Abnormal order monitoring and early warning method and system based on cloud platform

Abstract

The present invention discloses a cloud-based abnormal order monitoring and early warning method and system, which relates to the technical field of order monitoring, and solves the problem of not analyzing the actual packaging of the corresponding order and the fitting position of the label to confirm whether the corresponding order has been hit or the label has been swapped. The present invention is aimed at the confirmed abnormal order, and preferentially confirms the outbound images and inbound images related to such abnormal order, and then adopts the relevant method of feature confirmation to identify the feature display of the corresponding label in the outbound image, and performs correlation division of the area based on the determined feature points, determines different associated areas with different area parameters, and performs correlation comparison on the confirmed area sequences, so as to identify the specific abnormal reasons of such abnormal orders. By adopting this feature point and area confirmation method, it can be fully determined whether the area sequences of the two are completely consistent, so as to achieve a better evaluation effect.

Description

Technical Field

[0001] The present invention relates to the technical field of order monitoring, and in particular to a cloud platform-based abnormal order monitoring and early warning method and system. Background Art

[0002] Order monitoring refers to a series of activities to track, observe and manage the entire process of order generation to completion. After the corresponding order is generated, it will flow to various transit warehouses. In each transit warehouse or transfer station, the relevant orders need to be entered and transferred out. During the entry and transfer process, the activity track of such orders is generated for the convenience of relevant personnel to view;

[0003] The application with the publication number CN112215528A discloses a key material monitoring method of a visual portable device, which belongs to the field of monitoring. It binds the delivery notice, material order information, and key material transportation plan to the portable monitoring device, transmits the monitoring information during the transportation process to the Internet cloud platform at a fixed time frequency through the mobile network, and compares it according to the pre-set indicator warning threshold. By integrating GIS map information, it realizes real-time monitoring, real-time warning, and real-time processing of material transportation status, and shares logistics transportation status information to realize real-time monitoring of material transportation, strengthen monitoring of material transportation conditions, and greatly improve transportation efficiency and transportation quality. It has the advantages of real-time accurate positioning, real-time visualization of monitoring, RFID intelligent identification, and automatic warning of abnormalities throughout the process, which can be used in a sustainable and cyclical manner, realizing the continuous use mode of "order with goods, equipment with order", and improving the supply guarantee capability and comprehensive quality control capability. It can be widely used in the field of transportation management of power materials.

[0004] During the order delivery and monitoring process, its cloud platform generally identifies whether the corresponding order has stayed at a certain station for a long time or has not been recorded for a long time based on the corresponding monitoring process, so as to make relevant judgments on abnormal orders. However, in the actual monitoring process, it does not analyze the actual packaging of the corresponding order and the fitting position of the label to confirm whether the corresponding order has been hit or the label has been swapped. As a result, its comprehensive monitoring capabilities for delivery orders are insufficient and a better monitoring and early warning effect cannot be achieved. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a cloud platform-based abnormal order monitoring and early warning method and system, which solves the problem of not analyzing the actual packaging of the corresponding order and the fitting position of the label to confirm whether the corresponding order has been impacted or the label has been swapped.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a cloud platform-based abnormal order monitoring and early warning method, comprising the following steps:

[0007] Step 1: When each delivery order is generated and shipped out, collect the shipping image of this delivery order, and confirm the feature vector of this delivery order from the collected shipping image and record it. The shipping image is the image of the surface where the shipping order label is located. The specific method is as follows:

[0008] S11. Confirm the internal shipping order label image from the collected outbound image, and based on the confirmed shipping order label image, calibrate the overall outline of the shipping order label image as the order outline, and place the order outline in a set of two-dimensional coordinate systems, and based on a number of different coordinate points associated with the order outline, determine the corresponding mean point, and calibrate the mean point in the order outline as the feature midpoint of the order outline;

[0009] S12, confirming the overall image of the surface where the shipping order label is located from the outbound image, and determining the feature midpoint belonging to the overall image in the same manner as determining the feature midpoint of the order outline based on the overall edge contour of the overall image;

[0010] S13, starting from the feature midpoint of the whole image and ending at the feature midpoint of the order outline, a set of feature vectors is generated and recorded, and the order number of the corresponding delivery order is also recorded, and the order number is obtained from the delivery order label;

[0011] Step 2: Take the feature vector confirmed in the outbound image of the same delivery order as the standard vector, and compare the feature vector of the same delivery order with the standard vector in the feature vectors confirmed by other transit warehouses for the same delivery order in the future. Based on the comparison results, lock the abnormal order. The specific sub-steps are as follows:

[0012] S21, based on the order number of the shipping order entered by the corresponding transit warehouse, identifying the feature vector associated with the order number and marking it as a standard vector;

[0013] S22, based on the input image of the shipping order, the shipping order label and the center point of the surface where the label is located are confirmed from the input image in the same manner as the outbound image, and based on the determined two sets of center points, a feature vector about the input image is determined starting from the center point of the surface where the label is located to the center point of the shipping order label;

[0014] S23, compare the confirmed feature vector with the standard vector: make the feature vector coincide with the initial point of the standard vector, then make the feature vector rotate in a circle, with the fixed point of the circular rotation being the initial point of the feature vector, and identify whether the feature vector completely coincides with the standard vector during the rotation process:

[0015] If there is a completely overlapping rotation process, it means that there is no abnormality in the transportation process of this delivery order and no processing is required;

[0016] If there is no completely overlapping rotation process, it means that there is an abnormality in the transportation process of this delivery order, and this delivery order is marked as an abnormal order;

[0017] Step 3: Based on the determined abnormal order, the input image and the outbound image associated with the abnormal order are confirmed when the abnormal order is judged as abnormal. The outbound image is used as the standard template in the verification process to determine whether the shipping order label in the input image is abnormal, and the determination result is displayed. The specific sub-steps are:

[0018] S31, directly translate the shipping order label image existing on the input image surface to the outbound image surface, and overlap the center point of the shipping order label image with the center point of the shipping order label image on the outbound image surface, and identify whether the two sets of shipping order label images completely overlap:

[0019] If they completely overlap, it means that there is no abnormality in the shipping order label of this abnormal order, and a packaging abnormality signal about this abnormal order is directly generated and displayed;

[0020] If they are not completely overlapped, the standard feature values ​​are confirmed from the outbound image: the original delivery order label image inside the outbound image is calibrated as a sub-image, the outbound image is calibrated as a parent image, the four corner points of the sub-image are calibrated as pending points, the contour point closest to the corresponding pending point is confirmed from the overall edge contour of the parent image, the determined contour point is calibrated as the pending connection point corresponding to the pending point, the pending point is connected with the pending connection point, the area between the sub-image and the parent image is divided into four areas, and the area parameters of the four areas are determined, a group of areas are randomly selected as the starting area and the area parameters are sorted in a clockwise manner to determine the standard area sequence;

[0021] The shipping order label image moved above the outbound image is marked as a moving image, and the corner points of the moving image are determined, and the determined corner points are marked as pending points. The pending points are connected with the associated pending points in the same manner as the above method of determining the pending points corresponding to the pending points, and the area between the child image and the parent image is divided, and the area parameters of each different area are confirmed. A group of areas are randomly selected as the starting areas and are sorted in a clockwise order to obtain a comparison area sequence;

[0022] S32. If the number of areas in the comparison area sequence is consistent with that in the standard area sequence, a proofreading process is performed: whether the area parameter at the same position in the comparison area sequence is consistent with that in the standard area sequence:

[0023] If they are not consistent, the first group of area parameters in the comparison area sequence is swapped to the last group, and then it is identified whether the area parameters at the same position are consistent. If they are still inconsistent, the swapping is continued. When the swapping is repeated, it is stopped to identify whether there is a swapping process to determine whether the area parameters at the same position in the area sequence are consistent with the standard area sequence. If so, it means that there is no abnormality in the shipping order label of this abnormal order, and a packaging abnormality signal for this abnormal order is directly generated and displayed. If not, it means that the shipping order label of this abnormal order is abnormal, and an order label abnormality signal for this abnormal order is directly generated and displayed;

[0024] If they are consistent, it means that there is no abnormality in the shipping order label of this abnormal order, and a packaging abnormality signal about this abnormal order is directly generated and displayed;

[0025] S33. If the number of areas in the comparison area sequence is inconsistent with the number of areas in the standard area sequence, it means that the shipping order label of this abnormal order is abnormal, and an order label abnormality signal about this abnormal order is directly generated and displayed.

[0026] Preferably, an abnormal order monitoring and early warning system based on a cloud platform includes:

[0027] The outbound recording end collects the outbound image of each delivery order, and based on the collected outbound image, confirms the center point of the delivery order label inside the outbound image and the outbound image, locks and records the feature vector belonging to the delivery order, and simultaneously records the order number of the delivery order label;

[0028] The monitoring and judgment end performs feature analysis on the input images of the delivery orders taken by each transit warehouse, locks the feature vector associated with the input image, compares the confirmed feature vector with the standard vector, identifies whether the two sets of adjacent images completely overlap, and calibrates abnormal orders based on the identification results;

[0029] The abnormal determination end, based on the determined abnormal order, confirms the input image and the outbound image associated with the abnormal order when it is judged as abnormal, uses the outbound image as the standard template in the verification process, determines whether the shipping order label in the input image is abnormal, and displays the confirmed determination result.

[0030] The present invention provides a cloud platform-based abnormal order monitoring and early warning method and system. Compared with the prior art, it has the following beneficial effects:

[0031] The present invention performs image recognition on the related orders for outbound delivery, and confirms the center point of the label associated with the outbound delivery image and the center point of the corresponding image during the recognition process, performs relevant confirmation of the feature vector based on the center point associated between the two, and then determines the corresponding abnormal order through the vector change of each such order during the input process. This abnormal order confirmation method is convenient and fast, and can quickly and effectively lock the related orders with abnormal changes;

[0032] For the confirmed abnormal orders, the outbound images and inbound images related to such abnormal orders are confirmed first, and then the relevant methods of feature confirmation are used to identify the feature display of the corresponding labels in the outbound images, and based on the determined feature points, the associated division of the areas is performed to determine different associated areas with different area parameters. By correlating and comparing the confirmed area sequences, the specific abnormal causes of such abnormal orders can be identified. By using this method of feature point and area confirmation, it is possible to fully and effectively determine whether the area sequences of the two are completely consistent, so as to achieve better evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the process of the present invention;

[0034] Figure 2 Schematic diagram of determining image features of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0036] See also Figure 1 , the present application provides a cloud platform-based abnormal order monitoring and early warning method, comprising the following steps:

[0037] Step 1: When each delivery order is generated and shipped out, collect the shipping image of this delivery order, and confirm the feature vector of this delivery order from the collected shipping image and record it. The shipping image is the image of the side where the shipping order label is located. That is, after this delivery order is generated, it will be placed on the corresponding conveyor belt. When placed, the side with the order label faces up. The corresponding machine vision equipment will then shoot the label side of each running order. The corresponding image shot is the shipping image of the corresponding delivery order. From the shot shipping image, the relevant feature identification and locking can be performed to determine whether such delivery orders will be abnormal in the future. The specific method of determining the feature vector of this delivery order is as follows:

[0038] S11. Confirm the internal delivery order label image from the collected outbound image (when the corresponding order is outbound, the location of the delivery order label can be directly determined based on the relevant scanning results of the corresponding instrument, so that the corresponding delivery order label image can be directly determined based on the location and the scanning results, and the features of the image are also relatively obvious, so it can be quickly confirmed that it belongs to the label order), and based on the confirmed delivery order label image, calibrate the overall contour of the delivery order label image as the order contour, and place the order contour in a set of two-dimensional coordinate systems, and determine the corresponding mean point based on several different coordinate points associated with the order contour, and calibrate the mean point in the order contour as the feature midpoint of the order contour (this method of determining the center point is to confirm the coordinates of the edge points of the contour, and then lock the mean point based on the mean processing process of the coordinates of several points, so as to determine the center point of the corresponding order contour);

[0039] S12, confirming the overall image of the surface where the shipping order label is located from the outbound image (that is, the surface where the corresponding order label is affixed, based on the image resolution and clarity at different distances, the overall image of the surface where the shipping order label is located can be locked), and based on the overall edge contour of the overall image, determining the feature midpoint belonging to the overall image in the same way as determining the feature midpoint of the order contour;

[0040] S13. Generate and record a set of feature vectors starting from the feature midpoint of the whole image and ending at the feature midpoint of the order contour. Also record the order number of the corresponding delivery order, which is obtained from the delivery order label (assuming that the feature midpoint of the whole image is A and the feature midpoint of the order contour is B, then the expression of this feature vector is A to B, that is, the vector direction is the moving direction from A to B, and the vector length is the specific length of the line segment between AB. The reason for confirming the feature vector of each delivery order is to identify whether there is a difference in the vector changes of such orders in the subsequent transfer warehouse and other input links, that is, to monitor and warn such orders in real time, and quickly lock the specific moment when such orders change to abnormal orders, and handle them in time);

[0041] Step 2: Take the feature vector confirmed in the outbound image of the same delivery order as the standard vector, and in the feature vectors confirmed for the same delivery order in other subsequent transit warehouses, check the feature vector of the same delivery order with the standard vector. Based on the check result, lock the abnormal order. Specifically, when the corresponding order is outbound, the corresponding initial feature vector is confirmed, then the confirmed initial feature vector is the corresponding standard vector. For the delivery order associated with this standard vector, it needs to pass through the corresponding transit warehouse during transportation, and the transit warehouse is also provided with the corresponding information entry and image confirmation related processes, so that the feature vector can be confirmed in the same way, and then the abnormal order is confirmed by analyzing whether the feature vector has associated changes. The specific sub-steps of locking the abnormal order are as follows:

[0042] S21, based on the order number of the shipping order entered by the corresponding transit warehouse, identifying the feature vector associated with the order number and marking it as a standard vector;

[0043] S22, based on the input image of the shipping order, the shipping order label and the center point of the surface where the label is located are confirmed from the input image in the same manner as the outbound image, and based on the determined two sets of center points, a feature vector about the input image is determined starting from the center point of the surface where the label is located to the center point of the shipping order label (that is, starting from one point and ending at another point, because a vector has a direction);

[0044] S23, compare the confirmed feature vector with the standard vector: make the feature vector coincide with the initial point of the standard vector, then make the feature vector rotate in a circle, with the fixed point of the circular rotation being the initial point of the feature vector, and identify whether the feature vector completely coincides with the standard vector during the rotation process:

[0045] If there is a completely overlapping rotation process, it means that there is no abnormality in the transportation process of this delivery order and no processing is required;

[0046] If there is no completely overlapping rotation process, it means that there is an abnormality in the transportation process of this delivery order, and this delivery order is marked as an abnormal order. The abnormalities generally include: the order box is deformed, resulting in a change in the determined center point, thereby causing a change in the generated vector; and the order label is tampered with or maliciously removed and then re-attached, resulting in a change in the center point, causing a change in the corresponding vector;

[0047] The rotation process of complete overlap is: two vectors completely overlap in a certain direction, and the overlap synchronization includes the corresponding vector lengths being consistent. If two vectors of unequal lengths also overlap in a certain direction, this type of overlap does not belong to the case of complete overlap. The reason for rotation is that when the corresponding order is entered, there will be inconsistencies in the image entry angles, which will result in two completely identical vectors, but inconsistent directions, so rotation is required to complete the specific process of overlap analysis.

[0048] Step 3: Based on the determined abnormal order, the input image and the outbound image associated with the abnormal order are confirmed when the abnormal order is judged to be abnormal. The outbound image is used as a standard template in the verification process to determine whether the shipping order label in the input image is abnormal, and the determination result is displayed. The specific sub-steps of the determination are as follows:

[0049] S31, directly translate the shipping order label image existing on the input image surface to the outbound image surface, and overlap the center point of the shipping order label image with the center point of the shipping order label image on the outbound image surface, and identify whether the two sets of shipping order label images completely overlap: if they completely overlap (complete overlap here means that each area overlaps, and there is no missing area or wrinkle. When a certain order is missing a part, it is an overlap, but not a complete overlap), it means that the shipping order label of the abnormal order is normal, and a packaging abnormality signal about the abnormal order is directly generated and displayed (the order is normal, that is, the packaging is deformed by impact, etc.);

[0050] If they do not completely overlap, confirm the standard feature value from the image in the library: Combined Figure 2 , calibrate the original shipping order label image inside the outbound image as a sub-image, calibrate the outbound image as a parent image, calibrate the four corner points of the sub-image as pending points, confirm the contour point closest to the corresponding pending point from the overall edge contour of the parent image, calibrate the determined contour point as the pending connection point corresponding to the pending point, connect the pending point with the pending connection point, divide the area between the sub-image and the parent image into four areas, and determine the area parameters of the four areas, randomly select a group of areas as the starting area and sort the area parameters in a clockwise order to determine the standard area sequence;

[0051] The shipping order label image moved above the outbound image is marked as a moving image, and the corner points of the moving image are determined, and the determined corner points are marked as pending points (if this label is missing or there are other situations, the number of corner points may not be 4, so the number of corner points is not limited here), and the pending points are connected with the associated pending points in the same way as the above-mentioned method of determining the pending points corresponding to the pending points, and the area between the child image and the parent image is divided, and the area parameters of each different area are confirmed, and a group of areas are randomly selected as the starting area and sorted in a clockwise manner to obtain a comparison area sequence;

[0052] S32. If the number of areas in the comparison area sequence is consistent with that in the standard area sequence, a proofreading process is performed: whether the area parameter at the same position in the comparison area sequence is consistent with that in the standard area sequence:

[0053] If they are consistent, it means that the shipping order label of this abnormal order is normal, and a packaging abnormality signal about this abnormal order is directly generated and displayed. The proposed standard area sequence is {10, 15, 20, 25}, and its comparison area sequence is {10, 15, 20, 25}, then the two are completely consistent;

[0054] If they are not consistent, the first group of area parameters in the comparison area sequence will be swapped to the last group, and then the area parameters at the same position will be identified to be consistent. If they are still inconsistent, the swap will continue. When the swap is repeated, it will stop to identify whether there is a swap process to determine whether the area parameters at the same position in the area sequence are consistent with the standard area sequence. If so, it means that there is no abnormality in the shipping order label of this abnormal order, and a packaging abnormality signal for this abnormal order is directly generated and displayed. If not, it means that the shipping order label of this abnormal order is abnormal, and an order label abnormality signal for this abnormal order is directly generated and displayed to formulate a standard area. The product sequence is {10, 15, 20, 25}, and its comparison area sequence is {20, 25, 10, 15}, then the two are completely consistent. If the first comparison is inconsistent, the first group of area parameters 20 in the comparison area sequence is swapped to the last group, and it changes to {25, 10, 15, 20}. In this comparison, it is still inconsistent, so it is changed again, and 25 is swapped to the last group to get {10, 15, 20, 25}. Then the current adjusted comparison area sequence is consistent with the standard area sequence, and this type of comparison process is the corresponding replacement process for consistent comparison;

[0055] S33. If the number of areas in the comparison area sequence is inconsistent with the number of areas in the standard area sequence, it means that the shipping order label of the abnormal order is abnormal, and an order label abnormality signal about the abnormal order is directly generated and displayed;

[0056] Specifically, the above-mentioned comparison and verification analysis method can not only analyze whether there are abnormalities in the corresponding order label, but also fully analyze whether there are relevant problems in the corresponding product packaging. When the corresponding order is entered, the front and rear directions are in opposite states. This order is also in a normal fitting state, but when the two overlap, the overlapping features generated will have relevant abnormalities. Therefore, by using this feature point and area confirmation method, it is possible to fully and effectively determine whether the area sequences of the two are completely consistent, so as to achieve better evaluation results.

[0057] An abnormal order monitoring and early warning system based on a cloud platform, comprising:

[0058] The outbound recording end collects the outbound image of each delivery order, and based on the collected outbound image, confirms the center point of the delivery order label inside the outbound image and the outbound image, locks and records the feature vector belonging to the delivery order, and simultaneously records the order number of the delivery order label;

[0059] The monitoring and judgment end performs feature analysis on the input images of the delivery orders taken by each transit warehouse, locks the feature vector associated with the input image, compares the confirmed feature vector with the standard vector, identifies whether the two sets of adjacent images completely overlap, and calibrates abnormal orders based on the identification results;

[0060] The abnormal determination end, based on the determined abnormal order, confirms the input image and the outbound image associated with the abnormal order when it is judged as abnormal, uses the outbound image as the standard template in the verification process, determines whether the shipping order label in the input image is abnormal, and displays the confirmed determination result.

[0061] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0062] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A cloud platform-based abnormal order monitoring and early warning method, characterized in that: The following steps are involved: Step 1: When each shipping order is generated and shipped out, collect the shipping order image, and confirm the feature vector of the shipping order from the collected shipping order image and record it. The shipping order image is the image of the side where the shipping order label is located. Step 2: The feature vector confirmed in the outbound image of the same delivery order is used as the standard vector, and the feature vector of the same delivery order is checked against the standard vector in the feature vectors confirmed for the same delivery order in other transit warehouses in the future, and the abnormal order is locked based on the check result; Step 3: Based on the determined abnormal order, the input image and the outbound image associated with the abnormal order are confirmed when the abnormal order is judged as abnormal. The outbound image is used as the standard template in the verification process to determine whether the shipping order label in the input image is abnormal, and the determination result is displayed. The specific sub-steps are: S31, directly translate the shipping order label image existing on the input image surface to the outbound image surface, and overlap the center point of the shipping order label image with the center point of the shipping order label image on the outbound image surface, and identify whether the two sets of shipping order label images completely overlap: If they completely overlap, it means that there is no abnormality in the shipping order label of this abnormal order, and a packaging abnormality signal about this abnormal order is directly generated and displayed; If they are not completely overlapped, the standard feature values ​​are confirmed from the outbound image: the original delivery order label image inside the outbound image is calibrated as a sub-image, the outbound image is calibrated as a parent image, the four corner points of the sub-image are calibrated as pending points, the contour point closest to the corresponding pending point is confirmed from the overall edge contour of the parent image, the determined contour point is calibrated as the pending connection point corresponding to the pending point, the pending point is connected with the pending connection point, the area between the sub-image and the parent image is divided into four areas, and the area parameters of the four areas are determined, a group of areas are randomly selected as the starting area and the area parameters are sorted in a clockwise manner to determine the standard area sequence; The shipping order label image moved above the outbound image is marked as a moving image, and the corner points of the moving image are determined, and the determined corner points are marked as pending points. The pending points are connected with the associated pending points in the same manner as the above method of determining the pending points corresponding to the pending points, and the area between the child image and the parent image is divided, and the area parameters of each different area are confirmed. A group of areas are randomly selected as the starting areas and are sorted in a clockwise order to obtain a comparison area sequence; S32. If the number of areas in the comparison area sequence is consistent with that in the standard area sequence, a proofreading process is performed: whether the area parameter at the same position in the comparison area sequence is consistent with that in the standard area sequence: If they are not consistent, the first group of area parameters in the comparison area sequence is swapped to the last group, and then it is identified whether the area parameters at the same position are consistent. If they are still inconsistent, the swapping is continued. When the swapping is repeated, it is stopped to identify whether there is a swapping process to determine whether the area parameters at the same position in the area sequence are consistent with the standard area sequence. If so, it means that there is no abnormality in the shipping order label of this abnormal order, and a packaging abnormality signal for this abnormal order is directly generated and displayed. If not, it means that the shipping order label of this abnormal order is abnormal, and an order label abnormality signal for this abnormal order is directly generated and displayed; S33. If the number of areas in the comparison area sequence is inconsistent with the number of areas in the standard area sequence, it means that the shipping order label of this abnormal order is abnormal, and an order label abnormality signal about this abnormal order is directly generated and displayed.

2. According to the cloud platform-based abnormal order monitoring and early warning method according to claim 1, it is characterized in that: In step 1, the specific method of confirming the feature vector of the delivery order is: S11. Confirm the internal shipping order label image from the collected outbound image, and based on the confirmed shipping order label image, calibrate the overall outline of the shipping order label image as the order outline, and place the order outline in a set of two-dimensional coordinate systems, and based on a number of different coordinate points associated with the order outline, determine the corresponding mean point, and calibrate the mean point in the order outline as the feature midpoint of the order outline; S12, confirming the overall image of the surface where the shipping order label is located from the outbound image, and determining the feature midpoint belonging to the overall image in the same manner as determining the feature midpoint of the order outline based on the overall edge contour of the overall image; S13. Generate and record a set of feature vectors starting from the feature midpoint of the overall image and ending at the feature midpoint of the order contour. Also record the order number of the corresponding delivery order, which is obtained from the delivery order label.

3. The abnormal order monitoring and early warning method based on a cloud platform according to claim 2 is characterized in that: In step 2, the specific sub-steps of locking abnormal orders are: S21, based on the order number of the shipping order entered by the corresponding transit warehouse, identifying the feature vector associated with the order number and marking it as a standard vector; S22, based on the input image of the shipping order, the shipping order label and the center point of the surface where the label is located are confirmed from the input image in the same manner as the outbound image, and based on the determined two sets of center points, a feature vector about the input image is determined starting from the center point of the surface where the label is located to the center point of the shipping order label; S23, compare the confirmed feature vector with the standard vector: make the feature vector coincide with the initial point of the standard vector, then make the feature vector rotate in a circle, with the fixed point of the circular rotation being the initial point of the feature vector, and identify whether the feature vector completely coincides with the standard vector during the rotation process: If there is a completely overlapping rotation process, it means that there is no abnormality in the transportation process of this delivery order and no processing is required; If there is no completely overlapping rotation process, it means that there is an abnormality in the transportation process of this delivery order, and this delivery order is marked as an abnormal order.

4. The abnormal order monitoring and early warning method based on a cloud platform according to claim 1 is characterized in that: In step S32, if the area parameters at the same position in the compared area sequence are consistent with the standard area sequence, it means that there is no abnormality in the shipping order label of this abnormal order, and a packaging abnormality signal for this abnormal order is directly generated and displayed.

5. A cloud platform-based abnormal order monitoring and early warning system, the early warning system is used to execute the cloud platform-based abnormal order monitoring and early warning method according to any one of claims 1 to 4, characterized in that: include: The outbound recording end collects the outbound image of each delivery order, and based on the collected outbound image, confirms the center point of the delivery order label inside the outbound image and the outbound image, locks and records the feature vector belonging to the delivery order, and simultaneously records the order number of the delivery order label; The monitoring and judgment end performs feature analysis on the input images of the delivery orders taken by each transit warehouse, locks the feature vector associated with the input image, compares the confirmed feature vector with the standard vector, identifies whether the two sets of adjacent images completely overlap, and calibrates abnormal orders based on the identification results; The abnormal determination end, based on the determined abnormal order, confirms the input image and the outbound image associated with the abnormal order when it is judged as abnormal, uses the outbound image as the standard template in the verification process, determines whether the shipping order label in the input image is abnormal, and displays the confirmed determination result.

Images