CN115661146B - Production quality detection method of liquid propellant - Google Patents
Production quality detection method of liquid propellant Download PDFInfo
- Publication number
- CN115661146B CN115661146B CN202211670432.4A CN202211670432A CN115661146B CN 115661146 B CN115661146 B CN 115661146B CN 202211670432 A CN202211670432 A CN 202211670432A CN 115661146 B CN115661146 B CN 115661146B
- Authority
- CN
- China
- Prior art keywords
- point
- damage
- suspected
- sequence
- suspected damage
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the field of image processing, in particular to a production quality detection method of a liquid volatile agent, which comprises the following steps: acquiring a sequence of projection images of a bottle of liquid propellant; obtaining a bottleneck area of each projection image according to a judgment index of each line in each projection image in the projection image sequence; acquiring each corner feature triple in each bottleneck region; clustering according to the feature triples of each corner point to obtain a plurality of corner point clusters; obtaining the abnormal degree of each corner point according to the difference between the feature triples of each corner point in each corner point cluster, and further obtaining each suspected damage point; obtaining each suspected damage sequence; and calculating the damage probability of each suspected damage sequence to obtain each damage position. The invention can avoid the interference of the information texture of the filling bottle and ensure the accuracy of identifying the damaged position.
Description
Technical Field
The invention relates to the field of image processing, in particular to a production quality detection method of a liquid volatile agent.
Background
For some aromatic traditional Chinese medicine products capable of refreshing, such as snuff type essential balm, the main components of the aromatic traditional Chinese medicine products often comprise menthol, borneol and other components with strong volatility, the production quality of the products is required to be ensured, the quality of the products in the preparation process is required to be ensured, the quality of the products in the packaging and transportation process is also required to be ensured, the bottle body containing the volatile products is required to have a good sealing effect, the bottle opening of the containing bottle is a part for receiving the bottle cover to realize the sealing of the container, and therefore the quality of the bottle opening area can directly influence the sealing effect of the containing bottle. However, during the transportation or packaging of the bottle blank, the phenomena of collision, extrusion and the like inevitably occur, which damages the screw threads of the bottle mouth of the volatile medicament containing bottle, affects the sealing quality of the product, and further affects the production quality of the product, so that the bottle mouth quality of the volatile medicament containing bottle needs to be detected.
The existing detection of bottle mouth threads usually uses edge detection, but because some product related information is usually printed on the surface of a product containing bottle, if the edge detection is directly carried out on the surface of the bottle body, the character information on the bottle body can interfere the detection of the bottle mouth threads; the bottle mouth threads have transition areas and certain inclination angles, so that the positions of the bottle mouth threads collected in different directions are different, and the shapes of the threads in different areas are also different, so that the accuracy of the bottle mouth thread quality detection result by the traditional method is lower.
Disclosure of Invention
The invention provides a production quality detection method of a liquid volatile agent, which aims to solve the existing problems.
The production quality detection method of the liquid volatile agent adopts the following technical scheme:
one embodiment of the invention provides a production quality detection method of a liquid propellant, which comprises the following steps:
acquiring a sequence of projection images of a bottle of liquid propellant; acquiring each edge point in each projection image in the projection image sequence; taking the accumulated sum of the abscissa of all edge points contained in each line of each projection image as a judgment index of each line, and obtaining the bottleneck area of each projection image according to the difference of the judgment index of each line between the projection images;
acquiring each corner point in each bottleneck area; taking the average value of the horizontal coordinates of each corner point in each bottleneck area as the area segmentation threshold of each bottleneck area, dividing the corner point of which the horizontal coordinate of each corner point is smaller than the area segmentation threshold into a first area, and dividing the corner point of which the horizontal coordinate of each corner point is larger than the area segmentation threshold into a second area;
obtaining a first chain code and a second chain code of each angular point according to each angular point in each first region and the adjacent angular point of each angular point, wherein the coordinates of each angular point and the first chain code and the second chain code of each angular point form a feature triple of each angular point; clustering feature triples of all the corners in all the first regions to obtain a plurality of corner clusters; obtaining the abnormal degree of each corner according to the difference between the feature triples of each corner in each corner cluster; obtaining each suspected damage point according to the abnormal degree of each corner point; similarly, obtaining each suspected damage point in the second area;
obtaining each suspected damage sequence according to the coordinates of each suspected damage point and the serial number of the projection image where each suspected damage point is located; obtaining the damage probability of each suspected damage sequence according to the abscissa of each suspected damage point in each suspected damage sequence; and obtaining each damage position according to the damage probability of each suspected damage sequence.
Preferably, the method for acquiring the bottleneck area of each projection image comprises the following steps:
performing adaptive threshold segmentation on each projection image to obtain a projection area of each projection image; and calculating the variance of the judgment indexes of each line between the judgment indexes corresponding to the projection images, when the variance is 0, judging each line to be a bottleneck area, otherwise, judging each line in the projection images in sequence to obtain all line number sequences belonging to the bottleneck area, and taking the line of which the line number belongs to the line number sequence in the projection area of each projection image as the bottleneck area of each projection image.
Preferably, the obtaining expression of the abnormal degree of each corner point is as follows:
in the formula (I), the compound is shown in the specification,
the abnormal degree of the kth corner point in the first area of the ith bottleneck area is obtained;
the abscissa of the kth angular point is taken as the abscissa;
a first chain code of the kth angular point;
a second chain code of the kth angular point;
，
and
respectively taking the average value of the abscissa of all the angular points in the angular point cluster to which the kth angular point belongs, the average value of the first chain code and the average value of the second chain code;
is L2 norm;
is an exponential function with a natural constant as the base.
Preferably, the method for obtaining each suspected damage point according to the abnormal degree of each corner point comprises:
and setting a threshold value according to experience, wherein when the abnormal degree of each corner point is greater than the threshold value, each corner point is a suspected damage point, otherwise, each corner point is not considered as a suspected damage point.
Preferably, the method for acquiring each suspected damage sequence comprises:
for a suspected damage point which is not included in a suspected damage sequence, acquiring a suspected damage point which is closest to the adjacent projection image in the projection image sequence according to the serial number of the projection image where the suspected damage point is located in the projection image sequence, including the coordinates of the suspected damage point into one suspected damage sequence, and acquiring a suspected damage point which is closest to the newly included suspected damage point in the adjacent projection image until the suspected damage point which is closest to the newly included suspected damage point does not exist in the adjacent projection image; and sequentially processing each suspected damage point to obtain each suspected damage sequence.
Preferably, the method for determining the damage probability of each suspected damage sequence comprises:
obtaining a damage curve of each suspected damage sequence according to the serial number of each suspected damage point in each suspected damage sequence and the abscissa of each suspected damage point, and calculating the slope of each suspected damage point in each damage curve; and obtaining an average value of the ratio according to the ratio of the slope of the adjacent suspected damage points in each suspected damage sequence to the absolute value of the slope, and taking the average value as the damage probability of the suspected damage sequence.
The invention has the beneficial effects that: firstly, acquiring projected images of a containing bottle of a liquid propellant in the process of rotating for one circle, avoiding the interference of surface information textures of the containing bottle, and then extracting a bottle opening area in each projected image according to the characteristic that the coordinate of an edge point of the bottle opening area of the containing bottle in the projected images can change by combining the characteristic that the containing bottle is mostly a cylinder; performing angular point detection on each bottle mouth area, representing the shape of the bottle mouth thread by using a chain code between adjacent angular points, representing the structural information of each angular point by using a characteristic triple as the tooth height of the thread by using the abscissa of each angular point, thereby obtaining the shape characteristic of the whole thread; and finally, obtaining damage positions according to the damage probability, avoiding mistakenly identifying transition regions in the bottleneck region as the damage positions, ensuring the accuracy of identifying the damage positions of the bottleneck region, and further ensuring the production quality of products.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of the steps of the method for detecting the production quality of a liquid propellant;
FIG. 2 is a schematic view of a bottle mouth region of the method for detecting the production quality of a liquid propellant.
Detailed Description
In order to further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the method for detecting the production quality of a liquid propellant according to the present invention, the specific implementation manner, structure, characteristics and effects thereof will be provided in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the production quality detection method of the liquid propellant in detail with reference to the accompanying drawings.
Referring to fig. 1, a flow chart of steps of a method for detecting production quality of a liquid propellant according to an embodiment of the present invention is shown, the method includes the following steps:
step S001: and (3) acquiring continuous frame projection images by using a camera in the process of rotating the containing bottle of the liquid propellant for one circle to obtain a projection image sequence.
Because the surface of the containing bottle of the volatile medicament can also be printed with the related text information or pattern information of some products, the information can also generate textures on the surface of the containing bottle, and the interference of the surface textures of the containing bottle cannot be eliminated when the edge detection is directly carried out on the surface of the containing bottle to identify the damaged area.
The invention polishes the side surface of the containing bottle on a production line, arranges background cloth on the backlight side of the containing bottle and directly collects the image projected on the background cloth by the containing bottle so as to eliminate the interference of the information texture on the surface of the containing bottle. In order to obtain complete bottle opening threads, the camera is arranged in the backlight area, continuous frame projection image acquisition is carried out in the process of rotating the bottle for one circle, and gray processing is carried out on each acquired image to obtain a projection image sequence of the bottle.
Step S002: and obtaining the bottleneck area in each projection image according to the judgment index of each line of each projection image in the projection image sequence.
In the actual production process, in order to make product portable, the product containing bottle that has the refreshing efficiency is the cylinder mostly, regional width such as body is unanimous, and the screw thread is owing to there being certain inclination, at the in-process of the rotatory a week of containing bottle, pixel rigidity in the body region is unchangeable, pixel position on the outside edge can change along with the change of screw thread in the bottleneck region, consequently can confirm the bottleneck region according to the change of pixel coordinate in the different projection images in the projection image sequence, concrete process is as follows:
because the color difference between the containing bottle in the projection image and the background cloth is larger, the method firstly uses the OTSU algorithm to perform self-adaptive threshold segmentation on each projection image in the projection image sequence to obtain the projection area in each projection image, and each projection image corresponds to one projection area;
using a Canny operator to carry out edge detection on each projection area to obtain all edge points in each projection area, and taking the accumulated sum of the horizontal coordinates of all the edge points on each line in each projection area as a judgment index of whether the line belongs to a flat area, wherein the ith projection areaThe sum of the abscissas of all the edge points on the jth line of the shadow region is recorded as
Then, then
That is, the judgment index of the jth line of the ith projection area is obtained, the variance between the judgment indexes of the jth line in different projection images in the projection area is calculated, the obtained variance is taken as the probability that the jth line of the ith projection area belongs to the bottleneck area, and the probability that the jth line of the ith projection area belongs to the bottleneck area is taken as
。
Due to the columnar structure of the containing bottle, the coordinate of each pixel point in the bottle body area cannot change in different projection images, so that the judgment index of each line in the bottle body area cannot change in different projection images, the probability of belonging to the bottle opening area corresponding to each line is equal to 0, the coordinate of each line of pixel points can change due to the change of threads in the rotation process of the containing bottle in the bottle opening area, and the probability of belonging to the bottle opening area of each line is not 0;
thus when
And if the j line does not belong to the bottleneck area, otherwise, the j line belongs to the bottleneck area, sequentially processing each line in the projection area until the last line is processed, and taking all lines belonging to the bottleneck area in each projection image as the bottleneck area corresponding to each projection image, wherein the number of the bottleneck areas is equal to the number of the projection areas and the number of the projection images.
Step S003: carrying out angular point detection on each bottle mouth area; obtaining a feature triple of each angular point by combining the coordinates of each angular point according to the chain code between each angular point and two angular points adjacent to the angular point; and obtaining suspected damage points according to the difference between the feature triples of each corner point in the bottle mouth area.
Firstly, carrying out angular point detection on a bottleneck region of each projection image by using an SIFT operator to obtain all angular points in each bottleneck region and coordinates of each angular point;
because the threads surround the bottle mouth and show that the threads are arranged on two sides of the bottle mouth in a projection image, the thread edges on different sides have larger change difference, and the thread edges on the same side have more similar change, the invention firstly divides the area according to the position relation of each angular point relative to the middle part of the bottle mouth in each area of the bottle mouth, thereby dividing the angular points on the same side into one area, and then respectively judges the angular points contained in the two areas obtained by the division.
Taking the ith bottleneck area as an example, obtaining the average value of the horizontal coordinates of all corner points in the bottleneck area, taking the average value as an area division threshold, dividing the corner points of which the horizontal coordinates are smaller than the area division threshold into a first area, and dividing the corner points of which the horizontal coordinates are larger than the area division threshold into a second area;
sequencing all corner points in the first area from small to large according to the ordinate to obtain a first sequencing sequence
To do so by
The two chain codes are obtained according to the coordinates of edge points between each angular point and two adjacent angular points, wherein the chain code between each angular point and the angular point with the minimum sequence number in the two adjacent angular points is recorded as a first chain code, the chain code between the angular point with the maximum sequence number is recorded as a second chain code, the acquisition process of the chain codes is a known technology, and the process is not repeated herein.
Then to
For the kth corner point in (b), the chain code between the corner point and the (k-1) th corner point is recorded as the first chain code of the kth corner point
The chain code between the angular point and the (k + 1) th angular point is taken as the second chain code of the kth angular point
Then the feature triplet of the kth corner point in the first region can be represented as
Wherein
Coordinates of a kth angular point; and similarly, obtaining a feature triple of each corner point in the second area.
The thread can be divided into triangular thread, rectangular thread, trapezoidal thread, zigzag thread and the like according to the cross-sectional shape, i.e. the thread form, wherein the triangular thread is mainly used for connection, and the rectangular, trapezoidal and zigzag thread is mainly used for transmission, so that the thread form for connecting the bottle cap and the bottle body is mostly triangular thread, and each angular point in the mouth region of the bottle can be divided into three types according to the characteristic triad obtained by the method, for example, one thread form in the mouth region of the bottle comprises three angular points: a tooth top angular point A, a tooth bottom angular point with a smaller vertical coordinate is B, a tooth bottom angular point with a larger vertical coordinate is C, as shown in FIG. 2, because the coordinate of A, the first chain code and the second chain code are different from B and C, and the first chain code and the second chain code of B and C are also different, A, B and C respectively correspond to an angular point type, the first sample data is formed by characteristic triples of all angular points in the first region of all bottleneck regions, the first sample data is clustered by using a K-means clustering algorithm, the number of the angular point clusters is set to be 3, and therefore three angular point clusters are obtained, and each angular point cluster corresponds to one angular point type;
under normal conditions, the thread profile and the thread height of the thread are the same, so that the difference between feature triplets of the same type of corner points is smaller, and the method judges according to the difference between the feature triplets of the same type of corner pointsCutting off the abnormal degree of each corner point, the abnormal degree of the kth corner point in the first area corresponding to the ith bottleneck area
Can be expressed as:
in the formula (I), the compound is shown in the specification,
the abnormal degree of the kth corner point in the first area of the ith bottleneck area is obtained;
the abscissa of the kth angular point is taken as the abscissa;
a first chain code of the kth angular point;
a second chain code of the kth angular point;
，
and
respectively taking the average value of the abscissa of all the angular points in the angular point cluster to which the kth angular point belongs, the average value of the first chain code and the average value of the second chain code;
is L2 norm;
is based on a natural constant ofThe base exponential function.
The thread is not damaged, so that the tooth height of the thread is ensured to be normal, and the shape of the thread is normal. The thread crest can also spirally rise or fall along with the rotation of the filling bottle, but because the thread height is fixed, under normal conditions, the vertical coordinates of the corner points in the same corner point cluster have difference, and the horizontal coordinates are basically similar, so that the invention only takes the difference of the horizontal coordinates of each corner point in the same corner point cluster as an index of the abnormal degree, when the screw is used for filling the bottle, the screw is not screwed down, and the screw is not screwed down, so that the screw is not screwed down, and the screw is not screwed down, so that the screw is not easy to be broken
And
the larger the difference between the k-th corner points is, the higher the abnormal degree of the k-th corner point is;
although the angular point descriptor obtained by using the SIFT operator is also a shape feature description, the descriptor can only be used for describing features in a local area, no association exists between two adjacent angular points, and the change of the shape of the thread between the two angular points cannot be described, namely, according to the fact that the angular point cannot describe the shape of a single thread, and the chain code is a direction sequence obtained according to the coordinate change of adjacent points, different shapes can cause different edge changes, so that different chain codes are obtained, therefore, the chain code between the adjacent angular points can be used for representing the change of the thread shape, and the thread tooth form is also fixed, when the first chain code of the kth angular point is used as the first chain code
Average value of first chain codes of same angular point cluster
The greater the difference between, or the second chain code
Average value of second chain codes of same angular point cluster
The larger the difference between the two codes is, the larger the abnormal degree of the angular point is, the higher the probability that the angular point is a damaged point is, the more the first chain code and the second chain code are obtained according to the coordinates of the edge point between the two angular points, and the number of the edge points between the two angular points is multiple, namely the first chain code and the second chain code both comprise multiple data, so that the invention uses L2 norm comparison to compare the two numbers of the edge points with each other
And
and
and
the difference between them.
Similarly, in the second region of all the bottle mouth regions, the feature triples of each corner point form second sample data, and the abnormal degree of each corner point in the second region is calculated according to the second sample data.
Setting a first threshold value
When is coming into contact with
And if not, the corner point is considered as a normal point, and all the corner points in the bottleneck area are sequentially judged to obtain all the suspected damage points in all the bottleneck areas.
Step S004: obtaining each suspected damage sequence according to the coordinates of each suspected damage point; and obtaining the damage probability of each suspected damage sequence according to the coordinate change of each suspected damage point in each suspected damage sequence, thereby determining the damage position.
When the thread in the bottleneck region is damaged, the thread profile of the thread may be abnormal, however, since the thread start point and the thread end point have transition regions with the normal thread, the thread profile in the transition region and the thread profile at other positions also have a certain difference, according to the method in step S003, the corner point of the transition region of the thread is considered as a suspected damaged point, and therefore the suspected damaged point needs to be further analyzed, so as to obtain a true damaged point.
The difference between the transition area formed by the thread starting point or the thread ending point and the damage position is that the thread in the transition area is gradually convex, the change corresponding to the thread height is continuous, the damage point is the damage to the thread due to an external force factor, and the change of the thread height in the transition area is abrupt, so that the normal transition area and the damage position can be distinguished according to the height of the suspected damage point on the projection image of the adjacent frame, namely the trend consistency of the coordinate change of the suspected damage point.
When a bottle opening thread is damaged, a damaged area gradually disappears or gradually appears in the rotation process of a filling bottle, so that the coordinate change of a suspected damaged point corresponding to a damaged position in an adjacent projection image is small, for a certain suspected damaged point, a suspected damaged point closest to the adjacent projection image is obtained according to the serial number of the projection image in which the suspected damaged point is located in the projection image sequence, the obtained suspected damaged point is regarded as the same damaged position, the coordinates of the suspected damaged point are classified into a suspected damaged sequence, and then the suspected damaged point closest to the newly classified suspected damaged point in the adjacent projection image is obtained until the suspected damaged point closest to the newly classified suspected damaged point does not exist in the adjacent projection image; and then repeating the method for the suspected damage points which are not classified into any suspected damage sequence until all the suspected damage points are processed to obtain all the suspected damage sequences, wherein each suspected damage sequence corresponds to one suspected damage position.
The suspected injury in each suspected injury sequenceSorting the damage points from small to large according to the vertical coordinate, using the serial number of each suspected damage point in each sorted suspected damage sequence and the corresponding horizontal coordinate as sample data, performing curve fitting on the sample data by using a least square method to obtain a damage curve corresponding to each suspected damage sequence, calculating the slope of each suspected damage point on each damage curve, and then for the damage probability of the tth suspected damage sequence belonging to the damage position
Can be expressed as:
in the formula (I), the compound is shown in the specification,
is the slope of the r-th suspected damage point in the suspected damage sequence, and N is the total number of the suspected damage points contained in the suspected damage sequence.
Because the thread height of the thread in the thread transition region is gradually raised or gradually returns to be flat, when the thread in the transition region is gradually raised, the thread is gradually raised, the abscissa representing the angular point of the thread in the projection image is gradually increased, and the increasing degree is gradually reduced; when the transition area is that the thread exists or does not exist, the thread is gradually restored to be flat, the abscissa representing the corner point of the thread is gradually reduced in a projection image, the degree of reduction is gradually increased, the variation trend of adjacent suspected damage points in the sequence is corresponded, namely, the difference value between the slopes obtained according to the abscissas of the adjacent damage points is positive or negative, the variation trends of the adjacent suspected damage points in the sequence are accumulated and then averaged, the obtained result is 1 or-1, the damage caused by external force has certain randomness and is difficult to meet the conditions, and therefore when the absolute value of the probability that the suspected damage sequence belongs to the damage position is 1, namely the absolute value of the probability that the suspected damage sequence belongs to the damage position is 1
When the utility model is used, the water is discharged,the suspected damage sequence is not considered as a damage position, otherwise, the suspected damage point is considered as a damage point, if the containing bottle is used, the sealing effect of the product is poor, the production quality of the product is affected, and the containing bottle needs to be removed.
Through the steps, the production quality detection of the liquid volatile agent is completed.
The method comprises the steps of firstly, acquiring projected images of a containing bottle of the liquid propellant in the process of rotating for one circle, avoiding the interference of surface information textures of the containing bottle, and then extracting bottle opening areas in each projected image according to the characteristic that the coordinates of edge points of the bottle opening areas of the containing bottle in the projected images can change by combining the characteristic that the containing bottle is mostly cylindrical; performing angular point detection on each bottle mouth area, representing the shape of the bottle mouth thread by using a chain code between adjacent angular points, representing the structural information of each angular point by using a characteristic triple as the tooth height of the thread by using the abscissa of each angular point, thereby obtaining the shape characteristic of the whole thread; and finally, obtaining damage positions according to the damage probability, avoiding mistakenly identifying transition regions in the bottleneck region as the damage positions, ensuring the accuracy of identifying the damage positions of the bottleneck region, and further ensuring the production quality of products.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (6)
1. A production quality detection method of a liquid propellant is characterized by comprising the following steps:
acquiring a sequence of projection images of a bottle of liquid propellant; acquiring each edge point in each projection image in the projection image sequence; taking the accumulated sum of the abscissa of all edge points contained in each line of each projection image as a judgment index of each line, and obtaining the bottleneck area of each projection image according to the difference of the judgment index of each line between the projection images;
acquiring each corner point in each bottleneck area; taking the average value of the horizontal coordinates of each corner point in each bottleneck area as the area segmentation threshold of each bottleneck area, dividing the corner point of which the horizontal coordinate of each corner point is smaller than the area segmentation threshold into a first area, and dividing the corner point of which the horizontal coordinate of each corner point is larger than the area segmentation threshold into a second area;
obtaining a first chain code and a second chain code of each angular point according to each angular point in each first region and adjacent angular points of each angular point, wherein coordinates of each angular point and the first chain code and the second chain code of each angular point form a feature triple of each angular point; clustering feature triples of all corner points in all the first areas to obtain a plurality of corner point clusters; obtaining the abnormal degree of each corner according to the difference between the feature triples of each corner in each corner cluster; obtaining each suspected damage point according to the abnormal degree of each corner point; similarly, obtaining each suspected damage point in the second area;
obtaining each suspected damage sequence according to the coordinates of each suspected damage point and the serial number of the projection image where each suspected damage point is located; obtaining the damage probability of each suspected damage sequence according to the abscissa of each suspected damage point in each suspected damage sequence; and obtaining each damage position according to the damage probability of each suspected damage sequence.
2. The method for detecting the production quality of a liquid propellant according to claim 1, wherein the method for acquiring the bottleneck region of each projection image comprises:
performing adaptive threshold segmentation on each projection image to obtain a projection area of each projection image; and calculating the variance of the judgment indexes of each line between the judgment indexes corresponding to the projection images, when the variance is 0, judging each line as a bottleneck area, otherwise, judging each line in the projection images in sequence to obtain all line number sequences belonging to the bottleneck area, and taking the line of which the line number belongs to the line number sequence in the projection area of each projection image as the bottleneck area of each projection image.
3. The method for detecting the production quality of a liquid propellant as claimed in claim 1, wherein the expression for obtaining the degree of abnormality of each corner point is as follows:
in the formula (I), the compound is shown in the specification,
the abnormal degree of the kth corner point in the first area of the ith bottleneck area is obtained;
the abscissa of the kth angular point is taken as the abscissa;
a first chain code of the kth angular point;
a second chain code of the kth angular point;
，
and
respectively taking the average value of the abscissa of all the angular points in the angular point cluster to which the kth angular point belongs, the average value of the first chain code and the average value of the second chain code;
is L2 norm;
is an exponential function with a natural constant as the base.
4. The method for detecting the production quality of a liquid propellant according to claim 1, wherein the method for obtaining each suspected damage point according to the abnormal degree of each corner point comprises:
and setting a threshold value according to experience, wherein when the abnormal degree of each corner point is greater than the threshold value, each corner point is a suspected damage point, otherwise, each corner point is not considered as a suspected damage point.
5. The method for detecting the production quality of a liquid propellant according to claim 1, wherein the method for acquiring each suspected damage sequence comprises:
for a suspected damage point which is not included in a suspected damage sequence, acquiring a suspected damage point which is closest to the adjacent projection image in the projection image sequence according to the serial number of the projection image where the suspected damage point is located in the projection image sequence, including the coordinates of the suspected damage point into one suspected damage sequence, and acquiring a suspected damage point which is closest to the newly included suspected damage point in the adjacent projection image until the suspected damage point which is closest to the newly included suspected damage point does not exist in the adjacent projection image; and sequentially processing each suspected damage point to obtain each suspected damage sequence.
6. The method for detecting the production quality of a liquid propellant according to claim 1, wherein the method for detecting the damage probability of each suspected damage sequence comprises:
obtaining a damage curve of each suspected damage sequence according to the serial number of each suspected damage point in each suspected damage sequence and the abscissa of each suspected damage point, and calculating the slope of each suspected damage point in each damage curve; and obtaining an average value of the ratio according to the ratio of the slope of the adjacent suspected damage points in each suspected damage sequence to the absolute value of the slope, and taking the average value as the damage probability of the suspected damage sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211670432.4A CN115661146B (en) | 2022-12-26 | 2022-12-26 | Production quality detection method of liquid propellant |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211670432.4A CN115661146B (en) | 2022-12-26 | 2022-12-26 | Production quality detection method of liquid propellant |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115661146A CN115661146A (en) | 2023-01-31 |
| CN115661146B true CN115661146B (en) | 2023-03-10 |
Family
ID=85023313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211670432.4A Active CN115661146B (en) | 2022-12-26 | 2022-12-26 | Production quality detection method of liquid propellant |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115661146B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07121713A (en) * | 1993-10-21 | 1995-05-12 | Kobe Steel Ltd | Pattern recognition method |
| CN115239701A (en) * | 2022-09-15 | 2022-10-25 | 江苏鑫缘医疗科技有限公司 | Method for detecting foreign matters on surface of medical gauze |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2645375B2 (en) * | 1987-11-06 | 1997-08-25 | 株式会社鷹山 | How to vectorize shapes |
| JPH08327560A (en) * | 1995-05-30 | 1996-12-13 | Sanyo Electric Co Ltd | Device and method for inspecting shape |
| CN115115641B (en) * | 2022-08-30 | 2023-12-22 | 孙清珠 | Pupil image segmentation method |
| CN115311262A (en) * | 2022-10-09 | 2022-11-08 | 南通如东依航电子研发有限公司 | Printed circuit board defect identification method |
| CN115376032B8 (en) * | 2022-10-25 | 2023-05-26 | 金乡县林业保护和发展服务中心(金乡县湿地保护中心、金乡县野生动植物保护中心、金乡县国有白洼林场) | Forestry disease and insect intelligent identification method and system based on image understanding |
| CN115457067B (en) * | 2022-11-09 | 2023-04-07 | 乐比(广州)健康产业有限公司 | Nose-clearing and refreshing medicine liquid level detection method |
-
2022
- 2022-12-26 CN CN202211670432.4A patent/CN115661146B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07121713A (en) * | 1993-10-21 | 1995-05-12 | Kobe Steel Ltd | Pattern recognition method |
| CN115239701A (en) * | 2022-09-15 | 2022-10-25 | 江苏鑫缘医疗科技有限公司 | Method for detecting foreign matters on surface of medical gauze |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115661146A (en) | 2023-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115311292B (en) | Strip steel surface defect detection method and system based on image processing | |
| CN115082419B (en) | Blow-molded luggage production defect detection method | |
| CN107103320B (en) | Embedded medical data image identification and integration method | |
| CN110047109A (en) | A kind of camera calibration plate and its recognition detection method based on self-identifying label | |
| CN105117677A (en) | QR code feature detection method and system | |
| CN113706534B (en) | Injection molding product defect detection method based on multi-ray analysis | |
| CN109242858B (en) | Fabric printing cyclic pattern element segmentation method based on self-adaptive template matching | |
| CN102930279B (en) | For the image-recognizing method that product quantity detects | |
| CN102938062A (en) | Document image slant angle estimation method based on content | |
| CN115311277A (en) | Pit defect identification method for stainless steel product | |
| CN110648330B (en) | Defect detection method for camera glass | |
| CN114863492B (en) | Method and device for repairing low-quality fingerprint image | |
| CN111738271A (en) | Method for identifying shielded fruits in natural environment | |
| CN111242957A (en) | Data processing method and device, computer storage medium and electronic equipment | |
| CN105913067A (en) | Image contour characteristic extraction method and device | |
| CN111833362A (en) | Unstructured road segmentation method and system based on superpixel and region growing | |
| CN115661146B (en) | Production quality detection method of liquid propellant | |
| CN115631128A (en) | Circular medicine quantity detection method based on circular power theorem | |
| CN112381844B (en) | Self-adaptive ORB feature extraction method based on image blocking | |
| CN106897989B (en) | A kind of fingerprint image dividing method calculated based on line density | |
| CN108564020B (en) | Micro-gesture recognition method based on panoramic 3D image | |
| CN114550167B (en) | Artificial intelligence based pear quality classification method and device | |
| CN113902667B (en) | Thread turning identification method and system for machine vision | |
| CN114913152A (en) | Machine vision-based wood cutting surface deburring quality evaluation method and system | |
| CN113011498B (en) | Feature point extraction and matching method, system and medium based on color 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 |