CN114613098B - Tray stacking out-of-range detection method - Google Patents
Tray stacking out-of-range detection method Download PDFInfo
- Publication number
- CN114613098B CN114613098B CN202111565026.7A CN202111565026A CN114613098B CN 114613098 B CN114613098 B CN 114613098B CN 202111565026 A CN202111565026 A CN 202111565026A CN 114613098 B CN114613098 B CN 114613098B
- Authority
- CN
- China
- Prior art keywords
- tray
- classifier
- video
- frame
- range
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 17
- 230000003068 static effect Effects 0.000 claims abstract description 30
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000012544 monitoring process Methods 0.000 claims abstract description 14
- 230000008569 process Effects 0.000 claims abstract description 14
- 238000001914 filtration Methods 0.000 claims abstract description 12
- 230000000877 morphologic effect Effects 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000004140 cleaning Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 3
- 238000002955 isolation Methods 0.000 abstract description 3
- 230000004044 response Effects 0.000 abstract description 3
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000012804 iterative process Methods 0.000 description 3
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/0202—Child monitoring systems using a transmitter-receiver system carried by the parent and the child
- G08B21/0233—System arrangements with pre-alarms, e.g. when a first distance is exceeded
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/0202—Child monitoring systems using a transmitter-receiver system carried by the parent and the child
- G08B21/0236—Threshold setting
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/0202—Child monitoring systems using a transmitter-receiver system carried by the parent and the child
- G08B21/0261—System arrangements wherein the object is to detect trespassing over a fixed physical boundary, e.g. the end of a garden
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
-
- 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
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Child & Adolescent Psychology (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a tray stacking out-of-range detection method, which comprises the steps of constructing and training a self-adaptive lifting tree cascade classifier, wherein the self-adaptive lifting tree cascade classifier is a strong classifier formed after a group of weak classifiers are weighted, and the tray classifier is used for carrying out real-time positioning, monitoring and tracking on trays in a goods yard and filtering the trays which are placed in isolation and disorder in the goods yard; the pallet classifier is utilized to position, detect and track the pallet in real time, and alarm once when the pallet is in a static state and the time length exceeding the warning line reaches the trigger preset time, and only alarm once when the static state is not changed. The invention realizes the potential safety hazard caused by whether the automatic supervision tray crosses the guard line in the stacking process, reduces the risk of untimely manual supervision, improves the operation safety, and realizes the safe and efficient response in the operation process of the goods yard.
Description
Technical Field
The invention relates to the technical field of monitoring of goods yards, in particular to a tray stacking out-of-range detection method.
Background
In recent years, railway freight yards frequently suffer safety accidents, personnel, vehicles, cargoes, operation processes and the like in the freight yards are mainly controlled by means of personal air defense at present, the informatization and intellectualization degree is low, and the supervision strength and the supervision efficiency are required to be improved. Although the traditional video monitoring means are deployed at present, the advantages of the video are not fully exerted, and the informatization construction needs to be further optimized and improved.
For example, in chinese patent CN111311630a, publication date 2020, 6 and 19, a method for counting the number of goods through video intelligence in warehouse management, based on video monitoring, the SSD algorithm and KLT algorithm are adopted to identify and track the goods flowing through the warehouse entrance and exit area, so as to detect the variation condition of the number of goods in warehouse, and make the statistics of goods more convenient and faster. Specifically, monitoring videos of warehouse entrance and exit areas are collected through cameras. And then, acquiring the current frame image alpha and the previous frame image beta of the monitoring video, and calculating the difference ratio between the frame image alpha and the frame image beta. And when the difference proportion exceeds a preset threshold, detecting the cargo quantity of the multi-frame pictures after the frame pictures alpha through an SSD algorithm and a KLT algorithm. And finally, counting the quantity of goods coming in and going out of the warehouse, and transmitting the counted quantity to a visual terminal in real time for a manager to check and judge. The method mainly detects the quantity of the cargoes according to the monitoring video, but the warehouse management of the cargoes not only involves quantity statistics, but also has corresponding limiting measures for stacking the stacked cargoes, and the condition of personal safety accidents caused by collapse due to error border crossing when the cargoes are stacked is avoided.
Disclosure of Invention
The invention aims to solve the technical problems that: the existing warehouse management video monitoring method lacks the technical problem of personal safety accidents caused by collapse of goods stacks because of detecting whether goods are stacked out of range or not. The tray stacking boundary crossing detection method can monitor the stacking boundary crossing condition of the goods and timely early warn, and avoid personal safety accidents caused by collapse of the goods.
In order to solve the technical problems, the invention adopts the following technical scheme: a tray stacking out-of-range detection method comprises the following steps:
s1, constructing and training a self-adaptive lifting tree cascade classifier;
s2, filtering an isolated target tray in the monitoring video;
and S3, detecting and analyzing the tray crossing in the video, and giving an alarm when the trigger condition is met. A tray border crossing alarm method based on a self-adaptive lifting tree cascade classifier comprises the steps of building the self-adaptive lifting tree cascade classifier, selecting and training positive and negative samples, and filtering an isolated target tray and controlling alarm logic. The adaptive lifting tree cascade classifier is a strong classifier formed after a set of weak classifiers are weighted. The positive and negative samples manufactured in advance change the distribution of training samples adaptively in the iterative process, so that the weak classifier focuses on samples which are difficult to separate; real-time positioning, monitoring and tracking are carried out on trays in a goods yard by using a trained tray classifier, and filtering is carried out on the trays which are placed in isolation and disorder in the goods yard; and (3) carrying out real-time positioning detection tracking on the pallet in the goods yard by using the trained pallet classifier, and alarming once when the pallet is in a static state and the time length exceeding the warning line reaches the trigger preset time, wherein the alarm is carried out once when the static state is not changed. The tray can be crossed by a fork truck, a person and the like in the moving process in normal work and can select whether to alarm according to the state of the dynamic timer. The automatic supervision tray has the advantages that potential safety hazards are caused by whether the automatic supervision tray crosses a warning line in the stacking process, the risk of untimely manual supervision is reduced, the operation safety is improved, and safe and efficient response in the operation process of a goods yard is realized.
Preferably, the filtering process of the isolated target tray in step S2 includes the steps of:
a1, starting an algorithm;
a2, pushing a frame of video stream to a tray positioning algorithm;
a3, carrying out pallet target identification by utilizing an adaptive lifting tree cascade classifier;
a4, creating a pure black picture with the same size as the original video frame picture, marking the identified center point coordinates of each tray on the pure black picture, and setting the pixel value of the center point position;
a5, carrying out morphological opening operation by using the structural unit;
a6, setting a convolution kernel, and carrying out convolution operation on the marked pure black graph;
a7, performing threshold binarization on the marked pure black image after convolution operation, and taking the binarized image as a mask;
and A8, carrying out union operation on the original image of the video and the mask to obtain RIO, and cleaning all the isolated tray positions. The center point position pixel value may be set to 255.
Preferably, the trigger detection process of the alarm in step S3 includes the steps of:
b1, setting an alarm state after the algorithm is started and setting the alarm state as false, setting a static timer as zero and setting a dynamic timer as zero;
b2, sending video streams of a frame and a current frame before a section of interval a into a tray reasoning algorithm for detecting the crossing tray;
if the current frame does not detect the out-of-range tray, returning to the step B2, and continuing to wait for the video stream of the next frame to continue to be sent into a tray reasoning algorithm;
if the current frame detects that the out-of-range tray exists, continuously detecting whether the out-of-range tray of the current frame and the out-of-range tray of the previous frame of the current frame move or not;
b5, if the out-of-range tray is detected to move, starting a dynamic timer to start timing, and judging whether the dynamic timer reaches the preset dynamic trigger time or not; if the triggering time is reached, immediately alarming, and resetting the dynamic timer, wherein the alarming state is false; if the dynamic trigger time is not reached, returning to the step B2, and continuing to wait for the video stream of the next frame to continue to be sent into the tray reasoning algorithm;
b6, if the tray is detected to be not moved, the dynamic timer is cleared, and whether the alarm state is false is checked; if the alarm state is false, starting a static timer and judging whether the static timer reaches the static trigger time or not; if the static trigger time is reached, alarming, setting an alarm state as true, and resetting a static timer; and if the static trigger time is not reached, not alarming, returning to the step B2, and continuing to wait for the video stream of the next frame to continue to be sent into the tray reasoning algorithm. A frame with a distance a from the current frame and the current frame stream are firstly sent into a tray reasoning algorithm for comparison analysis, and a can be set and adjusted according to the video length. Dynamic triggering time and static triggering time such as 5 minutes and 10 minutes can be preset, and the alarm can be given when the corresponding triggering time is reached, and different alarm modes such as different alarm voices can be set to distinguish which timer triggers the alarm.
Preferably, the method for detecting the out-of-range tray in the step B2 comprises the following steps: dividing two intersecting straight lines as warning lines, and calibrating through three points on the warning lines: firstly, calculating one straight line through two points, wherein the calculation equation is as follows:
after transformation, the method comprises the following steps:
the other straight line is calculated as:
wherein (x) 1 ，y 1 )，(x 2 ，y 2 )，(x 3 ，y 3 ) Three points known on the guard line, respectively; detecting the position coordinates (x, y) of each tray by a tray classifier, and respectively bringing x into the calculation equation of two straight lines to obtain a corresponding value y p1 And y p2 If y satisfies the condition y>y p1 And condition y>y p1 At least one of the two judgment conditions judges that the tray is out of range. And determining the tray with the coordinate y meeting the judgment condition as an out-of-range tray.
Preferably, in step A5, morphological open operations are performed using the building blocks 3*3 or 5*5. Morphological opening operations are typically performed on the annotated, purely black plots using building blocks of 3*3 or 5*5 size.
Preferably, in step A6, the convolution kernel size is set to be between 15×15 and 20×20, and each kernel element size is 1 divided by the convolution kernel length-width product. The convolution kernel size may be set to be a certain determined size between 15×15 and 20×20, and the size of each kernel element is 1 divided by the convolution kernel length-width product, and the convolution operation is preferably performed on the marked pure black graph by using the convolution, where the threshold value of the binarization in the step A7 is selected to be between 200 and 230. The binarization rule is that a threshold value is set between 200 and 230, a value greater than the threshold value is set to 255, and a value not greater than the threshold value is set to 0.
The invention has the following substantial effects: the method comprises the steps of constructing and training a self-adaptive lifting tree cascade classifier, filtering an isolated target tray and controlling alarm logic, wherein the self-adaptive lifting tree cascade classifier is a strong classifier formed after a group of weak classifiers are weighted. The positive and negative samples manufactured in advance change the distribution of training samples adaptively in the iterative process, so that the weak classifier focuses on samples which are difficult to separate; carrying out real-time positioning, monitoring and tracking on trays in a goods yard by utilizing a tray classifier, and filtering the trays which are placed in isolation and disorder in the goods yard; the pallet classifier is utilized to perform real-time positioning detection tracking on the pallet in the goods yard, when the pallet is monitored to be in a static state and the time length exceeding the warning line reaches the trigger preset time, the alarm is given once only when the static state is not changed, and whether the pallet is in the normal working state or not can be selected according to the state of the dynamic timer when the pallet crosses the warning line in the moving process of a forklift, a person and the like. The automatic supervision tray has the advantages that potential safety hazards are caused by whether the automatic supervision tray crosses a warning line in the stacking process, the risk of untimely manual supervision is reduced, the operation safety is improved, and safe and efficient response in the operation process of a goods yard is realized.
Drawings
FIG. 1 is an iterative schematic of a cascade separator of the present embodiment;
FIG. 2 is a schematic diagram of an isolated tray filtration process according to the present embodiment;
fig. 3 is a schematic diagram of a tray border crossing alarm logic flow in this embodiment.
Detailed Description
The following description of the embodiments of the present invention will be made with reference to the accompanying drawings.
A tray stacking out-of-range detection method comprises the following steps:
s1, constructing and training a self-adaptive lifting tree cascade classifier;
s2, filtering an isolated target tray in the monitoring video;
and S3, detecting and analyzing the tray crossing in the video, and giving an alarm when the trigger condition is met.
The adaptive lifting tree cascade classifier is a strong classifier formed after a set of weak classifiers are weighted. The positive and negative samples manufactured in advance change the distribution of training samples adaptively in the iterative process, so that the weak classifier focuses on samples which are difficult to separate.
The method comprises the following steps:
(1-1) given a training sample (x 1, y 1),., (xi, yi), (xn, yn), where xi represents the i-th sample, yi=0 is represented as a negative sample, and yi=1 is represented as a positive sample. n is the total number of training samples;
(1-2) initializing weights of training samples;
(1-3) first iterating, firstly training a weak classifier, and calculating the error rate of the weak classifier; selecting a proper threshold value to minimize the error; updating the sample weight;
and (1-4) after T times of circulation, obtaining T weak classifiers, and carrying out weighted superposition according to the weight for evaluating the importance of each weak classifier to finally obtain the strong classifier.
The algorithm schematic diagram of the adaptive lifting tree cascade classifier is shown in fig. 1, and initially, a series of weak classifiers are given, initial weights of each weak classifier are distributed by AdaBoost, the AdaBoost distributes a weight alpha for each classifier, and an alpha value is calculated based on the error rate of each weak classifier. The initial weights for each sample are the same relative to the same weak classifier. The correct weight for sample classification is reduced after one round of iterative training, and the weight for samples with wrong classification is increased. And after T rounds of iterative loop training, T weak classifiers are obtained, weighted superposition is carried out according to the weight for evaluating the importance of each weak classifier, and finally a strong classifier is obtained.
In the embodiment, two adjacent edges of a cargo pile in a cargo yard monitoring video are divided into warning lines, the cargo being carried is used as an isolated tray, the cargoes which are piled on the ground and exceed the warning lines are filtered out during operation, and the cargoes are regarded as out-of-range trays.
An isolated tray filtering method based on an adaptive lifting tree cascade classifier is shown in figure 2,
(2-1) starting an algorithm;
(2-2) pushing a frame of video stream to a tray positioning algorithm;
(2-3) carrying out pallet target identification by using a trained self-adaptive lifting tree cascade classifier;
(2-4) creating a pure black picture with the same size as the original video frame picture, marking the identified coordinates of the central point of each tray on the pure black picture, and setting the pixel value of the central point position to 255;
(2-5) then performing morphological opening operations using the building blocks of 3*3 or 5*5;
(2-6) setting convolution kernels, wherein the size of each kernel element is 1 divided by the convolution kernel length-width product, and carrying out convolution operation on the marked pure black graph by using the convolution, wherein the size of each kernel element is set to be a certain determined size between 15 and 20 and 15;
(2-7) carrying out threshold binarization on the label graph after convolution operation, wherein the binarization rule is that a threshold value (generally between 200 and 230) is set, the label graph is 255 which is larger than the threshold value and 0 which is not larger than the threshold value, and the binarized picture is taken as a mask;
and (2-8) carrying out union operation on the original image and the mask generated in the step (2-7) to obtain RIO, and cleaning all the isolated tray positions.
The trained tray classifier is utilized to position, monitor and track trays in a cargo yard in real time, the trays which are isolated and scattered in the cargo yard need to be filtered, and only the integrally stacked trays are monitored. Generally, traversing according to the tray position information is time-consuming for judging the position of each tray relative to other trays, and the accuracy cannot be guaranteed. The central point coordinates of each tray are marked on a pure black picture with the same size as the original video frame picture, the pixel value of the central point position is set to 255, and then morphological open operation is carried out by using a structural unit. And setting convolution kernels, wherein the size of each kernel element is 1/convolution kernel length-width product, and carrying out convolution operation on the marked pure black graph. And performing threshold binarization on the label graph after convolution operation, wherein the binarization rule is that a threshold value is set, 255 is set when the threshold value is larger than the threshold value, and 0 is set when the threshold value is not larger than the threshold value. After the treatment method, the isolated tray positions are all cleaned.
The tray border crossing alarm method based on the self-adaptive lifting tree cascade classifier is shown in figure 3,
(3-1) setting an alarm state after the algorithm is started and setting false, setting a static timer to be zero, and setting a dynamic timer to be zero;
(3-2) inputting video streams of a frame and a current frame before an interval into a tray algorithm inference;
(3-3) if the current frame does not detect the out-of-range tray, returning to (3-2) and continuing to wait for the video stream of the next frame to continue to be fed into the tray reasoning algorithm;
(3-4) if the out-of-range tray is detected, continuing to detect whether the out-of-range tray of the current frame and the out-of-range tray of the previous frame move or not;
(3-5) if the tray moves, starting the dynamic timer to start timing, judging whether the dynamic timer reaches the preset dynamic trigger time, and selecting any time for practical use;
(3-6) if the dynamic trigger time is not reached, returning to (3-2) to continue waiting for the video stream of the next frame to continue to be sent into the tray reasoning algorithm;
(3-7) immediately alarming if the triggering time is reached, and resetting the dynamic timer to zero, wherein the alarm state position false;
(3-8) in the step (3-4), if the tray does not move, the dynamic timer is cleared, and whether the alarm state is false is checked;
(3-9) if the alarm state is false, starting a static timer and judging whether the static timer reaches the static trigger time or not, and selecting any time for practical use;
(3-10) alarming if the static trigger time is reached, and setting an alarm state as true, and resetting a static timer;
and (3-11) if the static trigger time is not reached, not alarming, returning to the step (3-2) to continuously wait for the video stream of the next frame to be continuously fed into the tray reasoning algorithm.
And (3) carrying out real-time positioning detection tracking on the pallet in the goods yard by using the trained pallet classifier, and alarming once when the pallet is in a static state and the time length exceeding the warning line reaches the trigger preset time, wherein the alarm is carried out once when the static state is not changed. The rear tray can select whether to alarm or not according to the state of the dynamic timer when the rear tray is crossed by a warning line in the moving process of a forklift, a person and the like in normal work.
The warning line is two intersecting straight lines which are defined in advance, and is calibrated through three points. The equation for one of the lines can be calculated by two points:
after transformation, the method comprises the following steps:
another linear equation is the same:
wherein (x) 1 ，y 1 )，(x 2 ，y 2 )，(x 3 ，y 3 ) Three points are known on the guard line, respectively.
Detecting the position coordinates (x, y) of each tray by a tray classifier, and respectively bringing x into equations of two straight lines to obtain a corresponding value y p1 And y p2 . If y>y p1 And y>y p2 At least one of the above is satisfied, it is determined that a tray has been out of range.
The foregoing examples illustrate only a few embodiments of the invention, which are described in detail and are not to be construed as limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention.
Claims (5)
1. The tray stacking out-of-range detection method is characterized by comprising the following steps of:
s1, constructing and training a self-adaptive lifting tree cascade classifier;
s2, filtering an isolated target tray in the monitoring video;
s3, detecting and analyzing the tray crossing in the video, and alarming when the trigger condition is met;
the step S2 comprises the following steps: carrying out pallet target identification by utilizing a self-adaptive lifting tree cascade classifier; creating a pure black picture with the same size as the original video frame picture; performing convolution operation and threshold binarization on the marked pure black image; carrying out union operation on the original video image and the binarized image to obtain RIO, and cleaning all the isolated tray positions;
the triggering detection process of the alarm in the step S3 comprises the following steps:
b1, setting an alarm state after the algorithm is started and setting the alarm state as false, setting a static timer as zero and setting a dynamic timer as zero;
b2, sending video streams of a frame and a current frame before a section of interval a into a tray reasoning algorithm for detecting the crossing tray;
if the current frame does not detect the out-of-range tray, returning to the step B2, and continuing to wait for the video stream of the next frame to continue to be sent into a tray reasoning algorithm;
if the current frame detects that the out-of-range tray exists, continuously detecting whether the out-of-range tray of the current frame and the out-of-range tray of the previous frame of the current frame move or not;
b5, if the out-of-range tray is detected to move, starting a dynamic timer to start timing, and judging whether the dynamic timer reaches the preset dynamic trigger time or not; if the triggering time is reached, immediately alarming, and resetting the dynamic timer, wherein the alarming state is false; if the dynamic trigger time is not reached, returning to the step B2, and continuing to wait for the video stream of the next frame to continue to be sent into the tray reasoning algorithm;
b6, if the tray is detected to be not moved, the dynamic timer is cleared, and whether the alarm state is false is checked; if the alarm state is false, starting a static timer and judging whether the static timer reaches the static trigger time or not; if the static trigger time is reached, alarming, setting an alarm state as true, and resetting a static timer; if the static trigger time is not reached, not alarming, returning to the step B2, and continuing to wait for the video stream of the next frame to continue to be sent into the tray reasoning algorithm;
the method for detecting the out-of-range tray in the step B2 comprises the following steps: dividing two intersecting straight lines as warning lines, and calibrating through three points on the warning lines: firstly, calculating one straight line through two points, wherein the calculation equation is as follows:
after transformation, the method comprises the following steps:
the other straight line is calculated as:
wherein (x) 1 ，y 1 )，(x 2 ，y 2 )，(x 3 ，y 3 ) Three points known on the guard line, respectively; detecting the position coordinates (x, y) of each tray by a tray classifier, and respectively bringing x into the calculation equation of two straight lines to obtain a corresponding value y p1 And y p2 If y satisfies the condition y>y p1 And condition y>y p1 At least one of the two judgment conditions judges that the tray is out of range.
2. The tray stacking border crossing detection method according to claim 1, wherein the filtering process of the isolated target tray in step S2 comprises the steps of:
a1, starting an algorithm;
a2, pushing a frame of video stream to a tray positioning algorithm;
a3, carrying out pallet target identification by utilizing an adaptive lifting tree cascade classifier;
a4, creating a pure black picture with the same size as the original video frame picture, marking the identified center point coordinates of each tray on the pure black picture, and setting the pixel value of the center point position;
a5, carrying out morphological opening operation by using the structural unit;
a6, setting a convolution kernel, and carrying out convolution operation on the marked pure black graph;
a7, performing threshold binarization on the marked pure black image after convolution operation, and taking the binarized image as a mask;
and A8, carrying out union operation on the original image of the video and the mask to obtain RIO, and cleaning all the isolated tray positions.
3. The tray stacking border crossing detection method according to claim 1, wherein in the step A5, a morphological opening operation is performed using a structural unit of 3*3 or 5*5.
4. The tray stacking border crossing detection method according to claim 1, wherein in the step A6, the convolution kernel size is set to be between 15 x 15 and 20 x 20, and each kernel element size is 1 divided by the convolution kernel length-width product.
5. The tray stacking border crossing detection method according to claim 1, wherein the threshold value of the binarization in the step A7 is selected to be between 200 and 230.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111565026.7A CN114613098B (en) | 2021-12-20 | 2021-12-20 | Tray stacking out-of-range detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111565026.7A CN114613098B (en) | 2021-12-20 | 2021-12-20 | Tray stacking out-of-range detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114613098A CN114613098A (en) | 2022-06-10 |
| CN114613098B true CN114613098B (en) | 2023-11-03 |
Family
ID=81857581
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111565026.7A Active CN114613098B (en) | 2021-12-20 | 2021-12-20 | Tray stacking out-of-range detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114613098B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003200927A (en) * | 2002-01-07 | 2003-07-15 | Japan Pallet Rental Corp | Pallet aligning unit |
| CN104709696A (en) * | 2014-12-31 | 2015-06-17 | 深圳市进林科技有限公司 | Goods placing and taking method, device and system for tray piling trolley |
| CN106144611A (en) * | 2016-08-31 | 2016-11-23 | 中国大冢制药有限公司 | Piling position bracket tray positioning apparatus |
| GB2538700A (en) * | 2015-04-27 | 2016-11-30 | Thermoteknix Systems Ltd | Conveyor belt monitoring system and method |
| CN109515478A (en) * | 2018-10-09 | 2019-03-26 | 高新兴创联科技有限公司 | Vehicle positioning method applied to railcar Safety of Shunting Operation supplemental protective system |
| CN112161821A (en) * | 2020-09-24 | 2021-01-01 | 中国铁路上海局集团有限公司科学技术研究所 | Railway empty and heavy vehicle mixed marshalling goods train acceleration test method |
| CN113139521A (en) * | 2021-05-17 | 2021-07-20 | 中国大唐集团科学技术研究院有限公司中南电力试验研究院 | Pedestrian boundary crossing monitoring method for electric power monitoring |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7822696B2 (en) * | 2007-07-13 | 2010-10-26 | Microsoft Corporation | Histogram-based classifiers having variable bin sizes |
-
2021
- 2021-12-20 CN CN202111565026.7A patent/CN114613098B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003200927A (en) * | 2002-01-07 | 2003-07-15 | Japan Pallet Rental Corp | Pallet aligning unit |
| CN104709696A (en) * | 2014-12-31 | 2015-06-17 | 深圳市进林科技有限公司 | Goods placing and taking method, device and system for tray piling trolley |
| GB2538700A (en) * | 2015-04-27 | 2016-11-30 | Thermoteknix Systems Ltd | Conveyor belt monitoring system and method |
| CN106144611A (en) * | 2016-08-31 | 2016-11-23 | 中国大冢制药有限公司 | Piling position bracket tray positioning apparatus |
| CN109515478A (en) * | 2018-10-09 | 2019-03-26 | 高新兴创联科技有限公司 | Vehicle positioning method applied to railcar Safety of Shunting Operation supplemental protective system |
| CN112161821A (en) * | 2020-09-24 | 2021-01-01 | 中国铁路上海局集团有限公司科学技术研究所 | Railway empty and heavy vehicle mixed marshalling goods train acceleration test method |
| CN113139521A (en) * | 2021-05-17 | 2021-07-20 | 中国大唐集团科学技术研究院有限公司中南电力试验研究院 | Pedestrian boundary crossing monitoring method for electric power monitoring |
Non-Patent Citations (1)
| Title |
|---|
| "智能物流托盘的研究和应用分析";黄欢, 毕晨静;《现代信息科技》;第5卷(第14期);154-157 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114613098A (en) | 2022-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111832415B (en) | Truck safety intelligent protection system for container hoisting operation | |
| CN112509315B (en) | Traffic accident detection method based on video analysis | |
| CN111242108B (en) | Belt transfer point coal blockage identification method based on target detection | |
| CN100454355C (en) | Video method for collecting information of vehicle flowrate on road in real time | |
| CN104573659B (en) | A kind of driver based on svm takes phone-monitoring method | |
| CN110895662A (en) | Vehicle overload alarm method and device, electronic equipment and storage medium | |
| CN102496275B (en) | Method for detecting overload of coach or not | |
| CN104966059B (en) | Driver based on intelligent monitor system drives detection method of making a phone call | |
| CN102982313A (en) | Smog detecting method | |
| CN109492548B (en) | Method for obtaining region mask picture based on video analysis | |
| CN102013013B (en) | Fatigue driving monitoring method | |
| CN113076899A (en) | High-voltage transmission line foreign matter detection method based on target tracking algorithm | |
| CN111523415A (en) | Image-based two-passenger one-dangerous vehicle detection method and device | |
| CN111696135A (en) | Intersection ratio-based forbidden parking detection method | |
| Srinivas et al. | Image processing edge detection technique used for traffic control problem | |
| CN118397571B (en) | Reservoir intelligent monitoring method and system based on AI video technology | |
| CN114613098B (en) | Tray stacking out-of-range detection method | |
| CN114565886A (en) | Method for detecting dangerous behavior of personnel crossing station tracks in railway industry | |
| CN114655655A (en) | Conveyor belt deviation detection method based on UNet network | |
| CN103177248A (en) | Rapid pedestrian detection method based on vision | |
| CN117789403A (en) | Real-time detection method and system for personnel invasion in dangerous area of driving operation | |
| CN115072510B (en) | Elevator car passenger intelligent identification and analysis method and system based on door switch | |
| Kim | Automated decision making in road traffic monitoring by on-board unmanned aerial vehicle system | |
| CN115171220A (en) | Abnormal traffic early warning method and device and electronic equipment | |
| CN113673383A (en) | Time-space domain obstacle detection method and system for complex road scene |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20220902 Address after: Building 536, No. 465, Middle Tianmu Road, Jing'an District, Shanghai, 200070 Applicant after: INSTITUTE OF SCIENCE AND TECHNOLOGY, CHINA RAILWAY SHANGHAI GROUP CO.,LTD. Applicant after: GOSUNCN CHUANGLIAN TECHNOLOGY Co.,Ltd. Address before: 310013 5 floor, 10 building, 30 Wan Tang Road, Xihu District, Hangzhou, Zhejiang. Applicant before: GOSUNCN CHUANGLIAN TECHNOLOGY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| CB02 | Change of applicant information |
Address after: Building 536, No. 465, Middle Tianmu Road, Jing'an District, Shanghai, 200070 Applicant after: INSTITUTE OF SCIENCE AND TECHNOLOGY, CHINA RAILWAY SHANGHAI GROUP CO.,LTD. Applicant after: Gaoxing Chuanglian Technology Co.,Ltd. Address before: Building 536, No. 465, Middle Tianmu Road, Jing'an District, Shanghai, 200070 Applicant before: INSTITUTE OF SCIENCE AND TECHNOLOGY, CHINA RAILWAY SHANGHAI GROUP CO.,LTD. Applicant before: GOSUNCN CHUANGLIAN TECHNOLOGY Co.,Ltd. |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |