CN117036340A - Counting method and system for high-speed conveying scene - Google Patents
Counting method and system for high-speed conveying scene Download PDFInfo
- Publication number
- CN117036340A CN117036340A CN202311269820.6A CN202311269820A CN117036340A CN 117036340 A CN117036340 A CN 117036340A CN 202311269820 A CN202311269820 A CN 202311269820A CN 117036340 A CN117036340 A CN 117036340A
- Authority
- CN
- China
- Prior art keywords
- counting
- product
- analysis
- time
- result
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000004458 analytical method Methods 0.000 claims abstract description 162
- 238000012937 correction Methods 0.000 claims abstract description 108
- 230000005540 biological transmission Effects 0.000 claims abstract description 94
- 238000001514 detection method Methods 0.000 claims abstract description 89
- 230000002159 abnormal effect Effects 0.000 claims abstract description 45
- 230000000694 effects Effects 0.000 claims abstract description 42
- 238000012545 processing Methods 0.000 claims abstract description 34
- 238000012795 verification Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 24
- 238000010276 construction Methods 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 14
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 description 5
- 238000013507 mapping Methods 0.000 description 5
- 238000004806 packaging method and process Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 230000000903 blocking effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30236—Traffic on road, railway or crossing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a counting method and a counting system for a high-speed conveying scene, which relate to the technical field of data processing, and are used for obtaining a real-time counting result from detection and counting of target products, counting a first counting time, judging a threshold value, marking abnormal products, obtaining photoelectric effect parameters and determining a first counting correction result; the image acquisition and the processing analysis are carried out to obtain a second counting correction result, the consistency judgment of the counting correction result is carried out, the correction and the alarm operation are carried out in the later step, the technical problems that the conventional photoelectric counting mode in the prior art is insufficient in counting support and cannot adapt to the counting requirement in a high-speed transmission state, the transmission counting of single products cannot be accurately carried out, the accuracy of the counting result is insufficient are solved, the counting accuracy judgment is carried out based on the shielding interval time of the single products aiming at the preliminary counting result, the photoelectric effect verification and the image detection verification are carried out step by step in a later position, the counting correction is accurately carried out, and the transportation state compliance of the counting result is ensured.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a counting method and system for a high-speed conveying scene.
Background
For products in high-speed transmission, such as the transmission and packaging of paper diapers on a transmission belt, when the products are transmitted and counted, the products are affected by non-directional factors and limited in technology, and stacking of the products possibly exists, so that the counting result deviates from the actual transmission condition, and the relative later packaging and other execution operations are affected.
In the prior art, the conventional photoelectric counting mode counts the intermittent blocking of laser through adjacent products and intervals between the adjacent products, the counting inaccuracy can occur when the products are stacked, the counting support degree is insufficient, the counting requirement in a high-speed transmission state cannot be met, the transmission counting of single products cannot be accurately performed, and the accuracy of the counting result is insufficient.
Disclosure of Invention
The application provides a counting method and a counting system for a high-speed conveying scene, which are used for solving the technical problems that the counting support degree is insufficient, the counting requirement in a high-speed conveying state cannot be met, the conveying counting of a single product cannot be accurately carried out, and the accuracy of a counting result is insufficient in the prior art.
In view of the above, the present application provides a counting method and system for high-speed conveying scenes.
In a first aspect, the present application provides a counting method for a high-speed delivery scenario, the method comprising:
in the high-speed transmission and conveying process of target products, detecting and counting according to gaps generated between adjacent target products through a preset first photoelectric detection device at a first point on a conveying device to obtain a real-time counting result and counting to obtain a first counting time;
acquiring a preset first time threshold, judging whether the first counting time is greater than the first time threshold, and if not, continuing counting;
if yes, marking to obtain an abnormal product, starting a second photoelectric detection device preset at a second point on the transmission device, obtaining photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, and performing photoelectric stacking processing analysis to obtain a first counting correction result and a first stacking mode;
acquiring images of a plurality of target products on the transmission device according to the first time threshold by starting an image detection device preset at a third point on the transmission device to obtain a product image set, wherein each product image comprises the abnormal products, and the first point, the second point and the third point are in front-back sequence in the transmission direction of the transmission device;
Processing the product image set to obtain a first product image set and a second product image set, and performing image stacking processing analysis according to the first stacking mode to obtain a second counting correction result;
and verifying whether the first counting correction result is consistent with the second counting correction result, if so, correcting the real-time counting result to obtain a corrected counting result, and if not, recording and alarming.
In a second aspect, the present application provides a counting system for high speed delivery scenarios, the system comprising:
the detection counting module is used for detecting and counting according to gaps generated between adjacent target products through a preset first photoelectric detection device at a first point on the transmission device in the high-speed transmission and conveying process of the target products, so as to obtain a real-time counting result and count to obtain a first counting time;
the threshold judging module is used for acquiring a preset first time threshold, judging whether the first counting time is greater than the first time threshold, and if not, continuing counting;
the first counting correction result acquisition module is used for marking to obtain an abnormal product if the abnormal product is detected, starting a second photoelectric detection device preset at a second point on the transmission device, acquiring photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, and performing photoelectric stacking processing analysis to obtain a first counting correction result and a first stacking mode;
The image acquisition module is used for acquiring images of a plurality of target products on the transmission device according to the first time threshold value by starting an image detection device preset at a third point on the transmission device to obtain a product image set, wherein each product image comprises the abnormal products, and the first point, the second point and the third point are in front-back sequence in the transmission direction of the transmission device;
the second counting correction result acquisition module is used for processing the product image set to obtain a first product image set and a second product image set, and performing image stacking processing analysis according to the first stacking mode to obtain a second counting correction result;
and the result verification module is used for verifying whether the first counting correction result is consistent with the second counting correction result, if so, correcting the real-time counting result to obtain a corrected counting result, and if not, recording and alarming.
One or more technical schemes provided by the application have at least the following technical effects or advantages:
In the counting method for the high-speed conveying scene provided by the embodiment of the application, in the high-speed conveying process of the target product, a first point on a conveying device is detected and counted through a preset first photoelectric detection device according to the gap generated between adjacent target products, a real-time counting result is obtained, a first counting time is obtained by statistics, whether the first counting time is larger than the first time threshold value or not is judged, and if not, the counting is continued; if yes, marking to obtain an abnormal product, starting a second photoelectric detection device preset at a second point on the transmission device, obtaining photoelectric effect parameters, inputting the photoelectric effect parameters into a first counting analysis module, and obtaining a first counting correction result; starting an image detection device preset at a third point on the transmission device, acquiring images of a plurality of target products on the transmission device according to the first time threshold, acquiring and processing a product image set, acquiring a first product image set and a second product image set, inputting the first product image set and the second product image set into a second counting analysis module, acquiring a second counting correction result, verifying whether the first counting correction result is consistent with the second counting correction result, and if so, correcting the real-time counting result; if not, recording and alarming are carried out, the technical problems that the conventional photoelectric counting mode in the prior art is insufficient in counting support, cannot adapt to the counting requirement under a high-speed transmission state, cannot accurately carry out transmission counting of single products, and the accuracy of a counting result is insufficient are solved, the counting accuracy judgment is carried out on the basis of shielding interval time of the single products according to a preliminary counting result, photoelectric effect verification and image detection verification are carried out step by step at a rear position, counting correction is carried out accurately, and the transportation live condition compliance of the counting result is guaranteed.
Drawings
FIG. 1 is a schematic flow chart of a counting method for a high-speed conveying scene;
fig. 2 is a schematic diagram of a first count correction result obtaining process in a counting method for a high-speed conveying scene according to the present application;
FIG. 3 is a schematic diagram of a second counting calibration result obtaining process in a counting method for a high-speed conveying scenario;
fig. 4 is a schematic diagram of a counting system for a high-speed conveying scenario.
Reference numerals illustrate: the device comprises a detection counting module 11, a threshold judging module 12, a first counting correction result obtaining module 13, an image acquisition module 14, a second counting correction result obtaining module 15 and a result verifying module 16.
Detailed Description
The application provides a counting method and a counting system for a high-speed conveying scene, which are used for detecting and counting target products to obtain a real-time counting result, counting first counting time and judging whether the first counting time is larger than a first time threshold value, and if not, continuing counting; if yes, marking an abnormal product, acquiring photoelectric effect parameters, and analyzing and acquiring a first counting correction result; the method comprises the steps of collecting images of a plurality of target products at a third point, carrying out image processing analysis to obtain a second counting correction result, carrying out counting correction result consistency judgment, and carrying out correction and alarm operation based on the judgment result, so that the technical problems that the conventional photoelectric counting mode in the prior art is insufficient in counting support degree, cannot adapt to the counting requirement in a high-speed transmission state, cannot accurately carry out transmission counting of single products, and is insufficient in counting result accuracy are solved.
Example 1
As shown in fig. 1, the present application provides a counting method for a high-speed delivery scenario, the method comprising:
step S100: in the high-speed transmission and conveying process of target products, detecting and counting according to gaps generated between adjacent target products through a preset first photoelectric detection device at a first point on a conveying device to obtain a real-time counting result and counting to obtain a first counting time;
further, the step S100 of the present application further includes:
step S110: detecting and counting according to gaps generated between adjacent target products to obtain a real-time counting result M and a real-time counting result M+1, wherein M is an integer greater than or equal to 1;
step S120: and obtaining the first counting time according to the time interval for generating the real-time counting result M and the real-time counting result M+1.
Specifically, for products in high-speed transmission, such as the transmission and packaging of paper diapers on a transmission belt, when the products are transmitted and counted, the counting result is influenced by an unoriented factor, so that deviation exists between the counting result and the actual transmission condition, the counting accuracy is insufficient, and the relative follow-up packaging and other execution operations are influenced. According to the counting method for the high-speed conveying scene, aiming at the abnormal detection time point determined by preliminary detection, the follow-up detection is carried out based on different detection counting modes, the counting correction is carried out, the result comparison is carried out, and if the results are consistent, the real-time counting adjustment is carried out so as to maximize the guarantee of the counting accuracy.
Specifically, the target product is a counting product to be conveyed at a high speed, the first point, for example, a detection point near an initial conveying point, is determined on the conveying device, and the conveying device is used for conveying the target product, for example, a conveying belt. And configuring the first photoelectric detection device, such as a light emitter and a light receiver, at the first point, and performing connection transmission judgment of the products based on the blocking of the target product to light and the photoelectric effect that the gap cannot block the light by emitting laser due to gaps between two products transmitted adjacently, so as to finish the transmission quantity measurement of the target product, and taking the measurement result as the real-time counting result. And determining the time of counting detection by the first photoelectric detection device as the first counting time.
Specifically, for the interval generated between adjacent target products, based on the generated continuous photoelectric effect, namely, no light shielding exists, no light shielding exists, the product is detected to generate the light shielding effect, the detection time, namely, the interval between the last counting and the next counting is determined, wherein the interval is the time when the product shields light, namely, the time when the last interval and the next interval are obtained by detection, and the time is used as the real-time counting result M and the real-time counting result M+1. And calculating an interval time difference between the real-time counting result M and the real-time counting result M+1 as the first counting time. And further performing a count accuracy determination analysis based on the first count time.
Step S200: acquiring a preset first time threshold, judging whether the first counting time is greater than the first time threshold, and if not, continuing counting;
further, a preset first time threshold is obtained, and whether the first count time threshold is greater than the first time threshold is determined, where step S200 of the present application further includes:
step S210: acquiring the transmission speed of the transmission device for transmitting the target product;
step S220: calculating to obtain the first time threshold according to the size of the target product and the transmission speed;
step S230: and judging whether the first counting time is larger than the first time threshold.
Specifically, the preset first time threshold is obtained, and the preset first time threshold is a standard for judging counting accuracy. Specifically, the transmission speed of the target product transmitted by the transmission device is determined, for example, the transmission speed can be determined by configured control parameters of the transmission device. And (3) measuring the size of the target product, wherein the measurement can be directly measured or invoked based on the production parameters of the target product, generally speaking, the target product transmitted in batches is a product with the same specification, the corresponding product size is the same, the size of the target product is divided by the transmission speed, and the calculation is performed as the first time threshold, namely, the transmission time interval when a single product completely passes through the first photoelectric detection device.
Further, the first time threshold is used as a limiting standard, whether the first counting time is larger than the first time threshold is judged, if not, the first counting time accords with the counting time of a single product, the counting time belongs to a normal transmission counting state, and the subsequent counting is continued; if yes, the first counting time does not accord with the counting time of a single product, and product overlapping possibly exists, and the first counting time belongs to an abnormal transmission counting state, and further checking and judging are needed to be carried out, so that counting errors are avoided.
Step S300: if yes, marking to obtain an abnormal product, starting a second photoelectric detection device preset at a second point on the transmission device, obtaining photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, and performing photoelectric stacking processing analysis to obtain a first counting correction result and a first stacking mode;
further, as shown in fig. 2, the method further includes obtaining parameters of photoelectric effect generated by the abnormal product in the light curtain area in the second photoelectric detection device, performing a photoelectric stacking analysis to obtain a first count correction result and a first stacking mode, and the step S300 further includes:
Step S310: acquiring photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, wherein the second photoelectric detection device comprises light emitting equipment and a light curtain area for receiving emitted light, and the photoelectric effect parameters are generated by shielding the emitted light by the abnormal product;
step S320: acquiring a sample photoelectric effect parameter set and a sample first count correction result set according to detection data of the second photoelectric detection device in the historical time;
step S330: the sample photoelectric effect parameter set and the sample first counting correction result set are used as construction data, and a first counting analysis module is obtained through construction training based on a feedforward neural network;
step S340: inputting the photoelectric effect parameters into the first counting analysis module to obtain the first counting correction result.
Specifically, if the first counting time is greater than the first time threshold, it indicates that there is abnormal transmission counting, for example, there is two or more overlapping products, and the current detected product is identified and used as the abnormal product. And further starting a second photoelectric detection device preset at a second point on the transmission device, wherein the second point is a position which is arranged at the first point behind the transmission device in the transmission direction, the detection mode of the second photoelectric detection device is the same as that of the first photoelectric detection device, the specific execution modes are different, and the contour identification is carried out based on the judgment of the shielding interval time or based on the shielding domain respectively. And counting, analyzing and correcting the detected photoelectric effect parameters.
Specifically, a time point when the first counting time is larger than the first time threshold is marked as abnormal time, and a target product positioned on one side of the transmission direction of the first point under the abnormal time is taken as the abnormal product. And at the second point, acquiring photoelectric effect parameters generated by the abnormal product in a light curtain area, such as a product shielding area, based on the second photoelectric detection device. Specifically, according to the transmission speed of the target product transmitted by the transmission device, determining the time when the abnormal product reaches the second point, and starting the second photoelectric detection device. The second photoelectric detection device comprises a light emitter and a light receiver, the emitted light and the light curtain area are all an area and can cover the whole target product, photoelectric effect parameters are formed according to light shielded by the target product and light which is not shielded, the product profile can be determined based on the light shielding state so as to carry out overlapping judgment, whether two or more target products are overlapped is judged, and a stacking mode is determined based on the product profile and used as the first stacking mode. For example, a plurality of light emitting devices may be included in the light emitter, and a photo resistor may be disposed in the light receiver as a photo-electric effect parameter by a resistance value change of the photo resistor due to a change in the amount of received light after the target product blocks light.
Further, the historical time is a self-defined detection time interval of the second photoelectric detection device, detection data in the historical time is called, the sample photoelectric effect parameter set and the sample first count correction result set are determined, the sample photoelectric effect parameter set corresponds to the sample first count correction result set one by one, and identification and acquisition can be directly carried out based on the detection data. Mapping and correlating the sample photoelectric effect parameter set with the sample first counting correction result set, and generating the first counting analysis module by performing feedforward neural network training as the construction data, wherein the first counting analysis module is a self-built auxiliary analysis tool for product contour recognition and overlap judgment. The photoelectric effect parameter is input into the first count analysis module, the first count correction result is determined, for example, count +1, count +2, etc. based on the number of overlapping products.
Step S400: acquiring images of a plurality of target products on the transmission device according to the first time threshold by starting an image detection device preset at a third point on the transmission device to obtain a product image set, wherein each product image comprises the abnormal products, and the first point, the second point and the third point are in front-back sequence in the transmission direction of the transmission device;
Specifically, the third point is a point location on the transmission device and placed at the second point, and the first point, the second point and the third point are sequentially in a front-back order in the transmission direction of the transmission device. And arranging an image detection device at the third point, wherein the image detection device is used for carrying out image acquisition of product transmission, such as monitoring equipment and the like. According to the transmission speed of the target product transmitted by the transmission device, based on the transmission distance between the second point and the third point, the time when the abnormal product reaches the third point can be determined by dividing the transmission speed, and further the time for acquiring images can be determined, and the image detection device is started to acquire images. The first time threshold, namely a time zone which is used for collecting images of the abnormal products and is set in a self-defined mode, images of a plurality of target products on the transmission device are collected based on the time period of collecting the images of the first time threshold, time sequence adjustment is conducted on the collected images, and a product image set is generated, wherein the product image set is source data for carrying out overlapping analysis judgment on the abnormal products.
Step S500: processing the product image set to obtain a first product image set and a second product image set, and performing image stacking processing analysis according to the first stacking mode to obtain a second counting correction result;
Step S600: and verifying whether the first counting correction result is consistent with the second counting correction result, if so, correcting the real-time counting result to obtain a corrected counting result, and if not, recording and alarming.
Specifically, the second counting analysis module is constructed and is an auxiliary tool constructed based on a slow network and used for performing image convolution feature recognition analysis, and the auxiliary tool comprises a first analysis channel, a second analysis channel and a feature analysis branch. Inputting the first product image set and the second product image set into the second counting analysis module in the counting analysis model, inputting the first product image set into the first analysis channel, inputting the second product image set into the second analysis channel, performing feature recognition extraction and analysis, and outputting the second counting correction result.
Further, the first counting correction result and the second counting correction result are analysis correction results based on different dimensions, the first counting correction result and the second counting correction result are corrected to avoid possible abnormal correction results, if the first counting correction result and the second counting correction result are consistent, the real-time counting result is corrected to obtain the correction counting result, and if the first counting correction result and the second counting correction result are in an accurate correction state, the real-time counting result is corrected to obtain the correction counting result; if the two are inconsistent, indicating that correction abnormality exists, recording correction processing data, alarming, scheduling on-site personnel for field inspection based on abnormality alarming, and maximizing the guarantee of counting accuracy.
Further, the processing of the product image set to obtain a first product image set and a second product image set, and the step S500 of the present application further includes:
step S510: selecting the product images in the product image set according to a first frequency, and performing downsampling processing to obtain the first product image set;
step S520: and selecting the product images in the product image set according to a second frequency to obtain the second product image set, wherein the first frequency is larger than the second frequency.
Further, as shown in fig. 3, according to the first stacking mode, an image stacking process analysis is performed to obtain a second count correction result, and the present application further includes step S530, including:
step S531: acquiring the plurality of stacking modes, wherein the plurality of stacking modes comprise the first stacking mode;
step S532: constructing a first count analysis unit corresponding to the first stacking mode according to detection data in the history time of the image detection device;
step S533: continuously constructing a plurality of counting analysis units corresponding to other plurality of stacking modes to obtain a second counting analysis module;
Step S534: inputting the first product image set and the second product image set into a first count analysis unit corresponding to the first stacking mode, and obtaining the second count correction result.
Further, according to the detection data in the history time of the image detection device, a first count analysis unit corresponding to the first stacking mode is constructed, and step S532 of the present application further includes:
step S5321: processing according to the first frequency and the second frequency according to a plurality of sample product image sets and a plurality of sample second count correction results acquired and obtained in the historical time of the image detection device, and obtaining a plurality of sample first product image sets and a plurality of sample second product image sets;
step S5322: and constructing the first count analysis unit by adopting the plurality of sample first product image sets, the plurality of sample second product image sets and the plurality of sample second count correction results as construction data.
Further, the step S5322 of constructing the first count analysis unit using the plurality of sample first product image sets, the plurality of sample second product image sets, and the plurality of sample second count correction results as construction data further includes:
Step S53221: based on a slow network, a first analysis channel, a second analysis channel and a characteristic analysis branch in the first counting analysis unit are constructed, wherein output layers of the first analysis channel and the second analysis channel are connected with input layers of the characteristic analysis branch, input data of the first analysis channel and the second analysis channel are a first product image set and a second product image set, and output data of the characteristic analysis branch is a second counting correction result;
step S53222: dividing and combining the plurality of sample first product image sets, the plurality of sample second product image sets and the plurality of sample second count correction results to obtain a training set, a verification set and a test set;
step S53223: performing supervised training on the first analysis channel, the second analysis channel and the feature analysis branch by adopting the training set until a preset convergence condition is reached;
step S53224: and testing the first analysis channel, the second analysis channel and the characteristic analysis branch by adopting the verification set and the test set, and obtaining the first counting analysis unit under the condition that the accuracy meets the preset requirement.
Specifically, the product image collection is processed based on the detection requirements to adapt to the later image analysis requirements. And setting the first frequency and the second frequency, namely, the standard of image extraction, wherein the first frequency is larger than the second frequency, for example, setting the ratio of the first frequency to the second frequency to be 4, for example, the first frequency is used for acquiring one image every 2 product images, and the second frequency is used for acquiring one image every 8 product images. And extracting the product images in the product image set based on the first frequency, for example, extracting every two images, and further performing downsampling processing on the selected images to reduce image resolution, facilitate dynamic image feature capture, and acquire the first product image set, wherein the first product image set is used for dynamic feature recognition. And selecting the product images in the product image set based on the second frequency, for example, extracting 1 image every 8 product images in the product image set, not performing downsampling processing, and reserving more image details as the second product image set, wherein the second product image set is used for performing static refinement feature recognition.
A second count analysis module is further constructed. Specifically, the second counting analysis module comprises a plurality of counting analysis units, and is used for performing targeted independent analysis of a plurality of stacking modes. Various states of the product stack, such as vertical stack, horizontal stack, vertical-horizontal composite stack, etc., are determined as various stack patterns, and differential analysis is performed for the various stack patterns. And extracting one of the plurality of stacking modes based on the plurality of stacking modes, and constructing the first count analysis unit which is adapted to the first stacking mode as the first stacking mode. And determining a historical time, namely, a time interval for calling a once acquired image, and acquiring a plurality of sample product image sets of the image detection device and a plurality of sample second count correction results, wherein the sample count correction results are for example counts of +1, +2 and the like, and the plurality of sample product image sets correspond to the plurality of sample second count correction result mappings based on the historical time. Performing image extraction and downsampling processing on the plurality of sample product images based on the first frequency, and generating a plurality of sample first product image sets in an integrated manner; and for the plurality of sample first product image sets which are not subjected to downsampling processing, performing image extraction based on the second frequency, and determining the plurality of sample second product image sets. Further, mapping and correlating the plurality of sample first product image sets, the plurality of sample second product image sets and the plurality of sample second count correction results, and constructing the first count analysis unit as construction data.
Specifically, based on the slow network, the first analysis channel, the second analysis channel and the feature analysis branch are configured, and an output layer of the first analysis channel and an output layer of the second analysis channel are connected with an input layer of the feature analysis branch to form the first counting analysis unit network architecture. Further, a dividing ratio, that is, a self-defined standard for dividing samples, is set, the first product image sets of the plurality of samples are divided based on the dividing ratio, mapping correspondence and division of the second product image sets of the plurality of samples and the second count correction results of the plurality of samples are synchronously performed based on the dividing results, and the mapping correspondence and division are combined based on the dividing results, that is, a plurality of determination training sets, verification sets and test sets are determined, and the plurality of determination training sets, verification sets and test sets correspond to associated sample sets of a plurality of groups of identical product images respectively. And based on the training set, matching and supervising the first analysis channel, the second analysis channel and the feature analysis branch until the preset convergence condition is reached, for example, the preset training times are met, training is stopped, the first counting analysis unit with the completed training is obtained, and in order to avoid the first counting analysis unit from being converged in place or having fitting, the verification set and the test set are further input into the first counting analysis unit with the completed training for verification test, so that the analysis accuracy of the first counting analysis unit is obtained. Judging whether the accuracy meets a preset requirement, for example, 80% or more, if so, indicating that the first counting analysis unit meets the analysis requirement; and if not, carrying out repartition and module training on the construction data until the accuracy meets the preset requirement, and acquiring the constructed first count analysis unit, wherein the first count analysis unit is used for carrying out recognition analysis on the image convolution characteristics.
Similarly, based on the plurality of stacking modes, a second stacking mode which is different from the first stacking mode is extracted, and an adaptive sample image is acquired to train and generate a counting analysis unit. And constructing a plurality of counting analysis units corresponding to the plurality of stacking modes, wherein the plurality of counting analysis units are identical in specific construction mode, similar to an execution mechanism in specific unit architecture and different in specific training data. And integrating the plurality of counting analysis units to perform parallel configuration, and generating the second counting analysis module.
Further, the first product image set and the second product image set are input into the first analysis channel and the second analysis channel of the first counting analysis unit corresponding to the first stacking mode, and are used for performing convolution feature extraction and flow to the feature analysis branch, outputting a counting correction result, and integrating a plurality of counting correction results of the plurality of counting analysis units to serve as the second counting correction result.
Further, the first counting analysis module and the second counting analysis module are peer analysis modules which are independently executed, the first counting analysis module and the second counting analysis module are integrated to generate the counting analysis model, and based on the counting analysis modules, the review analysis of abnormal products is carried out on different dimensions, so that analysis efficiency and objectivity can be effectively ensured.
Example two
Based on the same inventive concept as one of the counting methods for a high-speed transportation scenario in the foregoing embodiments, as shown in fig. 4, the present application provides a counting system for a high-speed transportation scenario, the system comprising:
the detection counting module 11 is configured to perform detection counting according to a gap generated between adjacent target products through a preset first photoelectric detection device at a first point on the transmission device in a high-speed transmission and conveying process of the target products, obtain a real-time counting result, and count to obtain a first counting time;
the threshold value judging module 12 is configured to obtain a preset first time threshold value, judge whether the first count time is greater than the first time threshold value, and if not, continue counting;
the first counting correction result obtaining module 13, where the first counting correction result obtaining module 13 is configured to, if yes, mark an abnormal product, start a second photoelectric detection device preset at a second point on the transmission device, obtain a photoelectric effect parameter generated by the abnormal product in a light curtain area in the second photoelectric detection device, and perform a photoelectric stacking process analysis to obtain a first counting correction result and a first stacking mode;
The image acquisition module 14 is configured to acquire images of a plurality of target products on the transmission device according to the first time threshold by starting an image detection device preset at a third point on the transmission device, so as to obtain a product image set, wherein each product image includes the abnormal product, and the first point, the second point and the third point are in a front-back sequence in the transmission direction of the transmission device;
the second counting correction result obtaining module 15 is configured to process the product image set to obtain a first product image set and a second product image set, and perform image stacking processing analysis according to the first stacking mode to obtain a second counting correction result;
the result verification module 16 is configured to verify whether the first count correction result and the second count correction result are consistent, if yes, correct the real-time count result to obtain a corrected count result, and if no, record and alarm.
Further, the system further comprises:
the real-time counting result acquisition module is used for detecting and counting according to gaps generated between adjacent target products to obtain a real-time counting result M and a real-time counting result M+1, wherein M is an integer greater than or equal to 1;
The first counting time acquisition module is used for acquiring the first counting time according to the time interval for generating the real-time counting result M and the real-time counting result M+1.
Further, the system further comprises:
the transmission speed acquisition module is used for acquiring the transmission speed of the target product transmitted by the transmission device;
the first time threshold calculating module is used for calculating and obtaining the first time threshold according to the size of the target product and the transmission speed;
the first counting time judging module is used for judging whether the first counting time is larger than the first time threshold value or not.
Further, the system further comprises:
the photoelectric effect parameter acquisition module is used for acquiring photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, wherein the second photoelectric detection device comprises light emitting equipment and a light curtain area for receiving emitted light, and the photoelectric effect parameters are generated by shielding the emitted light by the abnormal product;
The sample data acquisition module is used for acquiring a sample photoelectric effect parameter set and a sample first count correction result set according to the detection data of the second photoelectric detection device in the historical time;
the training construction module is used for constructing and training to obtain a first counting analysis module based on a feedforward neural network by adopting the sample photoelectric effect parameter set and the sample first counting correction result set as construction data;
and the parameter analysis module is used for inputting the photoelectric effect parameters into the first counting analysis module to obtain the first counting correction result.
Further, the system further comprises:
the first product image set acquisition module is used for selecting product images in the product image set according to a first frequency, and performing downsampling processing to obtain the first product image set;
the second product image set acquisition module is used for selecting the product images in the product image set according to a second frequency to obtain a second product image set, and the first frequency is larger than the second frequency.
Further, the system further comprises:
the mode acquisition module is used for acquiring the plurality of stacking modes, wherein the plurality of stacking modes comprise the first stacking mode;
the first counting analysis unit construction module is used for constructing a first counting analysis unit corresponding to the first stacking mode according to the detection data in the history time of the image detection device;
the counting analysis unit construction modules are used for continuously constructing a plurality of counting analysis units corresponding to other stacking modes to obtain a second counting analysis module;
the counting correction module is used for inputting the first product image set and the second product image set into a first counting analysis unit corresponding to the first stacking mode, and obtaining the second counting correction result.
Further, the system further comprises:
the sample image acquisition module is used for processing according to the first frequency and the second frequency according to a plurality of sample product image sets acquired and obtained in the historical time of the image detection device and a plurality of sample second count correction results to obtain a plurality of sample first product image sets and a plurality of sample second product image sets;
The model building module is used for building the first count analysis unit by adopting the first product image sets of the samples, the second product image sets of the samples and the second count correction results of the samples as building data.
Further, the system further comprises:
the construction module is used for constructing a first analysis channel, a second analysis channel and a characteristic analysis branch in the second counting analysis module based on a slow network, wherein output layers of the first analysis channel and the second analysis channel are connected with an input layer of the characteristic analysis branch, input data of the first analysis channel and the second analysis channel are a first product image set and a second product image set, and output data of the characteristic analysis branch is a second counting correction result;
the sample dividing and combining module is used for dividing and combining the plurality of sample first product image sets, the plurality of sample second product image sets and the plurality of sample second count correction results to obtain a training set, a verification set and a test set;
the monitoring training module is used for monitoring training the first analysis channel, the second analysis channel and the feature analysis branch by adopting the training set until a preset convergence condition is reached;
The testing module is used for testing the first analysis channel, the second analysis channel and the characteristic analysis branch by adopting the verification set and the testing set, and the first counting analysis unit is obtained under the condition that the accuracy rate meets the preset requirement.
The foregoing detailed description of a counting method for a high-speed conveying scenario will be clear to those skilled in the art, and the counting method and system for a high-speed conveying scenario in this embodiment are relatively simple for the device disclosed in the embodiments, and the relevant points refer to the description of the method section.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (9)
1. A counting method for a high-speed transport scenario, the method comprising:
in the high-speed transmission and conveying process of target products, detecting and counting according to gaps generated between adjacent target products through a preset first photoelectric detection device at a first point on a conveying device to obtain a real-time counting result and counting to obtain a first counting time;
acquiring a preset first time threshold, judging whether the first counting time is greater than the first time threshold, and if not, continuing counting;
if yes, marking to obtain an abnormal product, starting a second photoelectric detection device preset at a second point on the transmission device, obtaining photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, and performing photoelectric stacking processing analysis to obtain a first counting correction result and a first stacking mode;
acquiring images of a plurality of target products on the transmission device according to the first time threshold by starting an image detection device preset at a third point on the transmission device to obtain a product image set, wherein each product image comprises the abnormal products, and the first point, the second point and the third point are in front-back sequence in the transmission direction of the transmission device;
Processing the product image set to obtain a first product image set and a second product image set, and performing image stacking processing analysis according to the first stacking mode to obtain a second counting correction result;
and verifying whether the first counting correction result is consistent with the second counting correction result, if so, correcting the real-time counting result to obtain a corrected counting result, and if not, recording and alarming.
2. The method of claim 1, wherein statistically obtaining a first count time comprises:
detecting and counting according to gaps generated between adjacent target products to obtain a real-time counting result M and a real-time counting result M+1, wherein M is an integer greater than or equal to 1;
and obtaining the first counting time according to the time interval for generating the real-time counting result M and the real-time counting result M+1.
3. The method of claim 1, wherein obtaining a preset first time threshold, determining whether the first count time threshold is greater than the first time threshold, comprises:
acquiring the transmission speed of the transmission device for transmitting the target product;
calculating to obtain the first time threshold according to the size of the target product and the transmission speed;
And judging whether the first counting time is larger than the first time threshold.
4. The method of claim 1, wherein obtaining parameters of the photoelectric effect of the abnormal product in the light curtain area in the second photoelectric detection device, performing a photoelectric stacking process analysis, and obtaining a first count correction result and a first stacking mode, comprises:
acquiring photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, wherein the second photoelectric detection device comprises light emitting equipment and a light curtain area for receiving emitted light, and the photoelectric effect parameters are generated by shielding the emitted light by the abnormal product;
acquiring a sample photoelectric effect parameter set, a sample first count correction result set and a sample first stacking mode set according to detection data of the second photoelectric detection device in the historical time;
the sample photoelectric effect parameter set, the sample first count correction result set and the sample first stacking mode set are adopted as construction data, and a first count analysis module is obtained through construction training based on a feedforward neural network;
and inputting the photoelectric effect parameters into the first count analysis module to obtain the first count correction result and a first stacking mode.
5. The method of claim 1, wherein processing the set of product images to obtain a first set of product images and a second set of product images comprises:
selecting the product images in the product image set according to a first frequency, and performing downsampling processing to obtain the first product image set;
and selecting the product images in the product image set according to a second frequency to obtain the second product image set, wherein the first frequency is larger than the second frequency.
6. The method of claim 5, wherein performing an image stacking process analysis based on the first stacking mode to obtain a second count correction result comprises:
acquiring a plurality of stacking modes, wherein the plurality of stacking modes comprise the first stacking mode;
constructing a first count analysis unit corresponding to the first stacking mode according to detection data in the history time of the image detection device;
continuously constructing a plurality of counting analysis units corresponding to other plurality of stacking modes to obtain a second counting analysis module;
inputting the first product image set and the second product image set into a first count analysis unit corresponding to the first stacking mode, and obtaining the second count correction result.
7. The method according to claim 6, wherein constructing a first count analysis unit corresponding to the first stacking mode based on the detection data in the image detection device history time, comprises:
processing according to the first frequency and the second frequency according to a plurality of sample product image sets and a plurality of sample second count correction results acquired and obtained in the historical time of the image detection device, and obtaining a plurality of sample first product image sets and a plurality of sample second product image sets;
and constructing the first count analysis unit by adopting the plurality of sample first product image sets, the plurality of sample second product image sets and the plurality of sample second count correction results as construction data.
8. The method of claim 7, wherein constructing the first count analysis unit using the plurality of sample first product image sets, the plurality of sample second product image sets, and the plurality of sample second count correction results as construction data comprises:
based on a slow network, a first analysis channel, a second analysis channel and a characteristic analysis branch in the first counting analysis unit are constructed, wherein output layers of the first analysis channel and the second analysis channel are connected with input layers of the characteristic analysis branch, input data of the first analysis channel and the second analysis channel are a first product image set and a second product image set, and output data of the characteristic analysis branch is a second counting correction result;
Dividing and combining the plurality of sample first product image sets, the plurality of sample second product image sets and the plurality of sample second count correction results to obtain a training set, a verification set and a test set;
performing supervised training on the first analysis channel, the second analysis channel and the feature analysis branch by adopting the training set until a preset convergence condition is reached;
and testing the first analysis channel, the second analysis channel and the characteristic analysis branch by adopting the verification set and the test set, and obtaining the first counting analysis unit under the condition that the accuracy meets the preset requirement.
9. A counting system for a high speed transport scenario, the system comprising:
the detection counting module is used for detecting and counting according to gaps generated between adjacent target products through a preset first photoelectric detection device at a first point on the transmission device in the high-speed transmission and conveying process of the target products, so as to obtain a real-time counting result and count to obtain a first counting time;
the threshold judging module is used for acquiring a preset first time threshold, judging whether the first counting time is greater than the first time threshold, and if not, continuing counting;
The first counting correction result acquisition module is used for marking to obtain an abnormal product if the abnormal product is detected, starting a second photoelectric detection device preset at a second point on the transmission device, acquiring photoelectric effect parameters generated by the abnormal product in a light curtain area in the second photoelectric detection device, and performing photoelectric stacking processing analysis to obtain a first counting correction result and a first stacking mode;
the image acquisition module is used for acquiring images of a plurality of target products on the transmission device according to the first time threshold value by starting an image detection device preset at a third point on the transmission device to obtain a product image set, wherein each product image comprises the abnormal products, and the first point, the second point and the third point are in front-back sequence in the transmission direction of the transmission device;
the second counting correction result acquisition module is used for processing the product image set to obtain a first product image set and a second product image set, and performing image stacking processing analysis according to the first stacking mode to obtain a second counting correction result;
And the result verification module is used for verifying whether the first counting correction result is consistent with the second counting correction result, if so, correcting the real-time counting result to obtain a corrected counting result, and if not, recording and alarming.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311269820.6A CN117036340B (en) | 2023-09-28 | 2023-09-28 | Counting method and system for high-speed conveying scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311269820.6A CN117036340B (en) | 2023-09-28 | 2023-09-28 | Counting method and system for high-speed conveying scene |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117036340A true CN117036340A (en) | 2023-11-10 |
| CN117036340B CN117036340B (en) | 2023-12-15 |
Family
ID=88635757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311269820.6A Active CN117036340B (en) | 2023-09-28 | 2023-09-28 | Counting method and system for high-speed conveying scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117036340B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140348382A1 (en) * | 2013-05-22 | 2014-11-27 | Hitachi, Ltd. | People counting device and people trajectory analysis device |
| CN106097369A (en) * | 2016-06-16 | 2016-11-09 | 天津工业大学 | A kind of SMT material disc radioscopy image segmentation detection and counting statistics method and device |
| CN106897698A (en) * | 2017-02-24 | 2017-06-27 | 常州常工电子科技股份有限公司 | Classroom number detection method and system based on machine vision Yu binocular coordination technique |
| CN110147792A (en) * | 2019-05-22 | 2019-08-20 | 齐鲁工业大学 | The Key works Drug packing character high-speed detection system and method optimized based on memory |
| CN111563574A (en) * | 2020-05-14 | 2020-08-21 | 河北省机电一体化中试基地 | Machine vision-based white granulated sugar bag automatic counting device and method |
| CN112084964A (en) * | 2020-09-11 | 2020-12-15 | 浙江水晶光电科技股份有限公司 | Product identification apparatus, method and storage medium |
| CN113920073A (en) * | 2021-09-28 | 2022-01-11 | 深圳市雄帝科技股份有限公司 | Thin card counting method and device, electronic equipment and storage medium |
| KR102476048B1 (en) * | 2022-04-07 | 2022-12-12 | 세이지리서치 주식회사 | A method for detecting a process abnormality and a device for detecting a process abnormality |
-
2023
- 2023-09-28 CN CN202311269820.6A patent/CN117036340B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140348382A1 (en) * | 2013-05-22 | 2014-11-27 | Hitachi, Ltd. | People counting device and people trajectory analysis device |
| CN106097369A (en) * | 2016-06-16 | 2016-11-09 | 天津工业大学 | A kind of SMT material disc radioscopy image segmentation detection and counting statistics method and device |
| CN106897698A (en) * | 2017-02-24 | 2017-06-27 | 常州常工电子科技股份有限公司 | Classroom number detection method and system based on machine vision Yu binocular coordination technique |
| CN110147792A (en) * | 2019-05-22 | 2019-08-20 | 齐鲁工业大学 | The Key works Drug packing character high-speed detection system and method optimized based on memory |
| CN111563574A (en) * | 2020-05-14 | 2020-08-21 | 河北省机电一体化中试基地 | Machine vision-based white granulated sugar bag automatic counting device and method |
| CN112084964A (en) * | 2020-09-11 | 2020-12-15 | 浙江水晶光电科技股份有限公司 | Product identification apparatus, method and storage medium |
| CN113920073A (en) * | 2021-09-28 | 2022-01-11 | 深圳市雄帝科技股份有限公司 | Thin card counting method and device, electronic equipment and storage medium |
| KR102476048B1 (en) * | 2022-04-07 | 2022-12-12 | 세이지리서치 주식회사 | A method for detecting a process abnormality and a device for detecting a process abnormality |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117036340B (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3102018B1 (en) | Quality management device and method for controlling quality management device | |
| CN110214120B (en) | Belt inspection system and method | |
| US11105750B2 (en) | Method and system for the automatic measuring of physical and dimensional parameters of multi-segment articles | |
| CN109212618A (en) | Rays safety detection apparatus and safety inspection method | |
| CN104848786B (en) | A kind of glass geometric parameter measurement method | |
| CN113888480A (en) | MES-based quality tracing method and system | |
| CN117036340B (en) | Counting method and system for high-speed conveying scene | |
| CN104729423B (en) | A kind of measuring method for being used to detect special-shaped glass geometric parameter | |
| JP2007163340A (en) | Plate length measuring device and method for measuring plate length | |
| CN103504471A (en) | Device and method for evaluating an end surface of a rod-shaped product of the tobacco processing industry | |
| CN115575407B (en) | Detection method applied to track and tunnel | |
| CN112224788A (en) | Control method for automatically positioning intelligent logistics conveying line | |
| CN116720153A (en) | Information fusion system and method based on multiple sensors | |
| CN115508500A (en) | Visual layered detection method and system for carbon emission measurement | |
| JP7270322B2 (en) | data estimation controller | |
| CN114821025A (en) | Meter identification method and system based on deep learning | |
| CN113762707A (en) | Cargo tracking method and system based on block chain | |
| JP2022052003A (en) | Analysis device, analysis method, and analysis program | |
| CN105480772B (en) | The closing method of piece-rate system and offline removal equipment is picked in package label inspection | |
| CN116209000B (en) | Carrier communication quality detection and analysis method and system for dual-mode module | |
| CN117169282B (en) | Quality detection method and system of composite film for medicine packaging | |
| CN115471505B (en) | Intelligent regulation and control method for case sealer based on visual recognition | |
| CN117053727B (en) | Scanning detection device for accumulating volume of outer surface of material by ranging method | |
| JPH08159728A (en) | Shape measuring instrument | |
| CN117634538B (en) | Bread production counting analysis management system based on machine vision |
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 |