CN111415348A - Method for extracting characteristics of bubbles in automobile brake pipeline - Google Patents
Method for extracting characteristics of bubbles in automobile brake pipeline Download PDFInfo
- Publication number
- CN111415348A CN111415348A CN202010218489.5A CN202010218489A CN111415348A CN 111415348 A CN111415348 A CN 111415348A CN 202010218489 A CN202010218489 A CN 202010218489A CN 111415348 A CN111415348 A CN 111415348A
- Authority
- CN
- China
- Prior art keywords
- bubble
- bubbles
- matching
- automobile brake
- parameter
- 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
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
- G06T7/248—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20152—Watershed segmentation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Geometry (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for extracting characteristics of bubbles in an automobile brake pipe, which aims to track the positions of the bubbles in the automobile brake pipe and extract the characteristics of the bubbles. The method only needs small power consumption, performs bubble segmentation by using a bubble segmentation Algorithm combining a convex defect detection Algorithm and a Watershed Algorithm (Watershed Algorithm), completes the matching of front and rear frame bubbles by using a Hungary Algorithm (Hungarian Algorithm), completes the bubble classification by using a bubble parameter threshold value in an automobile brake pipeline, and finally completes the volume calculation of the bubbles by using a numerical integration method. The invention meets the real-time requirement of detecting the bubbles in the automobile brake pipeline, collects the important characteristics of the bubbles and provides help for analyzing the health state of the brake fluid in the automobile brake pipeline.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a method for extracting characteristics of bubbles in an automobile brake pipeline.
Background
The hydraulic brake line transmits braking force through brake fluid, which often contains air bubbles. The brake fluid contains bubbles which have important influence on the braking performance and the driving safety of the automobile. The detection of the bubbles in the automobile brake pipeline can provide support for the design of an automobile brake system and can also know the health state of brake fluid. Bubble identification relates to a plurality of fields such as flow pattern identification and target detection.
For the flow pattern identification, the most popular method at present is to apply an intelligent algorithm to perform bubble flow pattern classification. For example: and (3) performing bubble classification by using an SVM (support Vector machine) algorithm, a clustering algorithm and various neural networks. However, these methods tend to have relatively high power consumption, some methods have relatively high requirements on training sets, and the methods are black box means and are inconvenient to understand and apply.
In addition, dynamic detection algorithms such as Frame Difference Method (Frame Difference Method), background shearing Method and recently developed Vibe algorithm are often applied to moving targets, but the intelligent algorithm occupies more computing resources, and the dynamic detection requires the movement of the targets and generates pixel Difference when the movement needs to generate pixel Difference.
In addition, for the segmentation of bubbles involved in the processing of bubble images, the segmentation methods mentioned in various documents are as follows: 1. hough Transform Algorithm (Hough Transform Algorithm); 2. watershed algorithms (Watershed algorithms); 3. breakpoint detection Algorithm (Breakpoint Algorithm); 4. an Edge Intensity Gradient Algorithm (Edge Intensity Gradient Algorithm) or the like segments and reconstructs the bubbles. The Hough transform algorithm is mainly applied to circular or elliptical bubbles with regular shapes. The breakpoint detection algorithm and the edge pixel gradient algorithm are complex and are troublesome to execute. The internal texture of the large bubble image related in the invention is complex, and the bubble segmentation effect of a simple watershed algorithm is poor.
In summary, no detection method suitable for detecting air bubbles in an automobile brake pipe exists in the prior art.
Disclosure of Invention
The invention aims to provide a method for extracting characteristics of bubbles in an automobile brake pipe, which can solve one or more of the technical problems.
In order to achieve the purpose, the technical scheme provided by the invention is as follows:
a method for extracting the characteristics of bubbles in an automobile brake pipeline; the method comprises the following steps:
s1, establishing a bubble training set in the automobile brake pipeline;
s2, establishing a bubble test set in the automobile brake pipeline; the method comprises the steps of firstly adopting a convex defect detection algorithm to distinguish single bubbles and bubble groups, then using a watershed algorithm of a user-defined substrate to segment the bubble groups, separating the single bubbles in the bubble groups, and obtaining a single bubble test set in an automobile brake pipeline; so as to realize the complete detection of the air bubbles in the automobile pipeline;
s3, setting a bubble flow pattern parameter threshold value through the bubble training set in S1, and performing corresponding classification on the bubble flow pattern of each bubble in the bubble test set in the step S2 according to the bubble flow pattern parameter threshold value;
s4, performing maximum matching and optimal matching on the air bubbles of the front and rear frames in the air bubble test set obtained in the step S2 by using Hungarian algorithm (Hungarian Algorithm); the method comprises the steps of tracking the position of a bubble in each frame of a two-phase flow flowing video in an automobile brake pipeline, and calculating the speed of each bubble in a bubble test set and each characteristic corresponding to the bubble;
s5, calculating the volume of each bubble in the bubble test set in S2 by using a numerical integration method;
s6 extracts characteristic information of the bubbles obtained in S3, S4, S5.
According to the method, a training set of bubbles in the automobile brake pipeline is established, the parameter threshold of the bubbles in the automobile brake pipeline is obtained, and classification of the bubble flow pattern is completed. Compared with the traditional SVM and clustering algorithm, the method has the advantages of direct classification mode, visualization method, high classification speed and low requirement on the number of samples in a training set.
In addition, in the invention, the bubble segmentation is a watershed algorithm combining convex defects and a customized substrate, and the method is simple and effective. The segmentation may be done so that the calculation of the bubble volume is more accurate. The matching algorithm of the invention is simple and not cumbersome, tracks the position of the bubble in each frame of the video screen, and calculates the speed of the bubble. In addition, the bubble volume is calculated by adopting a numerical integration method, so that the accuracy of bubble volume estimation is improved.
Preferably, the specific steps in step S1 are as follows:
s11, shooting a two-phase flow flowing video in the automobile brake pipeline by using a high-speed camera;
s12, guiding the two-phase flow video in the automobile brake pipe into Opencv, and selecting sample bubbles in a bubble training set in the automobile brake pipe by using ROI (region of interest) so as to obtain a bubble training set with a time sequence multiframe diagram. Preferably, the specific steps in step S2 are as follows:
s21, introducing an image of each frame of bubbles in a two-phase flow flowing video in an automobile brake pipeline into Opencv, and binarizing the bubbles by using an Otsu' S algorithm to obtain a binarized image of each frame of bubbles;
s22, detecting the bubble binary image to obtain a bubble outline;
s23, distinguishing isolated bubbles and bubble groups concentrated in the bubble test by using a convex defect detection algorithm;
s24 is performed to segment the bubbles in the bubble group, specifically as follows:
s241, performing bubble segmentation on the bubble group with only two defects by adopting a method of connecting two defect points;
s242, for the bubble groups with three or more defects, acquiring inner layer contours of the bubble groups, and carrying out initial marking on a Watershed Algorithm (Watershed Algorithm);
s243, calculating the shortest distance from each pixel point in the bubble outline to the outline by using a pointPolygontest function in Opencv, and obtaining a self-creation base map;
s244 performs bubble segmentation on the self-created base map in step S24 using Watershed Algorithm (Watershed Algorithm), to obtain individual bubbles.
Preferably, the specific steps in step S3 are as follows:
s31, calculating classification parameters of sample bubbles in a bubble training set in the automobile brake pipe, wherein the classification parameters comprise:
area (S): obtaining the outline area of the sample bubble through a built-in function in Opencv;
aspect ratio (Ar): the ratio of the longitudinal length of the bubble to the transverse length of the bubble can be obtained by the ratio of the width and the length of a rectangle circumscribed by the bubble;
roundness (C): representing the degree of circularity of the bubble, expressed by the formulaCalculating, wherein S is the outline area of the bubble, and L is the circumference of the bubble;
left and right contour area ratio (L R _ ratio) of the area of the left and right halves of the bubble divided by the bounding rectangle of the bubble from the middle;
left and right upper half-outline area ratio (L R _ helf _ ratio) of the ratio of the left and right bisected areas by quartering the bubble circumscribing rectangle and by bisecting the upper surface of the circumscribing rectangle;
s32, drawing a parameter map of the bubble to obtain a bubble parameter threshold;
s33, collecting classification parameters for each bubble in the bubble test set in S2, comparing the classification parameters with parameter thresholds, and classifying the bubble flow pattern.
Preferably, the specific steps in step S4 are as follows:
s41, establishing a bubble library and a pre-storage library: firstly, defining a pre-storage library for storing new bubbles which appear at the first time, and a bubble library for storing bubbles which appear for many times and have a speed attribute;
the method comprises the following specific steps: performing maximum matching (all parameters are matched, and the most matched parameters are preferred) on the actual parameters of the bubbles in each frame in the bubble test set and the prediction parameters of the bubbles in the bubble library by using a Hungarian algorithm, and performing optimal matching on the successfully matched bubbles to minimize the matched power consumption;
the matching parameters comprise a bubble abscissa x, a bubble ordinate y and a bubble area S;
the deviation delta x ═ x of the bubble abscissa x of the matching parameter is determinedexpect-xactual|,xexpectExpected value, x, representing the abscissa x of the bubble of the matching parameteractualThe true value representing the x-coordinate of the bubble abscissa of the matching parameter;
deviation Δ y ═ y of the bubble ordinate y of the matching parameterexpect-yactual|,yexpectExpected value, y, representing the bubble ordinate y of the matching parameteractualRepresenting the true value of the bubble ordinate y of the matching parameter;
deviation of the matching parameter bubble area SSexpectExpected value, S, of bubble area S representing matching parameteractualRepresenting the true value of the matching parameter bubble area S;
when the bubble has the prediction parameter, the expected value is the value of the prediction parameter, and when the bubble does not have the prediction parameter, the expected value is the bubble parameter of the previous frame;
the prediction parameters are bubble parameters existing in a bubble library; the actual parameters are the parameters that are actually measured (i.e., real-time parameters).
When the maximum matching is carried out, deviation threshold values of delta x, delta y and delta S are respectively set, and three deviations are required to be within the threshold values; when the optimal matching is carried out, the three deviations are synthesized in proportion and used as the power consumption for matching between the two bubbles;
s42, performing maximum matching on the actual parameters of the unsuccessfully matched bubbles and the parameters of the last frame of bubbles in the pre-stored library, if the matching is successful, calculating the bubble speed and storing the bubbles in the bubble library; defining the bubble which fails to be matched as a new bubble; after the matching is finished, storing the new bubbles into a pre-storage library for matching of the next cycle;
s43, taking the bubble with the successful matching times (life cycle, which means the successful matching times of the bubble on the pictures of three continuous frames) more than or equal to 3 in the bubble library as a stable matching bubble, tracking and outputting the position of the bubble, and obtaining the real-time characteristics of aspect ratio, flow pattern, volume and speed; and removing the bubbles which fail to be matched for more than three times in the bubble library as disappearance treatment.
Preferably, the specific steps in step S5 are as follows:
s51, firstly, fitting a circumscribed rectangle of the outline of the bubble (the bubble in the bubble test set in S2), continuously drawing vertical lines parallel to the leftmost end of the rectangle at equal intervals from the leftmost end of the rectangle until the rightmost end of the rectangle is drawn, and searching for the intersection point of the vertical lines and the outline;
s52, when the vertical line is at the leftmost end and the rightmost end, the vertical line and the outline respectively have only one intersection point, and the height of the intersection point is 0; the rest vertical lines and the contour have two intersection points, and half of the distance between the two intersection points is used as a function value y for calculating the volume of the bubble;
s53 bubble volume integral formula isAnd (3) calculating the volume of the bubbles by adopting a composite trapezoidal formula:in the formula x0Is the abscissa, x, of the leftmost end of the bubblenIs the most bubbleAbscissa of right end, yiDenotes the abscissa as xiThe function value at the position of (a).
The invention has the technical effects that:
1. according to the invention, a training set is established for the bubbles in the automobile brake pipe, the parameter threshold of the bubbles in the automobile brake pipe is obtained, the classification of the bubbles can be rapidly completed, and the requirement on the number of samples of the training set is reduced on a classification task; and meanwhile, the visualization of brake fluid bubble identification is realized.
2. The invention adopts the convex defect detection algorithm and the watershed algorithm of the self-defined substrate to finish the segmentation of the bubble group, so that the bubbles in the bubble group can be independently detected, and the identification and measurement precision is improved.
3. The invention uses numerical integration method to calculate the approximate volume of the rotating body of each type of bubble, thereby improving the accuracy of bubble volume estimation.
4. The method combines two Hungarian algorithms of maximum matching and optimal matching and various selective rejection processes to complete the position tracking and speed calculation of the bubbles.
The feature extraction method provided by the invention is effective, the calculated amount is not large, and the operation flow is simple.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention.
In the drawings:
FIG. 1 is an overall flow chart of the present invention;
FIG. 2 is a flow chart of bubble segmentation;
FIG. 3 is a flow chart of frame matching before and after a bubble;
FIG. 4 is a block flow diagram of bubble classification;
FIG. 5 is a base map of a watershed algorithm;
FIG. 6 is a diagram illustrating the effect of bubble segmentation;
FIG. 7 is a diagram illustrating the effect of frame matching before and after a bubble;
FIG. 8 is the result of ROI interaction;
FIG. 9 is a schematic illustration of bubble parameters;
FIGS. 10 to 14 are distribution diagrams of parameters of bubble classification;
FIG. 15 is a schematic of the integral solution to the volume of the rotating body.
Detailed Description
The present invention will now be described in detail with reference to the drawings and specific embodiments, wherein the exemplary embodiments and descriptions are provided only for the purpose of illustrating the present invention and are not to be construed as unduly limiting the invention.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
The working process of the invention is shown in figure 1, and the method for extracting the characteristics of the bubbles in the automobile brake pipe establishes a training set of the bubbles in the automobile brake pipe to obtain the parameter thresholds of the bubbles in the automobile brake pipe and prepare for bubble classification.
And the bubbles in the test set are the bubbles needing to acquire characteristic data, and the bubbles in the two-phase flow video test set in the automobile brake pipeline are respectively subjected to bubble segmentation, front-frame and rear-frame bubble matching, bubble classification, volume calculation and other operations. From this, a large amount of useful bubble feature (e.g., area, aspect ratio, roundness, left-right contour area ratio, left-right upper-half contour area ratio, bubble type, etc.) information can be extracted.
Fig. 2 is a flow chart of bubble segmentation. Firstly, carrying out binarization on bubbles by using an Otsu's algorithm, and then respectively extracting bubble groups with two defects and three or more defects in the image and pasting the bubble groups into different image matrixes for waiting processing. And then, for the bubble groups with two defects, performing bubble segmentation by adopting a method of connecting two defect points. The bubble population of three or more defects requires further finding the inner profile and using the pointPolygonTest function to obtain the substrate, as shown in fig. 5. And then, carrying out initial marking on the base map by using the internal contour, and then carrying out watershed segmentation on the base map. The effect of the final segmentation can be seen in fig. 6.
Fig. 3 is a specific flow of the bubble matching of the previous and subsequent frames of the concentrated bubble. A pre-store library is first defined for storing new bubbles that occur the first time, while a bubble library is used for storing bubbles that occur multiple times with a velocity attribute. When a new frame of bubbles enters a program, the maximum matching is carried out on the real-time parameters (actual parameters) of the new frame of bubbles and the prediction parameters of the bubbles in the bubble library, and then the bubbles which are successfully matched are optimally matched, so that the matched power consumption is minimum.
And after matching is completed, updating the bubble information of the bubble database. And the remaining unmatched bubbles are maximally matched with the parameters of the bubble of the last frame in the prestored library by the real-time parameters of the remaining unmatched bubbles. Once the match is successful, the bubble of the pre-stored library can be updated into the bubble library. When the life cycle (successful matching times) of the bubbles in the bubble library is more than or equal to 3, the bubbles are stably matched, and when the disappearance cycle (matching failure times) of the bubbles in the bubble library exceeds 3, the bubbles are judged to disappear, and the bubbles are removed. The matching pursuit effect is shown in fig. 7, which shows the progress of the deformation of the tracked bubble number 149. And table 1 illustrates the change of each characteristic amount of the bubble during the deformation process.
The scenario of fig. 8 is a process of training set bubble acquisition using ROI, and the parameter distribution of each type of bubble acquired is shown in fig. 10 to fig. 14. The parameter thresholds of the air bubbles in the automobile brake pipe are determined according to the graphs and are shown in the table 2. The bubble of the test set is classified by using the parameter thresholds of the bubble of table 2. The flow is shown in fig. 4.
It is necessary to explain the parameters in conjunction with fig. 9:
the ellipse in fig. 9 is used to indicate the detected bubble profile, and the horizontal and vertical dividing lines are the median lines in the width and length of the circumscribed rectangle, respectively.
Area (S): the profile area can be directly derived from the built-in function in the longitudinal length ratio in opencv. I.e. the area of the oval portion in fig. 9.
The aspect ratio (Ar), which is the longitudinal length to the lateral length of the bubble, can be derived from the ratio of the width to the length of the circumscribed rectangle of the bubble, shown as W/L in FIG. 9.
Roundness (C): representsDegree of circularity of the bubble, in particular by formulaIt can be calculated that S is the elliptical area L is the elliptical perimeter in fig. 9.
Left and right contour area ratio (L R _ ratio) the ratio of the area of the left and right bubble halves dividing the bubble bounding rectangle from the middle is critical to discriminating bullet streams and this parameter is shown in FIG. 9 as min (R + Y, G + B)/max (R + Y, G + B)
The ratio of the left and right upper half-profile areas (L R _ helf _ ratio) that bisects the bubble into four, the ratio of the left and right bisected areas above the circumscribed rectangle, which is helpful in screening for transitional bubbles, this parameter is represented in FIG. 9 as min (R, G)/max (R, G);
as for FIG. 15, which is a diagram of the integral for solving the bubble volume, the function value y is a length of ac/2. The integral of the volume of the bubble is expressed asCalculating the volume of the bubble by using a composite trapezoidal formula
TABLE 1 feature acquisition for tracking bubbles
TABLE 2 parameter thresholds for various flow patterns of the bubbles
Type of bubble | Threshold value |
Annular flow | Ar<0.1and S>40000 |
Bubbly flow | (Ar>=0.9and S<1500and C>0.82)or area<800or C>0.875 |
Spring-like flow | (LR_ratio<0.93and C<=0.81)or C<0.66 |
Plug flow | LR_helf_ratio>0.8 |
Transition state | Else |
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. A method for extracting the characteristics of bubbles in an automobile brake pipeline; the method is characterized in that: the method comprises the following steps:
s1, establishing a bubble training set in the automobile brake pipeline;
s2, establishing a bubble test set in the automobile brake pipeline; the method comprises the steps of firstly adopting a convex defect detection algorithm to distinguish single bubbles and bubble groups, then using a watershed algorithm of a user-defined substrate to segment the bubble groups, separating the single bubbles in the bubble groups, and obtaining a single bubble test set in an automobile brake pipeline; so as to realize the complete detection of the air bubbles in the automobile pipeline;
s3, setting a bubble flow pattern parameter threshold value through the bubble training set in S1, and performing bubble flow pattern classification on each bubble in the bubble test set in the step S2 according to the bubble flow pattern parameter threshold value;
s4, performing maximum matching and optimal matching on the bubbles of the front and rear frames in the bubble test set obtained in the step S2 by using the Hungarian algorithm; the method comprises the steps of tracking the position of a bubble in each frame of a two-phase flow flowing video in an automobile brake pipeline, and calculating the speed of each bubble in a bubble test set and each characteristic corresponding to the bubble;
s5, calculating the volume of each bubble in the bubble test set in S2 by using a numerical integration method;
s6 extracts characteristic information of the bubbles obtained in S3, S4, S5.
2. The method for extracting the characteristics of the bubbles in the automobile brake pipe according to claim 1; it is characterized in that the specific steps in the step S1 are as follows:
s11, shooting a two-phase flow flowing video in the automobile brake pipeline by using a high-speed camera;
s12, guiding the two-phase flow video in the automobile brake pipe into Opencv, and selecting sample bubbles in the bubble training set in the automobile brake pipe by using the ROI to further obtain the bubble training set with a time-sequence multi-frame diagram.
3. The method for extracting the characteristics of the bubbles in the automobile brake pipe according to claim 1; it is characterized in that the specific steps in the step S2 are as follows:
s21, introducing an image of each frame of bubbles in a two-phase flow flowing video in an automobile brake pipeline into Opencv, and binarizing the bubbles by using an Otsu' S algorithm to obtain a binarized image of each frame of bubbles;
s22, detecting the bubble binary image to obtain a bubble outline;
s23, distinguishing isolated bubbles and bubble groups concentrated in the bubble test by using a convex defect detection algorithm;
s24 is performed to segment the bubbles in the bubble group, specifically as follows:
s241, performing bubble segmentation on the bubble group with only two defects by adopting a method of connecting two defect points;
s242, for the bubble groups with three or more defects, acquiring inner layer contours of the bubble groups, and carrying out initial marking on a watershed algorithm;
s243, calculating the shortest distance from each pixel point in the bubble outline to the outline by using a pointPolygontest function in Opencv, and obtaining a self-creation base map;
s344, performing bubble segmentation on the self-created base map in the step S24 by using a watershed algorithm to obtain single bubbles.
4. The method for extracting the characteristics of the bubbles in the automobile brake pipe according to claim 1; it is characterized in that the specific steps in the step S3 are as follows:
s31, calculating classification parameters of sample bubbles in a bubble training set in the automobile brake pipe, wherein the classification parameters comprise:
area (S): obtaining the outline area of the sample bubble through a built-in function in Opencv;
aspect ratio (Ar): the ratio of the longitudinal length of the bubble to the transverse length of the bubble can be obtained by the ratio of the width and the length of a rectangle circumscribed by the bubble;
roundness (C): representing the degree of circularity of the bubble, expressed by the formulaCalculating, wherein S is the outline area of the bubble, and L is the circumference of the bubble;
left and right profile area ratio: dividing a bubble circumscribed rectangle from the middle, and comparing the areas of the left half bubble and the right half bubble;
left-right upper half contour area ratio: dividing the bubble circumscribed rectangle into four equal parts, and dividing the left and right equal areas on the circumscribed rectangle into equal parts;
s32, drawing a parameter map of the bubble to obtain a bubble parameter threshold;
s33, collecting classification parameters for each bubble in the bubble test set in S2, comparing the classification parameters with parameter thresholds, and classifying the bubble flow pattern.
5. The method for extracting the characteristics of the bubbles in the automobile brake pipe according to claim 1; it is characterized in that the specific steps in the step S4 are as follows:
s41, establishing a bubble library and a pre-storage library: firstly, defining a pre-storage library for storing new bubbles which appear at the first time, and a bubble library for storing bubbles which appear for many times and have a speed attribute; the method comprises the following specific steps: performing maximum matching on the actual parameters of the bubbles in each frame in the bubble test set and the prediction parameters of the bubbles in the bubble library by using a Hungarian algorithm, and performing optimal matching on the successfully matched bubbles to minimize the matched power consumption;
the matching parameters comprise a bubble abscissa x, a bubble ordinate y and a bubble area S; the deviation delta x ═ x of the bubble abscissa x of the matching parameter is determinedexpect-xactual|,xexpectExpected value, x, representing the abscissa x of the bubble of the matching parameteractualThe true value representing the x-coordinate of the bubble abscissa of the matching parameter; deviation Δ y ═ y of the bubble ordinate y of the matching parameterexpect-yactual|,yexpectExpected value, y, representing the bubble ordinate y of the matching parameteractualRepresenting the true value of the bubble ordinate y of the matching parameter; deviation of the matching parameter bubble area SSexpectExpected value, S, of bubble area S representing matching parameteractualRepresenting the true value of the matching parameter bubble area S;
when the bubble has the prediction parameter, the expected value is the value of the prediction parameter, and when the bubble does not have the prediction parameter, the expected value is the bubble parameter of the previous frame;
the prediction parameters are bubble parameters existing in a bubble library; the actual parameters are actually measured parameters;
when the maximum matching is carried out, deviation threshold values of delta x, delta y and delta S are respectively set, and three deviations are required to be within the threshold values; when the optimal matching is carried out, the three deviations are synthesized in proportion and used as the power consumption for matching between the two bubbles;
s42, performing maximum matching on the actual parameters of the unsuccessfully matched bubbles and the parameters of the last frame of bubbles in the pre-stored library, if the matching is successful, calculating the bubble speed and storing the bubbles in the bubble library; defining the bubble which fails to be matched as a new bubble; after the matching is finished, storing the new bubbles into a pre-storage library for matching of the next cycle;
s43, taking the bubbles with the successful matching times (life cycle) of the bubbles in the bubble library more than or equal to 3 as stable matching bubbles, tracking and outputting the positions of the stable matching bubbles, and obtaining real-time characteristics of aspect ratio, flow pattern, volume and speed; and removing the bubbles which fail to be matched for more than three times in the bubble library as disappearance treatment.
6. The method for extracting the characteristics of the bubbles in the automobile brake pipe according to claim 1; it is characterized in that the specific steps in the step S5 are as follows:
s51, firstly, fitting a circumscribed rectangle of the bubble outline, continuously drawing vertical lines parallel to the leftmost end of the rectangle at equal intervals from the leftmost end of the rectangle until the rightmost end of the rectangle is drawn, and searching for an intersection point of the vertical lines and the outline;
s52, when the vertical line is at the leftmost end and the rightmost end, the vertical line and the outline respectively have only one intersection point, and the height of the intersection point is 0; the rest vertical lines and the contour have two intersection points, and half of the distance between the two intersection points is used as a function value y for calculating the volume of the bubble;
s53 bubble volume integral formula isAnd (3) calculating the volume of the bubbles by adopting a composite trapezoidal formula:
in the above formula, x0Is the abscissa, x, of the leftmost end of the bubblenIs the abscissa, y, of the rightmost end of the bubbleiDenotes the abscissa as xiThe function value at the position of (a).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010218489.5A CN111415348B (en) | 2020-03-25 | 2020-03-25 | Method for extracting bubble characteristics in automobile brake pipeline |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010218489.5A CN111415348B (en) | 2020-03-25 | 2020-03-25 | Method for extracting bubble characteristics in automobile brake pipeline |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111415348A true CN111415348A (en) | 2020-07-14 |
CN111415348B CN111415348B (en) | 2023-05-26 |
Family
ID=71494632
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010218489.5A Active CN111415348B (en) | 2020-03-25 | 2020-03-25 | Method for extracting bubble characteristics in automobile brake pipeline |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111415348B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112926606A (en) * | 2021-04-25 | 2021-06-08 | 南京工程学院 | Heterogeneous fusion bubble plume feature extraction method based on deep learning |
CN115937295A (en) * | 2023-03-09 | 2023-04-07 | 中国空气动力研究与发展中心低速空气动力研究所 | Bubble shape type discrimination method based on set thought |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1120163A (en) * | 1993-11-24 | 1996-04-10 | 尼若·施瓦茨 | Dynamic fluid level and bubble inspection for quality and process control |
CN103332183A (en) * | 2013-07-08 | 2013-10-02 | 中国计量学院 | System and method for eliminating bubbles in automotive hydraulic ABS in real time |
US9195875B1 (en) * | 2007-10-22 | 2015-11-24 | Data Recognition Corporation | Method and apparatus for defining fields in standardized test imaging |
CN105869154A (en) * | 2015-09-23 | 2016-08-17 | 长沙理工大学 | Visible foreign matter and bubble classification recognition detection method for medical (250ml) |
US20160252636A1 (en) * | 2013-07-26 | 2016-09-01 | Berger Geosciences, LLC | System for monitoring a surface for gas and oil flow |
CN106023153A (en) * | 2016-05-09 | 2016-10-12 | 山东省科学院海洋仪器仪表研究所 | Method for measuring bubbles in water body |
CN106227975A (en) * | 2016-08-04 | 2016-12-14 | 中国核动力研究设计院 | Flow pattern Experimental Identification method under ocean condition |
CN109741282A (en) * | 2019-01-16 | 2019-05-10 | 清华大学 | A kind of multiframe bubble stream image processing method based on Predictor Corrector |
CN109858501A (en) * | 2019-02-20 | 2019-06-07 | 云南农业大学 | A kind of two phase flow pattern feature extracting method |
CN110245672A (en) * | 2019-06-22 | 2019-09-17 | 福州大学 | Flotation bubble crushing detection method based on the matching of quick retinal feature point and multi-scale division |
CN110415257A (en) * | 2019-07-23 | 2019-11-05 | 东南大学 | A kind of biphase gas and liquid flow overlapping bubble image partition method |
-
2020
- 2020-03-25 CN CN202010218489.5A patent/CN111415348B/en active Active
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1120163A (en) * | 1993-11-24 | 1996-04-10 | 尼若·施瓦茨 | Dynamic fluid level and bubble inspection for quality and process control |
US9195875B1 (en) * | 2007-10-22 | 2015-11-24 | Data Recognition Corporation | Method and apparatus for defining fields in standardized test imaging |
CN103332183A (en) * | 2013-07-08 | 2013-10-02 | 中国计量学院 | System and method for eliminating bubbles in automotive hydraulic ABS in real time |
US20160252636A1 (en) * | 2013-07-26 | 2016-09-01 | Berger Geosciences, LLC | System for monitoring a surface for gas and oil flow |
CN105869154A (en) * | 2015-09-23 | 2016-08-17 | 长沙理工大学 | Visible foreign matter and bubble classification recognition detection method for medical (250ml) |
CN106023153A (en) * | 2016-05-09 | 2016-10-12 | 山东省科学院海洋仪器仪表研究所 | Method for measuring bubbles in water body |
CN106227975A (en) * | 2016-08-04 | 2016-12-14 | 中国核动力研究设计院 | Flow pattern Experimental Identification method under ocean condition |
CN109741282A (en) * | 2019-01-16 | 2019-05-10 | 清华大学 | A kind of multiframe bubble stream image processing method based on Predictor Corrector |
CN109858501A (en) * | 2019-02-20 | 2019-06-07 | 云南农业大学 | A kind of two phase flow pattern feature extracting method |
CN110245672A (en) * | 2019-06-22 | 2019-09-17 | 福州大学 | Flotation bubble crushing detection method based on the matching of quick retinal feature point and multi-scale division |
CN110415257A (en) * | 2019-07-23 | 2019-11-05 | 东南大学 | A kind of biphase gas and liquid flow overlapping bubble image partition method |
Non-Patent Citations (4)
Title |
---|
QINGTAI XIAO等: "Extraction and evolution of bubbles attributes in a two-phase direct contact evaporator" * |
XIAOLU LI等: "Visualization of bubble flow in the channel of a dimple-type embossing plate heat exchanger under different fluid inlet/outlet ports" * |
穆晶晶: "基于凹点检测与PCA混合降维的气液两相流气泡分割重构与流型识别研究" * |
许联锋等: "稀疏气泡流动的粒子跟踪测速技术研究" * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112926606A (en) * | 2021-04-25 | 2021-06-08 | 南京工程学院 | Heterogeneous fusion bubble plume feature extraction method based on deep learning |
CN115937295A (en) * | 2023-03-09 | 2023-04-07 | 中国空气动力研究与发展中心低速空气动力研究所 | Bubble shape type discrimination method based on set thought |
Also Published As
Publication number | Publication date |
---|---|
CN111415348B (en) | 2023-05-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109101924B (en) | Machine learning-based road traffic sign identification method | |
CN103714541B (en) | Method for identifying and positioning building through mountain body contour area constraint | |
CN101567046B (en) | Target recognition method of unmanned aerial vehicle based on minimum circle-cover matching | |
CN103886619B (en) | A kind of method for tracking target merging multiple dimensioned super-pixel | |
CN108985170A (en) | Transmission line of electricity hanger recognition methods based on Three image difference and deep learning | |
CN107392885A (en) | A kind of method for detecting infrared puniness target of view-based access control model contrast mechanism | |
CN106530347A (en) | Stable high-performance circle feature detection method | |
CN112825192B (en) | Object identification system and method based on machine learning | |
CN114972356B (en) | Plastic product surface defect detection and identification method and system | |
CN104809736B (en) | Medical Slice Images closure bone contours computational methods based on priori | |
CN103913166B (en) | A kind of punctate opacity of the cornea extracting method based on Energy distribution | |
CN111415348A (en) | Method for extracting characteristics of bubbles in automobile brake pipeline | |
CN101630407A (en) | Method for positioning forged region based on two view geometry and image division | |
CN114022483A (en) | Injection molding flash area identification method based on edge characteristics | |
Alturki | Traffic sign detection and recognition using adaptive threshold segmentation with fuzzy neural network classification | |
CN108875454A (en) | Traffic sign recognition method, device and vehicle | |
CN112198170B (en) | Detection method for identifying water drops in three-dimensional detection of outer surface of seamless steel tube | |
CN110309729A (en) | Tracking and re-detection method based on anomaly peak detection and twin network | |
CN102201060B (en) | Method for tracking and evaluating nonparametric outline based on shape semanteme | |
Fang et al. | Towards real-time crack detection using a deep neural network with a Bayesian fusion algorithm | |
CN106372593B (en) | Optic disk area positioning method based on vascular convergence | |
Saha et al. | Segmentation of cervical nuclei using SLIC and pairwise regional contrast | |
CN105761237B (en) | Chip x-ray image Hierarchical Segmentation based on mean shift | |
CN115131539A (en) | Aluminum template automatic identification and classification system based on machine vision | |
Kiratiratanapruk et al. | Segmentation algoritm for touching round grain image |
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 |