CN108830831B - Zinc flotation foam natural speed feature extraction method based on improved SURF matching - Google Patents
Zinc flotation foam natural speed feature extraction method based on improved SURF matching Download PDFInfo
- Publication number
- CN108830831B CN108830831B CN201810446667.2A CN201810446667A CN108830831B CN 108830831 B CN108830831 B CN 108830831B CN 201810446667 A CN201810446667 A CN 201810446667A CN 108830831 B CN108830831 B CN 108830831B
- Authority
- CN
- China
- Prior art keywords
- matching
- foam
- speed
- image
- flotation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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 provides a zinc flotation foam natural speed feature extraction method based on improved SURF matching, which comprises the steps of selecting a SURF feature operator with high scale invariance, rotation invariance and anti-noise capacity to extract features of an image, analyzing similarity of foam shapes and specific motion tracks of the foam shapes, providing a real-time iteration updating matching method according to foam motion conditions, performing motion matching by using improved matching conditions to obtain zinc flotation foam initial speed, calculating a matching result and considering periodic motion of a scraper in a roughing tank under actual conditions, and selecting bottom flow velocity to obtain flotation foam natural speed features. The invention overcomes the problems of large matching error, low speed and the like of the traditional foam speed extraction method, improves the application speed by more than 10 times, and greatly improves the matching success rate by more than 20 percent by the proposed iterative updating SURF operator matching method. The problems that the flotation froth characteristics of the zinc flotation site are difficult to extract and how to realize machine visualization are solved.
Description
Technical Field
The invention belongs to the technical field of froth flotation, and particularly relates to a method for extracting the froth speed characteristic of roughing in the zinc flotation process.
Background
The froth flotation is one of the most important methods in the current mineral separation industry, is a method for separating minerals by utilizing different hydrophilicity caused by different physical and chemical properties of the surfaces of mineral particles, and has strong practical value. In the flotation process, a large number of bubbles with different sizes, colors, forms, textures and other characteristics can be formed by utilizing the physical movement of the stirring barrel and controlling the air inflow, and mineral particles are adhered to the surface of the bubbles to further realize mineral separation. The grade value and the recovery rate of the zinc concentrate react with the condition of the working condition. The flotation froth image characteristics are an important basis for judging the flotation condition and contain a large amount of information related to operation variables and production processes. For years, the selection plant mainly depends on expert experience to observe the foam state of the surface of the pulp of the flotation machine so as to adjust the pulp liquid level of the flotation machine and change the chemical system. The production condition is evaluated by experience, and objective cognition of the comprehensive condition of the whole process cannot be realized. Although the concentrate grade can be obtained by off-line test analysis in a factory, the working condition fluctuates along with the flow of the production line, and the method is greatly influenced by human factors, and has complex inspection process and high cost. And because the flotation process flow is complex and long, and many influencing factors exist, the on-line detection and real-time monitoring of the concentrate grade cannot be realized, the instant adjustment of the chemical adding amount and other parameters is influenced, and the recovery rate of minerals is finally influenced. Therefore, the research on how to extract the effective zinc flotation froth image characteristics aims at optimizing production indexes to carry out real-time online detection, and the method has important significance for guiding the optimization and stable operation of production operation and process.
With the rapid development of computer technology and digital image processing technology, the application of the soft measurement technology based on machine vision to the flotation process brings a new breakthrough to the real-time monitoring of flotation indexes, and obtains more flotation indexes related to the grade. In the actual production process, an experienced worker judges the current working condition and carries out adaptability adjustment by observing the characteristics of the foam image, such as speed, stability, color and the like. To realize the automatic stable production of the zinc flotation, firstly, effective information is acquired through machine vision, and the information is intellectualized and regulated, so that the real-time monitoring of the process is realized. Research finds that the flotation froth speed is an important way for judging the concentrate grade. The existing foam image speed feature extraction method has the problems of low robustness, large error of a matching result, low matching speed and the like, and a traditional feature operator does not show good adaptability aiming at the characteristics of special shape similarity, large deformation amount, easy fusion and collapse and the like of foam; in addition, in the actual process, the periodic physical rotation of the scraper also has certain influence on the natural flow velocity of the foam, and the extraction of the speed characteristic of the flotation foam meets a plurality of problems in the aspect of practical application.
Disclosure of Invention
The invention aims to provide a method for extracting natural speed characteristics of zinc flotation froth, which improves SURF matching, aiming at the problems that the concentrate grade is difficult to detect on line in the zinc flotation process and the flotation froth image characteristics are difficult to express accurately.
The invention is realized by the following scheme:
a zinc flotation froth natural velocity feature extraction method based on improved SURF matching comprises the following steps:
s1: zinc flotation image data under the condition of different concentrate grade values are obtained through a factory flotation image acquisition system, the acquired data are analyzed, and a vacant data set and a data set with unclear images caused by camera vibration or other artificial reasons are removed;
s2: forming three different sample subspaces according to the processed zinc flotation image data set and the production data, and specifically comprising the following steps:
(1) acquiring a data set of a carefully selected zinc flotation image, and dividing the data set into three working conditions of high grade, normal grade and low grade;
(2) classifying the collected images, and eliminating foam images causing interference by class boundaries;
(3) dividing the final foam image data set into V1, V2 and V3 which respectively correspond to three different working conditions;
s3: for each different sample subspace, extracting SURF characteristics of each frame of foam image, and the method comprises the following specific steps:
(1) extracting the k frame image, converting into gray image, and obtaining integral image I of the imagek;
wherein L isxx(x, sigma) is in the frame imageThe point takes a Gaussian function with the mean value of 0 and the standard deviation of sigma as a filter, and the second-order partial derivative in the x direction is calculated through convolution operation, and the same principle is adopted for Lxy(x,σ)、Lyy(x, σ), calculatingObtaining a matrix determinant det (H (x, sigma));
(3) the method comprises the steps of establishing a scale space pyramid by changing the size of a template of a filter, and using three or four layers of filters according to the size of a given image;
(4) comparing a det (H (x, sigma)) value calculated according to the Hessian matrix of each point with points in a 3 multiplied by 3 field, preliminarily determining characteristic points by using a 3D-non-maximum value inhibition method, obtaining sub-pixel-level characteristic points by using three-dimensional linear interpolation, setting a threshold value to reduce the number of the detected characteristic points, and selecting a maximum value as a final characteristic point;
(5) selecting the final characteristic point as a center, 6s as a radius range to construct a circle, wherein s is the scale of the final characteristic point when the final characteristic point is detected in the step (3), and taking the response value d of the Haar wavelet with the size of 4 sigma in the horizontal directionxAnd a response value d in the vertical directionyWeighting with a Gaussian function with the standard difference of sigma-2 s, and scanning a circle in a sector with a central angle of 60 degrees to obtain the direction of the maximum response value as the main direction of the final characteristic point;
(6) and (4) taking the final characteristic point as a center, taking the main direction in the step (5) as the direction and taking 20s as a window to construct a region block, dividing the region block into 4 multiplied by 4 sub-regions, and calculating the response value d of the Haar wavelet in the horizontal direction of each sub-regionxAnd a response value d in the vertical directionySum of absolute values of sum response values ∑ dx|、∑|dyI jointly form the structure v ═ Σ dx,∑dy,∑|dx|,∑|dy| to form a 4 × 4 × 4 ═ 64-dimensional SURF feature point description operator;
s4: aiming at the foam image with the SURF characteristics extracted, motion matching is carried out by utilizing improved matching conditions to obtain the initial velocity of the zinc flotation foam, and the method comprises the following specific steps:
(1) extracting two continuous frames of images of the sample subspace under the L working condition, and performing S3 operation to obtain a SURF feature description operator v0、 v1;
(2) Definition of v from Euclidean distance0And v1Distance between two operators to obtain the nearest distance d between two operators0And a second close distance d1Selecting the ratio of the twoAs a criterion for evaluation, a threshold ratio is setTWhen ratio < ratioTIf so, the feature point matching is successful, otherwise, the feature point matching fails;
(3) removing mismatching points by using a RANSAC algorithm according to the characteristic point matching results of the two frames of foam images;
(4) according to the matching result after the mismatching points are removed, the foam displacement is obtained according to the pixel point difference of the matching points of the two frames of images, and the unit time is removed to obtain the initial speedWherein c is the number of the feature points successfully matched;
(5) to obtainThe average value SPE _ vel of medium velocity component and the average value SPE _ ang of direction component are selected as the weight coefficient omega0,ω1Updating the matching standard in the step (2), if ratio < ratioTAnd isThe matching is successful, otherwise the matching fails, whereinIs the average value of the initial velocity magnitude components of the current matching points,the average value of the initial speed direction components of the current matching point is obtained;
(6) calculating the initial foam speed of two continuous frames after the step (1) to the step (5) until the extraction of the speed characteristics of the subspace foam image is finished, and obtaining the initial flotation foam speed under the working condition of L
S5: calculating and considering the periodic motion of a scraper in a roughing tank under the actual production condition according to the matching result, and selecting the bottom flow speed to obtain the natural speed characteristic of the flotation foam, wherein the method comprises the following specific steps:
(1) for each component in the flotation froth initial velocity vectorCarrying out normalization processing to obtain a velocity component histogram;
(2) selecting more than 80% of velocity component intervals in the velocity component histogram to obtain an average valueTo obtain
(3) According to midle _ speedLObtaining a flotation froth speed curve through secondary linear interpolation;
(4) calculating all minimum value points of the speed curve in the step (3), and taking the average value to obtain the final speed
s6: and extracting the speed characteristics of all the foam images in the three sample subspaces according to the steps to form a speed characteristic set under the sample condition.
The invention provides a zinc flotation froth natural speed characteristic extraction method based on improved SURF matching, which solves the problems that the flotation froth characteristics of a zinc flotation field are difficult to extract and how to realize machine visualization; according to the method, the SURF characteristic operator with high scale invariance, rotation invariance and anti-noise capability is selected for characteristic extraction of the foam images, and compared with the traditional sub-block registration and phase registration method, the matching speed between the foam images is improved; because flotation froth is projected and imaged from 3D to 2D, the froth similarity is high, the deformation rate is high, the collapse rate is high, the existing SURF matching algorithm is directly applied to the flotation froth scene, the mismatching rate is high, and effective matching cannot be implemented, so that a real-time iterative updating matching method according to the froth motion condition is provided by analyzing the froth shape similarity and the specific motion track, and the matching rate is greatly improved; considering the speed noise caused by the periodic power action of the scraper in the actual process, the bottom flow speed is selected as the natural foam speed according to the periodic speed change. The method overcomes the problems of large matching error, low speed and the like of the traditional foam speed extraction method, improves the application speed of the traditional SIFT operator in the scene by more than 10 times, greatly improves the matching success rate by more than 20 percent through the proposed iterative updating SURF operator matching method, and is convenient for real-time operation on site.
Drawings
FIG. 1 is a flow chart of natural velocity feature extraction of zinc flotation froth in the practice of the invention.
Detailed Description
The technical solutions adopted in the present invention are described and explained in more detail and clearly with reference to the accompanying drawings. The invention provides a zinc flotation froth natural speed characteristic extraction method for improving SURF matching by utilizing speed characteristics closely related to zinc flotation yield and zinc flotation concentrate grade, aiming at the problems that the concentrate grade is difficult to detect on line in the zinc flotation process and flotation froth image characteristics are difficult to express accurately. It is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the relevant art without any inventive step based on the embodiments of the present invention, shall be within the scope of the present invention.
As shown in fig. 1, a method for extracting natural velocity characteristics of zinc flotation froth for improving SURF matching in an embodiment of the present invention includes the following specific steps:
s1: zinc flotation image data under the condition of different concentrate grade values are obtained through a factory flotation image acquisition system, and the acquired data are analyzed according to expert experience to remove interference items, such as a vacant data set and a data set with unclear images caused by camera vibration or other human factors.
S2: forming three different sample subspaces according to the processed zinc flotation image data set and the production data, and specifically comprising the following steps:
(1) acquiring a carefully selected zinc flotation image data set, and dividing the data set into three working conditions of high grade (the grade of zinc concentrate is higher than 55%), normal (the grade of zinc concentrate is about 54%) and low grade (the grade of zinc concentrate is lower than 53%) according to production data indexes and expert experience;
(2) classifying the collected and selected images, and eliminating foam images causing interference by class boundaries;
(3) and dividing the final foam image data set into V1, V2 and V3 which respectively correspond to three different working conditions of high grade, normal grade and low grade.
S3: for each different sample subspace, extracting SURF characteristics of each frame of foam image, and the method comprises the following specific steps:
(1) extracting the k frame image, converting the k frame image into a gray image, and traversing to obtain an integral image I of the imagek;
wherein L isxx(x, sigma) is in the frame imageThe point takes a Gaussian function with the mean value of 0 and the standard deviation of sigma as a filter, and calculates the second-order partial derivative in the x direction through convolution operation, and the same principle is that Lxy(x,σ)、Lyy(x, σ), calculatingObtaining a matrix determinant:
det(H(x,σ))=DxxDyy-(wDxy)2
wherein Dxx、DyyTo approximate a gaussian approximation solution with a 9 × 9 box filter where σ is 1.2, the operation can be simplified, where w is calculated by the following equation:
wherein |. YFIs a Frobenius norm;
(3) constructing a Gaussian pyramid scale space, establishing a scale space pyramid by changing the template size of a filter, wherein an original image does not need to be downsampled, and using three layers or four layers of filters according to the size of a given image, wherein the template size of a first layer of filters is 9 × 9, 15 × 15, 21 × 21 and 27 × 27, the difference between adjacent templates is 6 pixels, the template size of a second layer of filters is 15 × 15, 27 × 27, 39 × 39 and 51 × 51, the difference between adjacent templates is 12 pixels, the template size of a third layer of filters is 27 × 27, 51 × 51, 75 × 75 and 99 × 99, the difference between adjacent templates is 24 pixels, and the template size of a fourth layer of filters is 51 × 51, 99 × 99, 147 × 147 and 195 × 195, and the difference between adjacent templates is 48 pixels;
(4) calculating a det (H (x, sigma)) value according to the Hessian matrix of each point, setting a reasonable threshold value, removing points smaller than the threshold value, comparing the Hessian matrix response value with points in a 3 multiplied by 3 field by using a 3D-non-maximum value inhibition method, primarily determining characteristic points, obtaining sub-pixel-level characteristic points by using three-dimensional linear interpolation fitting for comparison, and selecting the maximum value as a final characteristic point;
(5) selecting the final characteristic point as a center, 6s as a radius range to construct a circle, wherein s is the scale of the final characteristic point when the final characteristic point is detected in the step (3), and taking the response value d of the Haar wavelet with the size of 4 sigma in the horizontal directionxAnd a response value d in the vertical directionyWeighted by a Gaussian function with a standard deviation of 2s, and is performed by scanning a circle with a sector having a central angle of 60 DEG in the direction of the maximum response valueIs the final characteristic point principal direction;
(6) and (4) taking the final characteristic point as a center, taking the main direction in the step (5) as the direction and taking 20s as a window to construct a region block, dividing the region block into 4 multiplied by 4 sub-regions, and calculating the response value d of the Haar wavelet in the horizontal direction of each sub-regionxAnd a response value d in the vertical directionySum of absolute values of sum response values ∑ dx|、∑|dyI jointly form the structure v ═ Σ dx,∑dy,∑|dx|,∑|dy| to form a 4 × 4 × 4 ═ 64-dimensional SURF feature point description operator;
s4: aiming at the foam image with the SURF characteristics extracted, motion matching is carried out by utilizing improved matching conditions to obtain the initial velocity of the zinc flotation foam, and the method comprises the following specific steps:
(1) extracting two continuous frames of images of the sample subspace under the L working condition, and performing S3 operation to obtain a SURF feature description operatorWherein m and n are each independentlyAndthe number of middle feature points;
(2) calculation from Euclidean distanceAndthe similar characteristic vector between them, the nearest distance d between two characteristic values is obtained0=min(dis(v0i,v1j) Dis (v) therein0i,v1j) Is v is0iAnd v1jThe Euclidean distance between the two points is obtained to obtain the matching point of the nearest Euclidean distance, the two points are removed, and then the nearest distance d is obtained1Selecting the ratio of the twoAs a criterion for evaluation, a threshold ratio is setTWhen ratio < ratioTIf so, the feature point matching is successful, otherwise, the feature point matching fails;
(3) removing mismatching points by using RANSAC algorithm according to the feature point matching result of the two frames of foam images, and obtaining a point set which is successfully matched in the step (2)Random sampling T1A matching corresponding point, wherein T1∈[4,20]Calculating a linear equation and then calculating a setSetting the distance threshold d between the inner point and the outer point according to the distance between the residual point and the straight lineT(obtained by multiple experiments), all are less than dTThe distance points are determined as inner points, the rest are outer points, and the optimal number T of the inner points is set2Wherein For successfully matched point setsWhen the number of inner points is more than T2Then, the model is judged to be optimal, and the statistical error of the model is calculatedminThe process is repeated, the model is iteratively updated and the statistical error is calculated such that error is mademinMinimum, model optimal, thereby cullingA medium mismatch point;
(4) according to the matching result after the mismatching points are removed and the pixel point difference of the matching points of the two frames of images, obtaining the foam displacement between the matching points, and removing the unit time to obtain the initial speedDegree of rotationWherein c is the number of the feature points successfully matched;
(5) to obtainThe average value SPE _ vel of medium velocity component and the average value SPE _ ang of direction component are selected as the weight coefficient omega0,ω1Updating the matching criteria in (2) if ratio < ratioTAnd isThe matching is successful, otherwise the matching fails, whereinIs the average value of the initial velocity magnitude components of the current matching points,the average value of the initial speed direction components of the current matching point is obtained;
(6) calculating the initial foam speed of two continuous frames after the step (1) to the step (5) until the extraction of the speed characteristics of the subspace foam image is finished, and obtaining the initial flotation foam speed under the working condition of L
S5: and calculating the scraper power speed under the actual production condition according to the matching result to obtain the natural speed characteristic of the flotation foam, and the method comprises the following specific steps:
(1) for each component in the flotation froth initial velocity vectorCarrying out normalization processing to obtain a velocity component histogram;
(2) selecting more than 80% of velocity component intervals in the velocity component histogram to obtain an average valueTo obtain
(3) According to midle _ speedLObtaining a flotation froth speed curve through secondary linear interpolation;
(4) considering the periodic movement of a scraper in a roughing tank, selecting the bottom flow velocity as the natural speed of flotation foam, removing outliers, solving all minimum values of the speed curve in the step (3), and taking the average value to obtain the final speedWherein xiIs the ith minimum value point, and n is the number of the minimum value points.
S6: and extracting the speed characteristics of all the foam images in the three sample subspaces according to the steps to form a speed characteristic set under the sample condition.
The scope of the invention is not limited to the description of the embodiments.
Claims (2)
1. A zinc flotation froth natural velocity feature extraction method based on improved SURF matching is characterized by comprising the following steps:
s1: zinc flotation image data under the condition of different concentrate grade values are obtained through a factory flotation image acquisition system, the acquired data are analyzed, a vacant data set and a data set with unclear images caused by camera vibration or other artificial reasons are removed, and a processed zinc flotation image data set and production data are obtained;
s2: forming three different sample subspaces according to the processed zinc flotation image data set and the production data, and specifically comprising the following steps:
(1) acquiring a data set of a carefully selected zinc flotation image, and dividing the data set into three working conditions of high grade, normal grade and low grade;
(2) classifying the collected images, and eliminating foam images causing interference by class boundaries;
(3) dividing the final foam image data set into V1, V2 and V3 which respectively correspond to three different working conditions;
s3: for each different sample subspace, extracting SURF characteristics of each frame of foam image, and the method comprises the following specific steps:
(1) extracting the k frame image, converting into gray image, and obtaining integral image I of the imagek;
wherein L isxx(x, sigma) is in the frame imageThe point takes a Gaussian function with the mean value of 0 and the standard deviation of sigma as a filter, and the second-order partial derivative in the x direction is calculated through convolution operation, and the same principle is adopted for Lxy(x,σ)、Lyy(x, σ), obtaining a matrix determinant det (H (x, σ));
(3) the method comprises the steps of establishing a scale space pyramid by changing the size of a template of a filter, and using three or four layers of filters according to the size of a given image;
(4) comparing a det (H (x, sigma)) value calculated according to the Hessian matrix of each point with points in a 3 multiplied by 3 neighborhood, preliminarily determining characteristic points by using a 3D-non-maximum value inhibition method, obtaining sub-pixel-level characteristic points by using three-dimensional linear interpolation, setting a threshold value to reduce the number of the detected characteristic points, and selecting a maximum value as a final characteristic point;
(5) selecting the final characteristic point as a center, 6s as a radius range to construct a circle, wherein s is the dimension of the final characteristic point detected in the step (3) and is 4 sigma in sizeThe response value d of the Haar wavelet in the horizontal directionxAnd a response value d in the vertical directionyWeighting with a Gaussian function with standard deviation of 2s, and scanning a circle in a circle by using a fan with a central angle of 60 degrees to obtain the direction of the maximum response value as the main direction of the final characteristic point;
(6) and (4) taking the final characteristic point as a center, taking the main direction in the step (5) as the direction and taking 20s as a window to construct a region block, dividing the region block into 4 multiplied by 4 sub-regions, and calculating the response value d of the Haar wavelet in the horizontal direction of each sub-regionxAnd a response value d in the vertical directionySum of absolute values of sum response values ∑ dx|、∑|dyI jointly form the structure v ═ Σ dx,∑dy,∑|dx|,∑|dy| to form a 4 × 4 × 4 ═ 64-dimensional SURF feature point description operator;
s4: aiming at the foam image with the SURF characteristics extracted, motion matching is carried out by utilizing improved matching conditions to obtain the initial velocity of the zinc flotation foam, and the method comprises the following specific steps:
(1) extracting two continuous frames of images of the sample subspace under the L working condition, and performing S3 operation to obtain a SURF feature description operator v0、v1;
(2) Definition of v from Euclidean distance0And v1Distance between two operators to obtain the nearest distance d between two operators0And a second close distance d1Selecting the ratio of the twoAs a criterion for evaluation, a threshold ratio is setTWhen ratio < ratioTIf so, the feature point matching is successful, otherwise, the feature point matching fails;
(3) removing mismatching points by using a RANSAC algorithm according to the characteristic point matching results of the two frames of foam images;
(4) according to the matching result after the mismatching points are removed, the foam displacement is obtained according to the pixel point difference of the matching points of the two frames of images, and the initial speed is obtained by removing the unit timeWherein c is the feature of successful matchingThe number of the sign points;
(5) to obtainThe average value SPE _ vel of medium-speed component and the average value SPE _ ang of direction component are selected as the weight coefficient omega0,ω1Updating the matching standard in the step (2), if ratio < ratioTAnd isThe matching is successful, otherwise the matching fails, whereinIs the average value of the initial velocity magnitude components of the current matching points,the average value of the initial speed direction components of the current matching point is obtained;
(6) calculating the initial foam speed of two continuous frames after the step (1) to the step (5) until the extraction of the speed characteristics of the subspace foam image is finished, and obtaining the initial flotation foam speed under the working condition of L
S5: calculating and considering the periodic movement of a scraper in a roughing tank under the actual production condition according to a matching result, and selecting the bottom flow velocity to obtain the natural speed characteristic of the flotation foam, wherein the method comprises the following specific steps:
(1) for each component in the flotation froth initial velocity vectorCarrying out normalization processing to obtain a velocity component histogram;
(2) selecting more than 80% of velocity component intervals in the velocity component histogram to obtain an average valueTo obtain
(3) According to midle _ speedLObtaining a flotation froth speed curve through secondary linear interpolation;
(4) calculating all minimum value points of the speed curve in the step (3), and taking the average value to obtain the final speedWherein xiIs the ith minimum value point, and n is the number of the minimum value points;
s6: and extracting the speed characteristics of all the foam images in the three sample subspaces to form a speed characteristic set under the sample condition.
2. The method for extracting the natural speed characteristics of the zinc flotation froth based on the improved SURF matching as claimed in claim 1, wherein: and (3) after the SURF features of the foam image are extracted, performing parameter ratio in step (2) of the improved matching algorithm by the S4TThe parameter omega in the step (5)0、ω1The value ranges are respectively as follows: ratio (R)T∈[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 CN108830831A (en) | 2018-11-16 |
CN108830831B true 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) |
Families Citing this family (5)
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 |
CN109772593B (en) * | 2019-01-25 | 2020-09-29 | 东北大学 | Ore pulp liquid level prediction method based on flotation foam dynamic characteristics |
CN110046623B (en) * | 2019-03-04 | 2021-09-10 | 青岛小鸟看看科技有限公司 | Image feature point extraction method and camera |
CN110109446B (en) * | 2019-05-28 | 2020-08-25 | 中南大学 | Zinc flotation process fuzzy fault diagnosis method based on time series characteristics |
CN110288591B (en) * | 2019-07-02 | 2021-03-09 | 中南大学 | Zinc flotation working condition discrimination method based on improved self-adaptive multi-population genetic algorithm |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102865859B (en) * | 2012-09-21 | 2014-11-05 | 西北工业大学 | Aviation sequence image position estimating method based on SURF (Speeded Up Robust Features) |
CN103426186B (en) * | 2013-09-05 | 2016-03-02 | 山东大学 | A kind of SURF fast matching method of improvement |
KR101531038B1 (en) * | 2013-12-05 | 2015-06-23 | 전자부품연구원 | Surf hardware apparatus and method for managing integral image memory thereof |
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 |
WO2016038446A1 (en) * | 2014-09-11 | 2016-03-17 | Preciflex Sa | System for providing an indication using a meniscus mobilizer and elastic reservoirs, and methods, indicators and timepieces related thereto |
CN105488816B (en) * | 2015-11-27 | 2018-09-11 | 中南大学 | A kind of mineral floating foam flow velocity on-line measuring device and method based on 3D vision 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 |
CN106951669B (en) * | 2017-05-19 | 2019-01-22 | 北京航空航天大学 | A kind of rolling bearing variable working condition method for diagnosing faults of view-based access control model cognition |
-
2018
- 2018-05-11 CN CN201810446667.2A patent/CN108830831B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN108830831A (en) | 2018-11-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108830831B (en) | Zinc flotation foam natural speed feature extraction method based on improved SURF matching | |
CN107657279B (en) | Remote sensing target detection method based on small amount of samples | |
CN103886589B (en) | Object-oriented automated high-precision edge extracting method | |
CN106204572B (en) | Road target depth estimation method based on scene depth mapping | |
CN113870235A (en) | Method for detecting defects of circular stamping part based on quantum firework arc edge extraction | |
CN110837768B (en) | Online detection and identification method for rare animal protection | |
CN111982910B (en) | Weak supervision machine vision detection method and system based on artificial defect simulation | |
CN109410248B (en) | Flotation froth motion characteristic extraction method based on r-K algorithm | |
CN104463199A (en) | Rock fragment size classification method based on multiple features and segmentation recorrection | |
CN112819772A (en) | High-precision rapid pattern detection and identification method | |
CN115797354B (en) | Method for detecting appearance defects of laser welding seam | |
CN109781737B (en) | Detection method and detection system for surface defects of hose | |
CN111860501B (en) | High-speed rail height adjusting rod falling-out fault image identification method based on shape matching | |
CN111402238A (en) | Defect identification system realized through machine vision | |
CN108647722B (en) | Zinc ore grade soft measurement method based on process size characteristics | |
CN110222661B (en) | Feature extraction method for moving target identification and tracking | |
CN116309847A (en) | Stacked workpiece pose estimation method based on combination of two-dimensional image and three-dimensional point cloud | |
CN115100116A (en) | Plate defect detection method based on three-dimensional point cloud | |
CN116740044B (en) | Copper pipe milling surface processing method and system based on visual detection and control | |
CN108399614B (en) | Fabric defect detection method based on non-sampling wavelet and Gumbel distribution | |
CN114820471A (en) | Visual inspection method for surface defects of intelligent manufacturing microscopic structure | |
CN102592290A (en) | Method for detecting moving target region aiming at underwater microscopic video | |
CN115222884A (en) | Space object analysis and modeling optimization method based on artificial intelligence | |
CN113674218B (en) | Weld feature point extraction method and device, electronic equipment and storage medium | |
Liu et al. | Flow velocity measurement and analysis based on froth image SIFT features and Kalman filter for froth flotation |
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 |