves 20% or more successful match rate.Solve the problems, such as zinc flotation site flotation froth feature be difficult to extract and how machine vision.
Description
Technical field
The invention belongs to froth flotation technical fields, and in particular to roughing foam velocity characteristic mentions in a kind of zinc floatation process
Take method.
Background technique
Froth flotation is one of most important method in current mining processing industry, is a kind of physics using mineral particle surface
Chemical property difference causes hydrophily different, and then to the method that mineral are sorted, has very strong practical value.Floating
During choosing, using the physical motion and control air inflow of agitator, can be formed largely has different sizes, color, form
And the bubble of Texture eigenvalue, mineral grain are attached on bubble surface and then realize sorting mineral.The grade value of zinc concentrate and
The fine or not situation of rate of recovery reaction operating condition.And floatation foam image is characterized in judging an important evidence of flotation operating condition, it is wrapped
Contain largely information related with performance variable and production technology.Factory is selected to rely primarily on expertise observation flotation device for many years
The foam state on ore pulp surface, to adjust flotation machine mineral pulp level and change regime of agent.Production status is carried out by rule of thumb
Evaluation cannot achieve the objective cognition of the comprehensive operating condition of whole process.Although selecting factory that can obtain concentrate product by offline assay
Position, but with the flowing of production line, operating condition is fluctuated, and the method is affected by human factor, and checkout procedure is multiple
It is miscellaneous and at high cost.Again since flotation process is complicated, tediously long, influence factor is more, the on-line checking of concentrate grade cannot achieve
And real time monitoring, the instant adjustment to dosage and other parameters is affected, the rate of recovery of mineral is finally affected.Therefore it grinds
Study carefully and how to extract effective zinc floatation foam image feature, so that facing to manufacture index optimization carries out real-time online detection, to guidance
Production operation and the optimization stable operation of process have great importance.
With the fast development of computer technology, digital image processing techniques, by the soft-measuring technique based on machine vision
Real-time monitoring applied to floatation process to floatation indicators brings new breakthrough, obtains more flotation relevant to grade
Index.In the actual production process, experience worker is worked as by the features judgement such as speed, stability, color of observation froth images
Preceding operating condition is simultaneously adaptively adjusted.The automation steady production for realizing zinc flotation first has to obtain by machine vision
Effective information, by information knowledge, regularization, to realize the real time monitoring of technique.The study found that flotation froth speed is
Judge the important channel of concentrate grade.Existing froth images velocity characteristic extracting method faces that robustness is low, matching result misses
The problems such as difference is big, matching speed is slow, for the distinctive shape similarity of foam, deformation quantity is big, easily merges the features such as collapsing, passes
The feature operator of system does not embody good adaptability；And the periodic physical rotation of scraper plate in actual process
It exerts a certain influence to foam nature flow velocity, in terms of the extraction of flotation froth velocity characteristic encounters many practical applications
The problem of.
Summary of the invention
The purpose of the present invention is more difficult for concentrate grade on-line checking in zinc floatation process, floatation foam image is special
Sign is difficult to accurate expression, proposes a kind of matched zinc flotation froth nature velocity characteristic extracting method of improvement SURF.
The present invention is given by following scheme and is realized：
One kind is based on improving the matched zinc flotation froth nature velocity characteristic extracting method of SURF, and this method includes：
S1：It is right by zinc flotation image data in the case of the different concentrate grade values of factory's flotation image capturing system acquisition
The data of acquisition are analyzed, remove wherein the data set of vacancy, schemed as caused by camera shake or other artificial origins
As unsharp data set；
S2：According to treated zinc flotation image data set and creation data, three different sample subspaces are formed,
Specific step is as follows：
(1) selected zinc flotation image data set is obtained, higher, normal, relatively low three operating conditions of grade are classified as；
(2) image of acquisition is classified, rejects the froth images that classification boundary interferes；
(3) final froth images data set is divided into V1, V2, V3 and respectively corresponds three kinds of different operating conditions；
S3：For each different sample subspace, SURF feature is extracted to its every frame froth images, specific steps have：
(1) it extracts kth frame image and is converted to gray level image, acquire image integration figure Ik；
(2) for integrogram IkIn give pointConstruct Hessian matrix:
Wherein Lxx(x, σ) is in the frame imageWith mean value for 0, the Gaussian function that standard deviation is σ leads to point as filter
It crosses convolution algorithm and calculates the direction x second-order partial differential coefficient, similarly Lxy(x,σ)、Lyy(x, σ), acquire matrix determinant det (H (x,
σ))；
(3) it takes the template size for changing filter to establish scale space pyramid, is used according to given image size
Three layers or four layers of filter；
(4) point in det (H (x, σ)) value and its 3 × 3 × 3 field after being calculated according to the Hessian matrix of each point
It is compared, primarily determines characteristic point using 3D- non-maxima suppression method, linear interpolation is recycled to obtain sub-pix
The characteristic point of grade, setting threshold value reduce detected characteristic point quantity, choose maximum value as final characteristic point；
(5) centered on choosing final characteristic point, 6s be building circle in radius, s is final characteristic point in step (3)
Scale when being detected, with the Haar small echo response d in the horizontal direction having a size of 4 σxWith the response d of vertical directionyWith
The Gaussian function that standard deviation is σ=2s weights, and obtains maximum response with the fan-shaped run-down in circle that central angle is 60 °
Direction as final characteristic point principal direction；
(6) principal direction is direction centered on final characteristic point, in step (5), 20s is window building region unit, then divides
For 4 × 4 sub-regions, Haar small echo response d in the horizontal direction is calculated to each subregionxWith the response of vertical direction
Value dyWith the sum of response absolute value ∑ | dx|、∑|dy| collectively constitute structure v=(∑ dx,∑dy,∑|dx|,∑|dy|), shape
SURF feature point description operator is tieed up at 4 × 4 × 4=64；
S4：For the froth images for extracting SURF feature, motion match is carried out using matching condition is improved, obtains zinc flotation
Foam initial velocity, specific step is as follows：
(1) the two continuous frames image for extracting sample subspace under L operating condition, carries out S3 operation, obtains the description of SURF feature and calculates
Sub- v0、 v1；
(2) v is defined according to Euclidean distance0And v1Between distance, obtain minimum distance d between two operators0With secondary short distance
d1, select the ratio of the twoAs judgment criteria, threshold value ratio is setT, as ratio < ratioTWhen, characteristic point
Otherwise successful match fails；
(3) it is gone according to the Feature Points Matching of two frame froth images as a result, carrying out Mismatching point using RANSAC algorithm to it
It removes；
(4) matching result after being mismatched a little according to removal, the pixel according to two field pictures match point is poor, obtains foam level
It moves, the removing unit time obtains initial velocityWherein c is the characteristic point of successful match
Number；
(5) it acquiresMiddle speed magnitude component average value SPE_vel, durection component average value SPE_ang, weight selection
Coefficient ω0, ω1, matching criteria in step (2) is updated, if ratio < ratioTAndThen
Successful match, otherwise it fails to match, whereinFor current matching point initial velocity magnitude component average value,
For current matching point initial velocity durection component average value；
(6) the foam initial velocity of two continuous frames after being calculated according to step (1)-(5), until the subspace froth images speed
Degree feature extraction finishes, and obtains flotation froth speed initial velocity under L operating condition
S5：The cycle movement for considering scraper plate in initial separatory cell under practical condition is calculated according to matching result, selects bottom
Flow velocity degree obtains flotation froth nature velocity characteristic, and specific step is as follows：
(1) for each component in flotation froth initial velocity vectorIt is normalized, obtains velocity component
Histogram；
(2) average is sought in 80% or more velocity component section in access speed histogram of componentIt obtains
(3) according to middle_speedLBy quadratic linear interpolation, flotation froth rate curve is obtained；
(4) all minimum points of rate curve in step (3) are found out, its average is taken to obtain final speed
Wherein xiFor i-th of minimum point, n is the number of minimum point；
S6：The velocity characteristic that all froth images in three sample subspaces are extracted according to above step, forms the sample
Under the conditions of velocity characteristic collection.
The invention proposes one kind based on the matched zinc flotation froth nature velocity characteristic extracting method of SURF is improved, and solves
Zinc flotation site flotation froth feature be difficult to extract and how machine vision the problem of；The present invention is directed to froth images
The characteristics of, choosing, there is scale invariability, rotational invariance and the stronger SURF feature operator of anti-noise ability to carry out to image
Feature extraction improves the matching speed between froth images compared to traditional sub-block registration and phase method for registering；Due to floating
Selecting foam is the projection imaging from 3D to 2D, and foam similarity is high, deformation ratio is high, collapsing rate is high, existing SURF matching algorithm
Being directly applied to flotation froth scene, to mismatch rate higher, cannot implement to be effectively matched, therefore pass through the similar of analysis foam shape
Property and its specific motion profile, propose to update matching process according to foam motion conditions real-time iterative, substantially increase
With rate；Consider that the periodical power of scraper plate in actual process acts on bring velocity noise, proposes to be changed according to speed period
Underflow speed is chosen as foam nature speed.Big, speed that the present invention overcomes conventional foam VELOCITY EXTRACTION method matching errors
The problems such as slow, improves 10 times or more compared to application speed of traditional SIFT operator under the scene, is updated by the iteration of proposition
SURF operator matching process greatly improves 20% or more successful match rate, is convenient for real-time operation at the scene.
Detailed description of the invention
Fig. 1 is the flow chart that zinc flotation froth nature velocity characteristic extracts during the present invention is implemented.
Specific embodiment
Here is to combine attached drawing of the present invention, in further detail, is clearly made that technical solution employed in the present invention
It describes and explains.The present invention is more difficult for concentrate grade on-line checking in zinc floatation process, and floatation foam image feature is difficult
With accurate expression, the present invention utilizes the velocity characteristic closely related with zinc flotation yield, zinc Floatation Concentrate Grade, proposes one kind
Improve the matched zinc flotation froth nature velocity characteristic extracting method of SURF.Obviously, described embodiment is only of the invention real
A part in example is applied, is not the whole of embodiment.Based on the embodiments of the present invention, those skilled in the relevant art exist
The premise for not making creative work obtains all other embodiment and all should be protection scope of the present invention.
As shown in Figure 1, improving the matched zinc flotation froth nature velocity characteristic of SURF for one of embodiment of the present invention
Extracting method, this method comprise the following specific steps that：
S1：Zinc flotation image data in the case of different concentrate grade values is obtained by factory's flotation image capturing system, according to
According to expertise, the data of acquisition are analyzed, remove distracter therein, such as the data set of vacancy, due to camera shake
Or the data set of fogging image caused by other artificial origins.
S2：According to treated zinc flotation image data set and creation data, three different sample subspaces are formed,
Specific step is as follows：
(1) it obtains selected zinc flotation image data set and product is classified as according to creation data index and expertise
Position higher (zinc concentrate grade is higher than 55%), normal (zinc concentrate grade is 54% or so), relatively low (zinc concentrate grade is lower than
53%) three operating conditions；
(2) image after selecting acquisition is classified, and the froth images that classification boundary interferes are rejected；
(3) final froth images data set is divided into V1, V2, V3 and respectively corresponds that grade is higher, grade is normal, grade is relatively low
Three kinds of different operating conditions.
S3：For each different sample subspace, SURF feature is extracted to its every frame froth images, specific steps have：
(1) it extracts kth frame image and is converted to gray level image, traversal acquires image integration image Ik；
(2) for image IkIn give pointConstruct Hessian matrix：
Wherein Lxx(x, σ) is in the frame imageWith mean value for 0, the Gaussian function that standard deviation is σ leads to point as filter
It crosses convolution algorithm and calculates the direction x second-order partial differential coefficient, similarly Lxy(x,σ)、Lyy(x, σ), acquires matrix determinant：
Det (H (x, σ))=DxxDyy-(wDxy)2
Wherein Dxx、Dyy, come the Gaussian function approximate solution of approximate σ=1.2, to can simplify fortune with 9 × 9 box filter
It calculates, wherein w is calculated by the following formula：
Wherein | * |FFor Frobenius norm；
(3) gaussian pyramid scale space is constructed, the template size for changing filter is taken to establish scale space gold word
Tower, original image do not need down-sampling, use three layers or four layers of filter according to given image size, wherein first layer is filtered
Wave device template size is 9 × 9,15 × 15,21 × 21,27 × 27, and adjacent forms differ 6 pixels, second layer filter template
Size is 15 × 15,27 × 27,39 × 39,51 × 51, and adjacent forms differ 12 pixels, third layer filter template size
It is 27 × 27,51 × 51,75 × 75,99 × 99, adjacent forms differ 24 pixels, and the 4th layer of filter template size is 51
× 51,99 × 99,147 × 147,195 × 195, adjacent forms differ 48 pixels；
(4) det (H (x, σ)) value is calculated according to the Hessian matrix of each point, reasonable threshold value is set, less than the threshold value
Point is removed, and recycles 3D- non-maxima suppression method by the click-through in Hessian matrix response and its 3 × 3 × 3 field
Row compares, and primarily determines characteristic point, and the characteristic point for recycling linear interpolation fitting to obtain sub-pixel is compared, and selects
It is maximized as final characteristic point；
(5) centered on choosing final characteristic point, 6s be building circle in radius, s is final characteristic point in step (3)
Scale when being detected, with the Haar small echo response d in the horizontal direction having a size of 4 σxWith the response d of vertical directionyWith
The Gaussian function that standard deviation is σ=2s weights, and obtains maximum response with the fan-shaped run-down in circle that central angle is 60 °
Direction as final characteristic point principal direction；
(6) principal direction is direction centered on final characteristic point, in step (5), 20s is window building region unit, then divides
For 4 × 4 sub-regions, Haar small echo response d in the horizontal direction is calculated to each subregionxWith the response of vertical direction
Value dyWith the sum of response absolute value ∑ | dx|、∑|dy| collectively constitute structure v=(∑ dx,∑dy,∑|dx|,∑|dy|), shape
SURF feature point description operator is tieed up at 4 × 4 × 4=64；
S4：For the froth images for extracting SURF feature, motion match is carried out using matching condition is improved, obtains zinc flotation
Foam initial velocity, specific step is as follows：
(1) the two continuous frames image for extracting sample subspace under L operating condition, carries out S3 operation, obtains the description of SURF feature and calculates
SonWherein m and n are respectivelyWithThe number of middle characteristic point；
(2) it is calculated according to Euclidean distanceWithBetween similar features vector, obtain minimum distance between two characteristic values
d0=min (dis (v0i,v1j)), wherein dis (v0i,v1j) it is v0iWith v1jBetween Euclidean distance, obtain nearest Euclidean distance
With point, two o'clock is removed, then finds out minimum distance d1, select the ratio of the twoAs judgment criteria, threshold value is set
ratioT, as ratio < ratioTWhen, otherwise Feature Points Matching success fails；
(3) it is gone according to the Feature Points Matching of two frame froth images as a result, carrying out Mismatching point using RANSAC algorithm to it
It removes, the point set of the successful match obtained in step (2)Random sampling T1A matching corresponding points, wherein T1∈ [4,
20], linear equation is calculated, then set of computationsInterior exterior point distance threshold d is arranged to the distance of the straight line in left pointT
(being obtained by many experiments), it is all to be less than dTThe point of distance is determined as interior point, remaining is exterior point, and optimal interior number is arranged
T2, wherein For the point set of successful matchNumber, when interior number be greater than T2
When, determine that the model is optimal, and calculate modeling statistics error e rrormin, the process is repeated, iteration more new model is simultaneously counted
Statistical error is calculated, so that errorminMinimum, model is optimal, to rejectIn mismatch a little；
(4) matching result after being mismatched a little according to removal, the pixel according to two field pictures match point is poor, obtains each
It is displaced with foam between point, the removing unit time obtains initial velocityWherein c is matching
Successful feature point number；
(5) it acquiresMiddle speed magnitude component average value SPE_vel, durection component average value SPE_ang choose power
Weight coefficient ω0, ω1, matching criteria in (2) is updated, if ratio < ratioTAndThen
With success, otherwise it fails to match, whereinFor current matching point initial velocity magnitude component average value,
For current matching point initial velocity durection component average value；
(6) the foam initial velocity of two continuous frames after being calculated according to step (1)-(5), until the subspace froth images speed
Degree feature extraction finishes, and obtains flotation froth speed initial velocity under L operating condition
S5：It is calculated according to matching result and considers practical condition lower scraping plate dynamic speed, it is naturally fast to obtain flotation froth
Feature is spent, specific step is as follows：
(1) for each component in flotation froth initial velocity vectorIt is normalized, obtains velocity component
Histogram；
(2) average is sought in 80% or more velocity component section in access speed histogram of componentIt obtains
(3) according to middle_speedLBy quadratic linear interpolation, flotation froth rate curve is obtained；
(4) cycle movement for considering scraper plate in initial separatory cell selects underflow speed as flotation froth nature speed, rejects
Outlier finds out all minimum points of rate curve in (3), its average is taken to obtain final speed
Wherein xiFor i-th of minimum point, n is the number of minimum point.
S6：The velocity characteristic that all froth images in three sample subspaces are extracted according to above step, forms the sample
Under the conditions of velocity characteristic collection.
The claimed range of the present invention is not limited only to the description of present embodiment.
Claims (2)
1. one kind is based on the improvement matched zinc flotation froth nature velocity characteristic extracting method of SURF, it is characterised in that including following
Step：
S1：By zinc flotation image data in the case of the different concentrate grade values of factory's flotation image capturing system acquisition, to acquisition
Data analyzed, remove wherein that the data set of vacancy, the image as caused by camera shake or other artificial origins be not
Clearly data set, obtain that treated zinc flotation image data set and creation data；
S2：According to treated zinc flotation image data set and creation data, three different sample subspaces are formed, specifically
Steps are as follows：
(1) selected zinc flotation image data set is obtained, higher, normal, relatively low three operating conditions of grade are classified as；
(2) image of acquisition is classified, rejects the froth images that classification boundary interferes；
(3) final froth images data set is divided into V1, V2, V3 and respectively corresponds three kinds of different operating conditions；
S3：For each different sample subspace, SURF feature is extracted to its every frame froth images, specific steps have：
(1) it extracts kth frame image and is converted to gray level image, acquire image integration figure Ik；
(2) for integrogram IkIn give pointConstruct Hessian matrix:
Wherein Lxx(x, σ) is in the frame imageFor point with mean value for 0, standard deviation is the Gaussian function of σ as filter, passes through volume
Product operation calculates the direction x second-order partial differential coefficient, similarly Lxy(x,σ)、Lyy(x, σ) acquires matrix determinant det (H (x, σ))；
(3) it takes the template size for changing filter to establish scale space pyramid, uses three layers according to given image size
Or four layers of filter；
(4) point in det (H (x, σ)) value and its 3 × 3 × 3 field after being calculated according to the Hessian matrix of each point carries out
Compare, primarily determines characteristic point using 3D- non-maxima suppression method, linear interpolation is recycled to obtain the spy of sub-pixel
Point is levied, setting threshold value reduces detected characteristic point quantity, chooses maximum value as final characteristic point；
(5) centered on choosing final characteristic point, 6s be building circle in radius, s is that final characteristic point is detected in step (3)
Scale when survey, with the Haar small echo response d in the horizontal direction having a size of 4 σxWith the response d of vertical directionyWith standard
The Gaussian function that difference is σ=2s weights, and obtains the direction of maximum response with the fan-shaped run-down in circle that central angle is 60 °
As final characteristic point principal direction；
(6) principal direction is direction centered on final characteristic point, in step (5), 20s is window building region unit, it is further divided into 4 ×
4 sub-regions calculate Haar small echo response d in the horizontal direction to each subregionxWith the response d of vertical directionyAnd sound
The sum of absolute value ∑ should be worth | dx|、∑|dy| collectively constitute structure v=(∑ dx,∑dy,∑|dx|,∑|dy|), form 4 × 4 × 4
=64 dimension SURF feature point description operators；
S4：For the froth images for extracting SURF feature, motion match is carried out using matching condition is improved, obtains zinc flotation froth
Initial velocity, specific step is as follows：
(1) the two continuous frames image for extracting sample subspace under L operating condition, carries out S3 operation, obtains SURF feature and describe operator v0、
v1；
(2) v is defined according to Euclidean distance0And v1Between distance, obtain minimum distance d between two operators0With secondary short distance d1, selection
The ratio of the twoAs judgment criteria, threshold value ratio is setT, as ratio < ratioTWhen, Feature Points Matching at
Otherwise function fails；
(3) according to the Feature Points Matching of two frame froth images as a result, carrying out Mismatching point removal using RANSAC algorithm to it；
(4) matching result after being mismatched a little according to removal, the pixel according to two field pictures match point is poor, obtains foam displacement,
The removing unit time obtains initial velocityWherein c is the feature point number of successful match；
(5) it acquiresMiddle speed magnitude component average value SPE_vel, durection component average value SPE_ang, weight selection coefficient
ω0, ω1, matching criteria in step (2) is updated, if ratio < ratioTAndThen matching at
Function, otherwise it fails to match, whereinFor current matching point initial velocity magnitude component average value,It is current
Match point initial velocity durection component average value；
(6) the foam initial velocity of two continuous frames after being calculated according to step (1)-(5), until the subspace froth images speed is special
Sign is extracted and is finished, and flotation froth speed initial velocity under L operating condition is obtained
S5：The cycle movement for considering scraper plate in initial separatory cell under practical condition is calculated according to matching result, selects underflow speed
Degree obtains flotation froth nature velocity characteristic, and specific step is as follows：
(1) for each component in flotation froth initial velocity vectorIt is normalized, obtains velocity component histogram
Figure；
(2) average is sought in 80% or more velocity component section in access speed histogram of componentIt obtains
(3) according to middle_speedLBy quadratic linear interpolation, flotation froth rate curve is obtained；
(4) all minimum points of rate curve in step (3) are found out, its average is taken to obtain final speed
Wherein xiFor i-th of minimum point, n is the number of minimum point；
S6：The velocity characteristic for extracting all froth images in three sample subspaces, forms the velocity characteristic under the sample conditions
Collection.
2. a kind of according to claim 1 be based on improving the matched zinc flotation froth nature velocity characteristic extracting method of SURF,
It is characterized in that：The S4 to froth images extract SURF feature after improve matching algorithm the step of (2) in parameter
ratioT, parameter ω in step (5)0、ω1Value range is respectively：ratioT∈ [0.55,0.85], ω0∈ [1.2,2], ω1
∈ [0.5,1.5].
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810446667.2A CN108830831B (en) | 2018-05-11 | 2018-05-11 | Zinc flotation foam natural speed feature extraction method based on improved SURF matching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810446667.2A CN108830831B (en) | 2018-05-11 | 2018-05-11 | Zinc flotation foam natural speed feature extraction method based on improved SURF matching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108830831A true CN108830831A (en) | 2018-11-16 |
| CN108830831B CN108830831B (en) | 2021-11-30 |
Family
ID=64147912
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810446667.2A Active CN108830831B (en) | 2018-05-11 | 2018-05-11 | Zinc flotation foam natural speed feature extraction method based on improved SURF matching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108830831B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109580618A (en) * | 2018-11-23 | 2019-04-05 | 鞍钢集团矿业有限公司 | A method of floating product grade is judged based on foam color |
| CN109772593A (en) * | 2019-01-25 | 2019-05-21 | 东北大学 | A kind of mineral pulp level prediction technique based on flotation froth behavioral characteristics |
| CN110046623A (en) * | 2019-03-04 | 2019-07-23 | 青岛小鸟看看科技有限公司 | A kind of characteristics of image point extracting method and camera |
| CN110109446A (en) * | 2019-05-28 | 2019-08-09 | 中南大学 | A kind of zinc floatation process Fuzzy Fault Diagnosis based on time series feature |
| CN110288591A (en) * | 2019-07-02 | 2019-09-27 | 中南大学 | Zinc flotation work condition judging method based on improved adaptive Multiple-population Genetic Algorithm |
| US20220414901A1 (en) * | 2021-03-31 | 2022-12-29 | Tianjin Research Institute For Water Transport Engineering, M.O.T | Real-time detection method of block motion based on feature point recognition |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102865859A (en) * | 2012-09-21 | 2013-01-09 | 西北工业大学 | Aviation sequence image position estimating method based on SURF (Speeded Up Robust Features) |
| CN103426186A (en) * | 2013-09-05 | 2013-12-04 | 山东大学 | Improved SURF fast matching method |
| CN104050687A (en) * | 2014-06-26 | 2014-09-17 | 中国矿业大学(北京) | Analyzing and processing method for flotation bubble motion pattern |
| WO2015083857A1 (en) * | 2013-12-05 | 2015-06-11 | 전자부품연구원 | Surf hardware apparatus, and method for managing integral image |
| CN105488816A (en) * | 2015-11-27 | 2016-04-13 | 中南大学 | On-line detection device and method of mineral flotation froth flow velocity on the basis of three-dimensional visual information |
| CN105976399A (en) * | 2016-04-29 | 2016-09-28 | 北京航空航天大学 | Moving object detection method based on SIFT (Scale Invariant Feature Transform) feature matching |
| CN106340008A (en) * | 2016-08-09 | 2017-01-18 | 杭州健培科技有限公司 | Feature value selection and SVM parameter optimization-based lung image recognition method |
| CN106951669A (en) * | 2017-05-19 | 2017-07-14 | 北京航空航天大学 | A kind of cognitive rolling bearing variable working condition method for diagnosing faults of view-based access control model |
| US20170248917A1 (en) * | 2014-09-11 | 2017-08-31 | Preciflex Sa | System for providing an indication using a meniscus mobilizer and elastic reservoirs, and methods, indicators and timepieces related thereto |
-
2018
- 2018-05-11 CN CN201810446667.2A patent/CN108830831B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102865859A (en) * | 2012-09-21 | 2013-01-09 | 西北工业大学 | Aviation sequence image position estimating method based on SURF (Speeded Up Robust Features) |
| CN103426186A (en) * | 2013-09-05 | 2013-12-04 | 山东大学 | Improved SURF fast matching method |
| WO2015083857A1 (en) * | 2013-12-05 | 2015-06-11 | 전자부품연구원 | Surf hardware apparatus, and method for managing integral image |
| CN104050687A (en) * | 2014-06-26 | 2014-09-17 | 中国矿业大学(北京) | Analyzing and processing method for flotation bubble motion pattern |
| US20170248917A1 (en) * | 2014-09-11 | 2017-08-31 | Preciflex Sa | System for providing an indication using a meniscus mobilizer and elastic reservoirs, and methods, indicators and timepieces related thereto |
| CN105488816A (en) * | 2015-11-27 | 2016-04-13 | 中南大学 | On-line detection device and method of mineral flotation froth flow velocity on the basis of three-dimensional visual information |
| CN105976399A (en) * | 2016-04-29 | 2016-09-28 | 北京航空航天大学 | Moving object detection method based on SIFT (Scale Invariant Feature Transform) feature matching |
| CN106340008A (en) * | 2016-08-09 | 2017-01-18 | 杭州健培科技有限公司 | Feature value selection and SVM parameter optimization-based lung image recognition method |
| CN106951669A (en) * | 2017-05-19 | 2017-07-14 | 北京航空航天大学 | A kind of cognitive rolling bearing variable working condition method for diagnosing faults of view-based access control model |
Non-Patent Citations (5)
| Title |
|---|
| JINPING LIU 等: "Flow velocity measurement and analysis based on froth image SIFT features and Kalman filter for froth", 《TURKISH JOURNAL OF ELECTRICAL ENGINEERING & COMPUTER SCIENCES》 * |
| YALIN WANG 等: "Sulfur Flotation Performance Recognition Based on Hierarchical Classification of Local Dynamic and Static Froth Features", 《IEEE ACCESS》 * |
| ZHOU HE-PENG 等: "Exploration on Mineral Flotation Bubble Velocity and Extraction Method of Size Distribution Feature", 《2ND INTERNATIONAL CONFERENCE ON MACHINERY, MATERIALS ENGINEERING, CHEMICAL ENGINEERING AND BIOTECHNOLOGY (MMECEB 2015)》 * |
| 冯洁: "基于浮选泡沫图像特征提取方法的研究及应用", 《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》 * |
| 唐朝晖 等: "基于数字图像处理的浮选泡沫速度特征提取及分析", 《中南大学学报(自然科学版)》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109580618A (en) * | 2018-11-23 | 2019-04-05 | 鞍钢集团矿业有限公司 | A method of floating product grade is judged based on foam color |
| CN109772593A (en) * | 2019-01-25 | 2019-05-21 | 东北大学 | A kind of mineral pulp level prediction technique based on flotation froth behavioral characteristics |
| CN110046623A (en) * | 2019-03-04 | 2019-07-23 | 青岛小鸟看看科技有限公司 | A kind of characteristics of image point extracting method and camera |
| CN110109446A (en) * | 2019-05-28 | 2019-08-09 | 中南大学 | A kind of zinc floatation process Fuzzy Fault Diagnosis based on time series feature |
| CN110109446B (en) * | 2019-05-28 | 2020-08-25 | 中南大学 | Zinc flotation process fuzzy fault diagnosis method based on time series characteristics |
| CN110288591A (en) * | 2019-07-02 | 2019-09-27 | 中南大学 | Zinc flotation work condition judging method based on improved adaptive Multiple-population Genetic Algorithm |
| US20220414901A1 (en) * | 2021-03-31 | 2022-12-29 | Tianjin Research Institute For Water Transport Engineering, M.O.T | Real-time detection method of block motion based on feature point recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108830831B (en) | 2021-11-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108830831A (en) | One kind is based on the improvement matched zinc flotation froth nature velocity characteristic extracting method of SURF | |
| CN107657279B (en) | Remote sensing target detection method based on small amount of samples | |
| CN107067415B (en) | A kind of object localization method based on images match | |
| CN107392232B (en) | Flotation working condition classification method and system | |
| CN107230203B (en) | Casting defect identification method based on human eye visual attention mechanism | |
| Liu et al. | Recognition methods for coal and coal gangue based on deep learning | |
| CN108898610A (en) | A kind of object contour extraction method based on mask-RCNN | |
| CN113870235B (en) | Circular stamping part defect detection method based on quantum firework arc edge extraction | |
| CN107423760A (en) | Based on pre-segmentation and the deep learning object detection method returned | |
| CN108280856A (en) | The unknown object that network model is inputted based on mixed information captures position and orientation estimation method | |
| CN109410248B (en) | Flotation froth motion characteristic extraction method based on r-K algorithm | |
| Weimer et al. | Learning defect classifiers for textured surfaces using neural networks and statistical feature representations | |
| Zhou et al. | Exploring faster RCNN for fabric defect detection | |
| CN107590427B (en) | Method for detecting abnormal events of surveillance video based on space-time interest point noise reduction | |
| Liu et al. | A classification method of glass defect based on multiresolution and information fusion | |
| CN108647722B (en) | Zinc ore grade soft measurement method based on process size characteristics | |
| CN110598030A (en) | Oracle bone rubbing classification method based on local CNN framework | |
| CN106874929A (en) | A kind of pearl sorting technique based on deep learning | |
| CN110826575A (en) | Underwater target identification method based on machine learning | |
| CN111260609A (en) | Cigarette appearance defect detection method based on deep learning | |
| CN108416801A (en) | A kind of Har-SURF-RAN characteristic point matching methods towards stereo vision three-dimensional rebuilding | |
| CN112734761A (en) | Industrial product image boundary contour extraction method | |
| CN114241372A (en) | Target identification method applied to sector-scan splicing | |
| CN116309847A (en) | Stacked workpiece pose estimation method based on combination of two-dimensional image and three-dimensional point cloud | |
| CN115984186A (en) | Fine product image anomaly detection method based on multi-resolution knowledge extraction |
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 |