CN107093181B - multiple label detection method for identifying existence of redundant label by using visual system - Google Patents
multiple label detection method for identifying existence of redundant label by using visual system Download PDFInfo
- Publication number
- CN107093181B CN107093181B CN201710160476.5A CN201710160476A CN107093181B CN 107093181 B CN107093181 B CN 107093181B CN 201710160476 A CN201710160476 A CN 201710160476A CN 107093181 B CN107093181 B CN 107093181B
- Authority
- CN
- China
- Prior art keywords
- label
- area
- bits
- tag
- redundant
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/14—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation using light without selection of wavelength, e.g. sensing reflected white light
- G06K7/1404—Methods for optical code recognition
- G06K7/1408—Methods for optical code recognition the method being specifically adapted for the type of code
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Electromagnetism (AREA)
- General Health & Medical Sciences (AREA)
- Toxicology (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
multi-label detection method for recognizing existence of redundant labels by using a vision system comprises the following steps of 1) carrying out image acquisition on a label feeding box by an industrial camera to obtain an image of the label feeding box and respectively extracting a plurality of label bit areas in the label feeding box, 2) screening label bits without labels, 3) screening the label bits with the redundant labels, 4) giving a value 0 to the label bits with the redundant labels and without labels, giving a value 1 to the rest label bits, carrying out binary and decimal numerical calculation on the numerical values of all the label bits in sequence to obtain integrated information of all the label bits in the label feeding box, and 5) transmitting the obtained integrated information of the label boxes to an electric control system which controls a robot to absorb an in-label with a binary value of 1.
Description
Technical Field
The invention belongs to the field of industrial automation, and particularly relates to multi-label detection methods for identifying redundant labels by using a vision system, which can be applied to the industries of assembly, packaging, detection and the like.
Background
At present, automation technology is widely applied to industrial production processes, intelligent production becomes more and more important trends of field operation in the industrial field, labels are generally required to be pasted at fixed positions in the fields of assembly, packaging, detection and the like, the manual operation efficiency of the process is low, the phenomena of wrong pasting, missing pasting and the like are easy to occur, the problems of label feeding and detection need to be solved firstly when the process is automatically modified, label detection robots applied in the industry at present are mostly used for detecting whether a single label is defective or not and whether the label is defective or not, and the simultaneous detection of multiple labels with higher efficiency and the detection of the label with a redundant label are not disclosed in related technologies.
Disclosure of Invention
In order to overcome the defects in the background art, the invention provides multi-label detection methods which can accurately acquire the label existence information of a plurality of label positions and eliminate the label positions with redundant labels and identify the redundant labels by using a vision system.
The multi-label detection method for identifying redundant labels by using a vision system mainly comprises two parts, namely multi-label vision detection and multi-label bit information integration.
(1) Wherein, the multi-label visual detection comprises the following steps:
step 1) obtaining an image of a label feeding box, and extracting a plurality of label position areas respectively;
step 2) processing each label position area in sequence, and eliminating label positions without labels, wherein the specific steps are as follows:
step 2.1) carrying out threshold segmentation on each area, and adopting different thresholds for different areas to finally obtain an interested suspected label area;
step 2.2) removing noise interference of the suspected label area by a corrosion operation method to obtain an interference-free area;
step 2.3) carrying out regional characteristic screening on the corroded region, and respectively adopting characteristics such as area, central point coordinates and the like to screen and remove label positions without labels;
step 3) processing the label positions with labels in sequence, and rejecting the label positions with redundant labels, wherein the specific steps are as follows:
step 3.1) performing expansion treatment on the area screened in the step 2.3) to obtain an area containing the label;
step 3.2) carrying out sub-pixel edge extraction on the region of each label position containing the label to obtain the label edge;
step 3.3) connecting the label edges in each area obtained in the step 3.2) respectively to obtain complete label edges;
and 3.4) carrying out feature screening on the label edges, and respectively adopting feature screening such as edge closure, edge length and the like to reject label positions with redundant labels.
(2) The multi-tag bit information integration comprises the following steps:
assigning a number 0 to the label-free label bit obtained in step 2.3) of the multi-label visual detection method; assigning a number 0 to the label bit of the redundant label obtained in the step 3.4) of the multi-label visual detection method;
assigning a number 1 to the tag bits after no tag and redundant tags are screened;
and sequentially carrying out binary and decimal numerical calculation on the label digit numbers to obtain label integrated information of the whole label feeding box.
The invention has the beneficial effects that:
1. the method adopted by the invention simultaneously detects a plurality of labels through sets of vision systems, improves the detection efficiency of the robot, and simultaneously performs distributed elimination on the label positions with redundant labels, thereby enhancing the robustness of the label detection system.
2. The invention adopts a mode of combining two screening objects and various characteristic screening methods, can effectively eliminate the label positions without labels and with redundant labels, and accurately obtains the label information of a plurality of label positions.
3. The invention integrates information of a plurality of label bits by adopting a binary and decimal value combined calculation mode, thereby improving the reliability and convenience of the information.
Drawings
FIG. 1 is a flow chart of the present invention patent;
FIG. 2 is an image taken of 12 tag bits;
FIG. 3 is a view of the th tagbit suspect label in the right column of FIG. 2;
FIG. 4 is a view of a suspected label area after an etching process;
FIG. 5 is a view after an expansion process has been performed on the eroded region;
FIG. 6 is a partial image view of a sub-pixel edge within an expansion region.
Detailed Description
The invention provides visual label detection methods capable of quickly and accurately acquiring the label existence information of a plurality of label positions and eliminating redundant label positions.
When the invention is adopted, an industrial camera is adopted as image acquisition equipment, an industrial personal computer is adopted as communication equipment of image processing and control system, a PLC electrical control system is adopted as a communication center between the robot and the industrial personal computer, and the robot is adopted as execution equipment.
During operation, the industrial camera gathers images, the industrial personal computer utilizes image processing technology to obtain label information and carries out information integration, information after will integrating sends PLC for, PLC judges this information, converts accurate information into robot execution command and transmits for the robot, and the robot carries out the label to the label position that has the correct label in the label feed box and absorbs and carries out automation mechanized operations down.
The specific method of the invention is as follows:
step 1) an industrial camera acquires images of a label box, acquires images of a label feeding box, and extracts a plurality of label position areas respectively;
step 2) processing each label position area in sequence, and eliminating label positions without labels;
the specific flow of the step is as follows:
step 2.1) performing threshold segmentation on each area, and adopting different thresholds for different areas to finally obtain an interested suspected label area, for example, the label feeding box in fig. 2 has 12 label positions, the left column and the right column have 6 label positions respectively, 10 labels are placed in the left column and the right column, so that the left column can see the 6 normal label positions, the right column has 3 label positions with redundant labels and 3 label positions without labels, and the suspected label area obtained by performing threshold segmentation on the th label position in the right column is shown in a black area in fig. 3;
step 2.2) removing noise interference of the suspected label area by a corrosion operation method to obtain an interference-free area, wherein an area obtained by performing corrosion operation on the th label position suspected label area in the right column of fig. 3 is a black area in fig. 4, and the specific implementation of the corrosion operation is defined as formula 1:
in the formula:
A1-the suspected label area obtained in step 2.1);
B1-structural element B1Is a pair A1A structure to be etched;
A1ΘB1——A1quilt B1Aggregate area after etching, if p1Is A1ΘB1A point in (B), it satisfies when the structural element B1Shift to p1After point, should be contained in A1And (4) the following steps.
Step 2.3) carrying out regional characteristic screening on the corroded region, respectively adopting characteristics such as area and central point coordinates to screen and remove label positions without labels, comparing the regional area and the central point characteristics with a set value in sequence, and removing the characteristics without the set value range, wherein the calculation processes of the regional area and the central point are respectively shown in formula 2 and formula 3:
A＝∫∫dxdy………………………………(2)
in the formula:
a-area of the region;
dx, dy-the line segment differential formed by the current calculation point in the x direction and the y direction, and the area differential is obtained by multiplying;
in the formula:
x0，y0-coordinates of centroid inside the region;
Sy，Sx-static moments of the internal section of the zone to the y and x axes, respectively;
a-area of the region;
x, y-the coordinates of the current calculation point in the image coordinate system;
dx, dy-the line segment differential formed by the current calculation point in the x direction and the y direction, and the area differential is obtained by multiplying;
and 2.4) endowing the screened label-free label bits with a value of 0.
Step 3) processing the label positions with labels in sequence, and rejecting the label positions with redundant labels;
the specific flow of the step is as follows:
step 3.1) performing expansion processing on the area screened in the step 2.3) to obtain an area containing a label, for example, performing expansion processing on the area subjected to corrosion in fig. 4 to obtain a black area as shown in fig. 5, wherein the expansion processing is specifically defined as formula 4:
in the formula:
A2-the screened areas obtained in step 2.3);
B2-structural element B2Is a pair A2A structure that undergoes expansion;
Z2-a two-dimensional integer space;
Step 3.2) performing sub-pixel edge extraction on the region of each label position containing the label to obtain a label edge, wherein the sub-pixel edge extraction is to subdivide the pixel level edge, extract the sub-pixel edge by adopting a method of interpolating the gray value of a pixel point, and perform the sub-pixel edge extraction on the expansion region obtained in the figure 5 to obtain all edges of the region which are amplified to be like black edges in the figure 6;
step 3.3) connecting the label edges in each area obtained in the step 3.2) respectively to obtain complete label edges;
step 3.4) feature screening is carried out on the label edges, label positions with redundant labels are respectively screened and removed by adopting edge closing and edge length feature screening, the obtained feature values are compared with a set value, and the label positions which are not in the range of the set value are removed, wherein the calculation processes of the edge closing features and the edge length features are respectively shown in formula 5 and formula 6:
in the formula:
delta L is the pixel difference between the edge starting point and the final point, the pixel difference is compared with a set threshold value, and if the pixel difference is smaller than the threshold value, the pixel difference is closed;
xend，yend-edge final point image coordinates;
x0，y0-edge start point image coordinates;
L＝∫ds………………………………(6)
in the formula:
l is the edge length;
ds — differential edge length.
And 3.5) endowing the screened tag bits with the redundant tags with a value of 0.
And 4) endowing the screened residual label bits with a value 1.
Step 5) carrying out binary and decimal numerical calculation on the numerical value of each label bit in sequence to obtain integrated information of all label bits in the label feeding box;
the specific flow of the step is as follows:
step 5.1) carrying out decimal value summarization on the n label bit information, wherein the summarization process of the label bit information in the step is mainly completed by the following formula 7:
in the formula:
resulti-calculating summary information after the ith tag bit;
i-total n label bits, wherein the i is the ith label bit from 1,2, … … n;
ai-assignment of the ith tag bit;
resultn-calculating the summary information after the nth tag bit, i.e. all tag bit information.
Step 5.2) decimal number result of the final summary information obtained in the step 5.1nThe conversion into binary values, the decimal to binary calculation process is as follows:
in the formula:
resultn-5.1) calculating summary information after n label bits;
i-total n label bits, wherein the i is the ith label bit from 1,2, … … n;
xithe ith pair yi-1A value obtained by dividing by 2;
yito xiTaking the value obtained by rounding;
bi——yi-1the remainder obtained by dividing by 2;
the whole calculation process is just through the resultnDividing by 2 to obtain remainder, and obtaining n-bit binary value Result of data integrationn＝bnbn-1bn-2…b2b1From the low positionThe integrated information of all label bits in the label feeding box is obtained from the 1 st to the n th in sequence from the high order, for example, the label information of 12 label bits shown in fig. 2 is summarized, the sequence of the corresponding label bits from the low order to the high order is set from right 1 to right 6 to left 1 to left 6, and the obtained label integrated information is 000000111111, namely six normal label bits and 6 abnormal label bits.
And 6) transmitting the acquired label box integrated information to an electric control system, controlling the robot to absorb the label position inner label with the binary value of 1 by the electric control system, and continuing to carry out automatic operation.
Claims (7)
1, A method for multi-tag detection using a vision system to identify the presence of redundant tags, comprising the steps of:
1) the industrial camera acquires images of the label feeding box, and extracts a plurality of label position areas in the label feeding box respectively;
2) screening label bits without labels;
3) screening the tag bits with redundant tags;
4) assigning a value 0 to the tag bits with redundant tags and no tags, and assigning a value 1 to the rest tag bits; carrying out binary and decimal numerical calculation on the numerical value of each label bit in sequence to obtain integrated information of all label bits in the label feeding box;
the specific process is as follows:
4.1) carrying out decimal numerical value summarization on the information of the n label bits, wherein the formula is as follows:
in the formula:
resulti-calculating summary information after the ith tag bit;
i-total n label bits which are respectively 1,2 and … … n, wherein i is the ith label bit;
ai-assignment of the ith tag bit;
resultn-calculating summary information after the nth tag bit, i.e. all tag bit information;
4.2) final summary information result obtained in the step 4.1)nThe formula for conversion from decimal to binary is:
in the formula:
resultn4.1) calculating summary information after n label bits;
i-total n label bits, wherein the i is the ith label bit from 1,2, … … n;
xithe ith pair yi-1A value obtained by dividing by 2;
yito xiTaking the value obtained by rounding;
bi——yi-1the remainder obtained by dividing by 2;
the whole calculation process is just through the resultnDividing by 2 to obtain remainder, and obtaining n-bit binary value Result of data integrationn＝bnbn-1bn-2…b2b1Sequentially integrating information of 1 st to n th label positions in the label feeding box from a low position to a high position;
5) and transmitting the acquired tag box integrated information to an electrical control system, and controlling the robot to absorb the tag in the tag position with the binary value of 1 by the electrical control system.
2. The multi-tag detection method for recognizing the existence of redundant tags by using a vision system as claimed in claim 1, wherein said step 2) is specifically characterized by:
2.1) carrying out threshold segmentation on each label bit area, and adopting different thresholds for different areas to obtain suspected label areas;
2.2) removing noise interference of the suspected label area by a corrosion operation method to obtain an interference-free area;
2.3) carrying out regional characteristic screening on the corroded region, and respectively screening and eliminating label positions without labels by adopting the area characteristics and the center point coordinates;
2.4) assigning a value of 0 to the selected non-labeled tag bits.
3. method for detecting multiple tags by using visual system to identify the existence of redundant tags, as claimed in claim 2, wherein said etching operation of step 2.2) is specifically:
in the formula:
A1-the suspected label area obtained in step 2.1);
B1-structural element B1Is a pair A1A structure to be etched;
A1ΘB1——A1quilt B1Aggregate area after etching, if p1Is A1ΘB1A point in (B), it satisfies when the structural element B1Shift to p1After point, should be contained in A1And (4) the following steps.
4. The method for detecting multiple tags, according to claim 2, wherein said step 2.3) of area calculation is:
the area calculation process is as follows:
A＝∫∫dxdy
in the formula:
a-area of the region;
dx, dy-the line segment differential formed by the current calculation point in the x direction and the y direction, and the area differential is obtained by multiplying;
the step 2.3) of calculating the central point comprises the following steps:
in the formula:
x0，y0-coordinates of centroid inside the region;
Sy，Sx-static moments of the internal section of the zone to the y and x axes, respectively;
x, y-the coordinates of the current calculated point in the image coordinate system.
5. The multi-tag detection method for recognizing the existence of redundant tags by using a vision system as claimed in claim 1, wherein said step 3) is specifically characterized by:
3.1) carrying out expansion treatment on the area screened in the step 2.3) to obtain an area containing the label;
3.2) carrying out sub-pixel edge extraction on the region of each label position containing the label to obtain the label edge;
3.3) respectively connecting the label edges in each area obtained in the step 3.2) to obtain complete label edges;
and 3.4) carrying out feature screening on the label edges, and respectively adopting edge closing features and edge length features to screen and reject label positions with redundant labels.
6. The method of claim 5, wherein the step of identifying the presence of redundant tags using a vision system comprises: the expansion treatment in the step 3.1) is specifically as follows:
in the formula:
A2-the screened areas obtained in step 2.3);
B2-structural element B2Is a pair A2A structure that undergoes expansion;
Z2-a two-dimensional integer space;
7. The multiple label detection method using visual system to identify the existence of redundant label according to claim 5, wherein the edge closing feature calculation process in step 3.4) is:
in the formula:
delta L is the pixel difference between the edge starting point and the final point, the pixel difference is compared with a set threshold value, and if the pixel difference is smaller than the threshold value, the pixel difference is closed;
xend，yend-edge final point image coordinates;
x0，y0-edge start point image coordinates;
the edge length feature calculation process in the step 3.4) comprises the following steps:
L＝∫ds
in the formula:
l is the edge length;
ds — differential edge length.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2016112608869 | 2016-12-30 | ||
| CN201611260886 | 2016-12-30 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107093181A CN107093181A (en) | 2017-08-25 |
| CN107093181B true CN107093181B (en) | 2020-01-31 |
Family
ID=59646217
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710160476.5A Active CN107093181B (en) | 2016-12-30 | 2017-03-17 | multiple label detection method for identifying existence of redundant label by using visual system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107093181B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109598724B (en) * | 2019-02-01 | 2023-04-07 | 中珂湳海(佛山)智能科技有限公司 | Cloth label detection and identification method |
| CN111985480B (en) * | 2020-06-29 | 2022-04-15 | 联宝(合肥)电子科技有限公司 | Multi-label detection method and device and storage medium |
| CN112506499B (en) * | 2020-12-04 | 2022-09-30 | 合肥工业大学 | Method for automatically arranging measurement software tags |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894268A (en) * | 2010-07-16 | 2010-11-24 | 西安理工大学 | Seal authenticity identification method capable of eliminating receipt interference |
| CN102663726A (en) * | 2012-03-06 | 2012-09-12 | 中国矿业大学(北京) | Method and device for material segmentation of material accumulation images |
| CN104021546A (en) * | 2014-05-14 | 2014-09-03 | 北京理工大学 | On-line rapid workpiece label positioning method based on image processing |
| CN105741308A (en) * | 2016-03-07 | 2016-07-06 | 英华达（上海）科技有限公司 | Method and system for obtaining multidimensional visual image |
-
2017
- 2017-03-17 CN CN201710160476.5A patent/CN107093181B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894268A (en) * | 2010-07-16 | 2010-11-24 | 西安理工大学 | Seal authenticity identification method capable of eliminating receipt interference |
| CN102663726A (en) * | 2012-03-06 | 2012-09-12 | 中国矿业大学(北京) | Method and device for material segmentation of material accumulation images |
| CN104021546A (en) * | 2014-05-14 | 2014-09-03 | 北京理工大学 | On-line rapid workpiece label positioning method based on image processing |
| CN105741308A (en) * | 2016-03-07 | 2016-07-06 | 英华达（上海）科技有限公司 | Method and system for obtaining multidimensional visual image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107093181A (en) | 2017-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108305243B (en) | Magnetic shoe surface defect detection method based on deep learning | |
| EP3176751B1 (en) | Information processing device, information processing method, computer-readable recording medium, and inspection system | |
| CN116703907B (en) | Machine vision-based method for detecting surface defects of automobile castings | |
| CN111257341B (en) | Underwater building crack detection method based on multi-scale features and stacked full convolution network | |
| CN107093181B (en) | multiple label detection method for identifying existence of redundant label by using visual system | |
| CN106709524B (en) | Element symbol detection and identification method in electrical engineering drawing | |
| CN115880248B (en) | Surface scratch defect identification method and visual detection equipment | |
| CN115063430B (en) | Electric pipeline crack detection method based on image processing | |
| CN113516619B (en) | Product surface flaw identification method based on image processing technology | |
| CN114581782A (en) | Fine defect detection method based on coarse-to-fine detection strategy | |
| CN114820626B (en) | Intelligent detection method for automobile front face part configuration | |
| CN114255212A (en) | FPC surface defect detection method and system based on CNN | |
| CN115775246A (en) | Method for detecting defects of PCB (printed circuit board) components | |
| CN107886093B (en) | Character detection method, system, equipment and computer storage medium | |
| CN108537778B (en) | Improved random round hole detection method for flexible substrate | |
| CN106778844B (en) | Method and system for matching cracks in tunnel | |
| CN107710229B (en) | Method, device and equipment for recognizing shape in image and computer storage medium | |
| CN113177922A (en) | Raw material foreign matter identification method, system, medium and electronic terminal | |
| CN111291712B (en) | Forest fire recognition method and device based on interpolation CN and capsule network | |
| CN115511884B (en) | Punching compound die surface quality detection method based on computer vision | |
| CN115690104B (en) | Wafer crack detection method and device and storage medium | |
| CN109584212B (en) | MATLAB-based SLM powder bed powder laying image scratch defect identification method | |
| CN115619799B (en) | Grain image segmentation method and system based on transfer learning | |
| CN115439446A (en) | Appearance defect detection method and device, storage medium and electronic equipment | |
| JP5993100B2 (en) | Image processing apparatus and specific figure detection method |
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 |