CN109829944A - Particulate matter accumulated partial size statistical method based on image procossing - Google Patents
Particulate matter accumulated partial size statistical method based on image procossing Download PDFInfo
- Publication number
- CN109829944A CN109829944A CN201910219590.XA CN201910219590A CN109829944A CN 109829944 A CN109829944 A CN 109829944A CN 201910219590 A CN201910219590 A CN 201910219590A CN 109829944 A CN109829944 A CN 109829944A
- Authority
- CN
- China
- Prior art keywords
- particulate matter
- size
- image procossing
- image
- statistic
- 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.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
This patent is entitled " the particulate matter accumulated partial size statistical method based on image procossing ", belongs to computer vision, field of image processing.In production activity, it is often necessary to which, in face of counting the problem of a large amount of desirable particle sizes are distributed, this patent proposes a kind of method for carrying out size statistic to a large amount of particulate matter accumulated using image procossing.And caused objects in images edge adhesion, fuzzy situation are stacked with for being detected between particulate matter, this patent uses a kind of New Image Processing Algorithm that core is filtered into column, fully automatically the particle object image of edge adhesion can be split, be identified, size statistic, solve the problems, such as the automation size statistic of bulk deposition particulate matter.The present invention can achieve the effect that include: fully automatically to carry out counting statistics by the number of statistics particulate matter in given size section, be not necessarily to manual intervention；Accurate for each size section and real-time statistical result.
Description
Technical field
The present invention relates to a kind of partial size statistical methods for particulate matter accumulated.Particulate matter accumulated to be counted is clapped
According to obtaining the particulate matter quantity in each given section to achieve the purpose that partial size is counted by carrying out image procossing to it.This hair
It is bright to belong to computer vision, field of image processing.
Background technique
In production activity, it is often necessary in face of counting the problem of a large amount of desirable particle sizes are distributed.Traditional way is to use
The descending particulate matter for filtering out different-grain diameter of the different sieve of diameter, then count the particulate count filtered out under each diameter
Amount；Or a large amount of particulate matters to be counted are sampled, it is distributed using the method statistic size that manual sorting counts.Both sides
Method is although simple and easy, but it can not accomplish real-time statistics.And the method for this patent, it can use and take pictures, scheme in real time in real time
As the method for processing, real-time size statistic and monitoring are carried out to a large amount of particulate matter accumulated.
It is stacked with caused objects in images edge adhesion, fuzzy situation between detected particulate matter,
The invention patent uses a kind of New Image Processing Algorithm that core is filtered into column, can be fully automatically to edge adhesion
Particle object image be split, identify, size statistic, solve the problems, such as the automation size statistic of bulk deposition particulate matter.
Summary of the invention
It is an object of the invention to solve problems faced, such as user during the size statistic of current a large amount of particulate matters
Work takes time and effort, the lag of statistical result caused by sampling Detection with it is inaccurate etc..To solve these problems, this patent with image at
Reason is technical way, and the counting of each size section object is completed using the captured image by statistics particulate matter accumulated,
And it proposes a set of pair of image and carries out processing to reach the process of size statistic purpose.The process includes: that pre-processing takes
The column filtering of noise spot, specific dimensions except rough profile, removal edge, count the particle detected under the size
Object number.Summary of the invention is related to the image processing algorithm of process flow and key link.
The present invention can achieve the effect that include: fully automatically to counted in size section by the number of statistics particulate matter
Number statistics, is not necessarily to manual intervention；Accurate for each size section and real-time statistical result.
Detailed description of the invention
Fig. 1: the original sample grayscale image of pre-processing
Fig. 2: the sample effect picture after the completion of pre-processing
Fig. 3: filtered sample effect picture is counted
Fig. 4: the schematic diagram of the influence for the point that frame is 0 to boundary upper value in threshold radius filtering
Fig. 5: the schematic diagram of the influence for the noise spot that frame is 0 to value in threshold radius filtering
Fig. 6: the filtered sample effect picture of threshold radius
Fig. 7: column filters schematic diagram
Fig. 8: column Filtering position relation schematic diagram when difference is 255
Fig. 9: column Filtering position relation schematic diagram when difference is 0
Figure 10: the filtered sample effect picture of column
Figure 11: column Filtering position relation schematic diagram when there is noise spot in inside
Figure 12: the filtered center of circle motion track sample effect picture of column
Figure 13: system flow chart
Figure 14: test sample grayscale image
Figure 15: the filtered result figure of large-size particle object column in test sample
Figure 16: the result figure in the center of circle after the large-size particle object column filtering in test sample
Figure 17: the filtered result figure of intermediate sizes particulate matter column in test sample
Figure 18: the result figure in the center of circle after the centre cun particulate matter column filtering in test sample
Figure 19: the filtered result figure of small sized particles object column in test sample
Figure 20: the result figure in the center of circle after the small sized particles object column filtering in test sample
Specific embodiment
One, pre-processing
This part processing intent is the binaryzation profile that particulate matter accumulated to be counted is obtained from original image.Sample original image
As shown in Figure 1, treated effect such as Fig. 2 after gray proces.By texture and the background institute of body surface to be counted
The interference of introducing can introduce noise spot after the processing of the part.
Two, noise spot is removed
Since the region that column filtering needs to guarantee to surround inside edge does not occur the noise spot of local minimum, so
First noise spot (noise spot especially in particulate matter profile) should be eliminated before column filtering.This part is specifically divided
For two steps: counting filtering and threshold radius filtering.After the filtering of two steps, the noise spot in image can almost disappear completely
It removes.
1. counting filtering
It is filtered as the first order, it is therefore an objective to remove a large amount of trifling noise spots.
A window is chosen, the pixel for being 0 to its intrinsic value counts, if the count results in window are less than setting
The center pixel of window is then set to 255 by percentage.Same Fig. 2 of sample effect picture before counting filtering, effect that treated is as schemed
Shown in 3.It can be seen that the lesser noise spot of diameter is largely removed in filtered image.
2. threshold radius filters
The noise spot for still having some larger particles by counting filtered image, they can seriously affect subsequent
The effect of column filtering, this step are handled for this problem.
Image is obtained after first part is handled, and binaryzation, value only have 0 and 255 two kind of situation, are 0 for being worth
Pixel may be invocation point on boundary (needing to retain), it is also possible to which noise spot (needs to remove).Filter in this step
Wave method in two kinds of situation and can remove noise spot in area well.
Consider to be located at the point that on boundary and value is 0 first, is easy discovery for continuous boundary, a window can not be found
Allow to frame the pixel that its upper value is 0, that is to say, that centainly there is the pixel that value is 0 to pass through in four sides of window extremely
Few one side (as shown in Figure 4).Even if the sufficiently large entire edge contour for allowing to frame a target object of frame, but by
In by statistics object it is a large amount of stacking cause the edge contour of other objects that can also pass through four side of window.
And for isolating and being worth the noise spot for being 0 relatively, it is clear that can find a window frames it completely, such as Fig. 5
It is shown.
Radius threshold filtering can be carried out as feature, sample effect picture such as Fig. 3 before step execution, treated
Effect picture is as shown in Figure 6.There may be two problems after processing: from the closer noise spot of profile because profile influence without
The interrupted profile in removal, part has been taken as noise spot to be removed.For the former, the closer noise spot of profile is to subsequent column
Shape filtering influences little；And although the latter causes some profiles to be removed, but also there is no too to the integrality of entire profile
It is big to influence, subsequent column will not be filtered and be impacted.
Three, column filters
By the processing of both the above step, the figure of general profile is had been obtained almost without noise spot and remained with
Picture, as shown in Figure 6.It considers how to obtain object to be detected from incomplete edge contour at this time.It is endless due to edge
It is whole and there are interference caused by adjacent object edge, so directlying adopt effect if Hough transformation can be very poor.
The thinking of column filtering is to be slightly less than the current circle for being intended to detecting size with a radius to move inside particulate matter, by
It is slightly less than particulate matter radius in its radius, so circle can be still stuck in profile when the profile of particulate matter occurs interrupted
Portion, thus can dispose edge it is interrupted caused by influence, to successfully detect the particulate matter of given radius, schematic diagram
As shown in Figure 7.
This thinking is realized, can be carried out by the following method.Since the value at edge is 0 and internal value is 255, then
The circle that a radius is slightly less than radius to be detected and value is 255 is constructed, which is moved and made the difference on the image, is found on image
The maximum value of each pixel difference in each pixel and circle；If its value illustrates that the edge of particulate matter have passed through round inside for 255
(as shown in Figure 8), does not operate at this time；Illustrate that circle is placed exactly in the inside (as shown in Figure 9) of particulate matter if its value is 0.This
When the circle is projected in another Zhang Changkuan black background image equal with original image, when circle smoothly moved inside edge when,
The particulate matter of the size section can be depicted out by its circle projected.Four great circles in detection Fig. 6, knot are filtered using column
Fruit is as shown in Figure 10.In addition it can be found that if occurring noise spot inside particulate matter, can make become for 0 maximum difference
It is 255, leads to the object detection failure (as shown in figure 11) to be counted.This is namely why one in the processing of second part
Surely the reason of removing particulate matter internal interference point to be counted.
When counting size, count from big to small, it is every counted a size after the testing result of the size is existed
Be set to 0 in original image (to avoid subsequent detection more minor radius object when generate repeat to detect), so repeat that institute can be counted
The number of particulate matter in some size sections.
Four, number statistical
Processing result in previous step can show the particulate matter under certain radius size well, if but directly to it
Counted two pelletizing edge adhesions for being also difficult to solve the problems, such as to be likely to occur.Also, due to having used radius to be slightly less than
The circle of radius to be detected carries out column filtering, and the tangent object of two scripts can be presented as the entirety of a connection, make the phenomenon
It is more obvious.So the counting operation of this step is directly using the particle object image detected, but column is used to filter
The center of circle of core (i.e. along that circle of interior of articles to be counted sliding).Due to apart from each other between the center of circle, two balls can be solved
The problem of group's adhesion, as shown in figure 12.
For the number statistical in the center of circle, the method counted using connected domain.This method is since " 1 " by the picture of each connection
Plain region carries out label, is designated as consistent number if pixel connection, if compiling after not being connected to number plus 1 to new region
Number.Also the meter of object to be counted under the size is achieved that after the completion of the connected component labeling for being so 255 by all values in figure
Number.
The process flow diagram of above each step is as shown in figure 13.
Sample operation result
This part dimensions indicated above statistical method carries out size statistic to a sample, to examine the accuracy of this method.
The original image of sample is as shown in figure 14, having a size of 541*474.Wherein it is divided into large, medium and small three size sections, counterpart
Body number is respectively 9,27,36.
For large scale, the filtered result figure of column and center of circle result figure are respectively as shown in Figure 15, Figure 16, by number
It shows that result is 9 after mesh statistics, consistent with the actual number in original image；It is 123 milliseconds time-consuming to detect large scale.For intermediate ruler
Very little, the filtered result figure of column and center of circle result figure are respectively as shown in Figure 17, Figure 18, its display is tied after number statistical
Fruit is 27, consistent with the actual number in original image；It is 148 milliseconds time-consuming to detect intermediate sizes.For small size, after column filtering
Result figure and center of circle result figure respectively as shown in Figure 19, Figure 20, it shows result in 36, with original image after number statistical
Actual number it is consistent；It is 157 milliseconds time-consuming to detect small size.For executing time, the device configuration run in experiment above are as follows:
Operating system is 10 family's Chinese edition of Windows, CPU is Intel Core i7-6700HQ, RAM 16GB.As a result summarize such as 1 institute of table
Show.
1 sample operation result of table summarizes
| Greatly | It is medium and small | It is small | |
| Actual quantity (a) | 9 | 27 | 36 |
| Success identifies quantity (a) | 9 | 27 | 36 |
| It executes time (millisecond) | 123 | 129 | 147 |
For the sample, the number of object has obtained accurate statistics under three different size sections, presents this
The high degree of accuracy of method in the application.And the execution time of each size section also embodies this method in 0.15 second
With good real-time statistics ability.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (3)
1. the particulate matter accumulated partial size statistical method based on image procossing, which is characterized in that using image procossing to a large amount of heap
Product particulate matter carries out size statistic, fully automatically carries out counting statistics, nothing by the number of statistics particulate matter in given size section
Need manual intervention；Accurate for each size section and real-time statistical result.
2. the particulate matter accumulated partial size statistical method according to claim 1 based on image procossing, characterized in that be directed to quilt
It is stacked with caused objects in images edge adhesion, fuzzy situation between detection particulate matter, this patent uses one
Kind is filtered into the New Image Processing Algorithm of core with column, can fully automatically divide the particle object image of edge adhesion
It cuts, identify, size statistic, solving the problems, such as the automation size statistic of bulk deposition particulate matter.
3. the particulate matter accumulated partial size statistical method according to claim 1 based on image procossing, characterized in that comprising with
Lower process flow: pre-processing obtains the column filtering of noise spot, specific dimensions except rough profile, removal edge, unites
Count the particulate matter number detected under the size.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910219590.XA CN109829944A (en) | 2019-03-21 | 2019-03-21 | Particulate matter accumulated partial size statistical method based on image procossing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910219590.XA CN109829944A (en) | 2019-03-21 | 2019-03-21 | Particulate matter accumulated partial size statistical method based on image procossing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109829944A true CN109829944A (en) | 2019-05-31 |
Family
ID=66870652
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910219590.XA Pending CN109829944A (en) | 2019-03-21 | 2019-03-21 | Particulate matter accumulated partial size statistical method based on image procossing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109829944A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110455825A (en) * | 2019-08-27 | 2019-11-15 | 无锡领先针测电子有限公司 | The method that particle on a kind of pair of glass is counted |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101620060A (en) * | 2009-08-13 | 2010-01-06 | 上海交通大学 | Automatic detection method of particle size distribution |
| CN105931277A (en) * | 2016-04-01 | 2016-09-07 | 天津工业大学 | Partial differential equation-based nano-particle size measurement method |
| CN107909138A (en) * | 2017-11-14 | 2018-04-13 | 江苏大学 | A kind of class rounded grain thing method of counting based on Android platform |
| CN108734731A (en) * | 2018-05-29 | 2018-11-02 | 中科院广州电子技术有限公司 | A kind of class rounded grain object count method based on machine vision |
-
2019
- 2019-03-21 CN CN201910219590.XA patent/CN109829944A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101620060A (en) * | 2009-08-13 | 2010-01-06 | 上海交通大学 | Automatic detection method of particle size distribution |
| CN105931277A (en) * | 2016-04-01 | 2016-09-07 | 天津工业大学 | Partial differential equation-based nano-particle size measurement method |
| CN107909138A (en) * | 2017-11-14 | 2018-04-13 | 江苏大学 | A kind of class rounded grain thing method of counting based on Android platform |
| CN108734731A (en) * | 2018-05-29 | 2018-11-02 | 中科院广州电子技术有限公司 | A kind of class rounded grain object count method based on machine vision |
Non-Patent Citations (2)
| Title |
|---|
| YI ZHANG ET AL: "Two-Stage Obstacle Detection Based on Stereo Vision in Unstructured Environment", 《2014 SIXTH INTERNATIONAL CONFERENCE ON INTELLIGENT HUMAN-MACHINE SYSTEMS AND CYBERNETICS》 * |
| 王培昌 等: "基于局域最大亮度的靶状激光图像边缘检测", 《光电工程》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110455825A (en) * | 2019-08-27 | 2019-11-15 | 无锡领先针测电子有限公司 | The method that particle on a kind of pair of glass is counted |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115239704B (en) | Accurate detection and repair method for wood surface defects | |
| CN111179225B (en) | Test paper surface texture defect detection method based on gray gradient clustering | |
| DE69610925T2 (en) | DEVICE FOR EXAMINING BOTTLE THREADS AND METHOD FOR THE OPERATION THEREOF | |
| DE69328533T2 (en) | Method for checking a solid for foreign matter | |
| WO2022088620A1 (en) | State detection method and apparatus for camera lens, device and storage medium | |
| CN106778668B (en) | A kind of method for detecting lane lines of robust that combining RANSAC and CNN | |
| CN107220649A (en) | A kind of plain color cloth defects detection and sorting technique | |
| CN107328787A (en) | A kind of metal plate and belt surface defects detection system based on depth convolutional neural networks | |
| CN104458766B (en) | A kind of cloth surface flaw detection method based on structural texture method | |
| CN104215639A (en) | Pear surface defect detection method based on machine vision | |
| Heydari et al. | An industrial image processing-based approach for estimation of iron ore green pellet size distribution | |
| CN106485696B (en) | A kind of detection method of the explosive dangerous material stitch defect based on machine vision | |
| CN104056789B (en) | Carrot defect image quantitative detection method | |
| CN101063660A (en) | Method for detecting textile defect and device thereof | |
| CN106780437B (en) | A kind of quick QFN chip plastic packaging image obtains and amplification method | |
| CN111047556A (en) | Strip steel surface defect detection method and device | |
| CN106529551B (en) | Intelligent recognition counting detection method for round-like objects in packaging industry | |
| CN116228651A (en) | Cloth defect detection method, system, equipment and medium | |
| CN114820626B (en) | Intelligent detection method for automobile front face part configuration | |
| WO2023231262A1 (en) | Hoisting wire rope tension measurement method based on visual vibration frequency identification | |
| CN113706566A (en) | Perfuming spray performance detection method based on edge detection | |
| CN109829944A (en) | Particulate matter accumulated partial size statistical method based on image procossing | |
| Choi et al. | Real-time defects detection algorithm for high-speed steel bar in coil | |
| CN113112458A (en) | Metal surface defect detection method based on support vector machine | |
| Khule et al. | Automated object counting for visual inspection applications |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190531 |