CN110379108A - Method and system for monitoring theft prevention of unmanned store - Google Patents

Abstract

The invention relates to the technical field of anti-theft monitoring, and specifically discloses an anti-theft monitoring method and system for unmanned stores, including performing preset processing and interconnecting with the unmanned store management system, obtaining static information resources of monitoring points, and comparing two adjacent Information resources to judge whether there is a change; if there is a change, submit the recorded video to extract the change feature, and analyze and judge the change feature; if it is an abnormal change, lock the behavior object and start the anti-theft early warning process, so as to pass Image static analysis can accurately identify whether the situation in the unmanned store has changed. If there is a change, the recorded video of the time period will be submitted for dynamic analysis. There is no need to analyze the entire recorded video, and it is targeted for changes. Perform dynamic identification and analysis in the time period to improve identification efficiency, and lock abnormally changing behavior objects to facilitate background monitoring, and can carry out anti-theft early warning processing to prevent problems before they happen.

Description

A method and system for anti-theft monitoring of an unmanned store

technical field

The invention relates to the technical field of anti-theft monitoring, and specifically discloses an anti-theft monitoring method and system for an unmanned store.

Background technique

With the development of science and technology, people's shopping methods have become more and more diverse, and unmanaged stores have gradually appeared on the market. The emergence of unmanned stores has brought convenience to people. The store owner does not need special management, and shoppers can enter the unmanned store for shopping at any time within 24 hours a day, and can pay independently without queuing.

At present, the common unmanned store anti-theft monitoring system generally installs a camera in the unmanned store, and then manually monitors the unmanned store in the background, and calls the police when someone steals goods. However, the current anti-theft monitoring system for unmanned stores is not perfect and relies heavily on manual monitoring in the background. Not only can it not prejudge the suspected behavior, but it is also prone to untimely discovery of theft, which brings a lot of trouble to the owner of the unmanned store. loss.

Therefore, there is a need for a solution that can solve the above problems.

Contents of the invention

In order to overcome the shortcomings and deficiencies in the prior art, an object of the present invention is to provide a method and system for anti-theft monitoring of an unmanned store.

In order to achieve the above object, the present invention adopts the following solutions.

A method for anti-theft monitoring of an unmanned store, comprising:

Preset the monitoring points in the unmanned store space, and interconnect with the unmanned store management system;

Obtain the static information resources of the monitoring point at all times, and compare the information resources obtained twice adjacent to determine whether the situation in the unmanned store corresponding to the monitoring point has changed;

If there is a change, submit the recorded video of the monitoring point to extract the change feature, and analyze and judge the extracted change feature;

If it is an abnormal change, the behavior object is locked, and the anti-theft warning processing is started.

Further, the preset processing of monitoring points in the unmanned store space and interconnection with the unmanned store management system include:

Divide the space in the unmanned store into several areas, and set up several monitoring points in each area;

Number each area, establish a list of monitoring points in each numbered area, and associate the list of monitoring points with monitoring point resources;

Connect each monitoring point to the unmanned store management system through relevant protocols.

Further, the static information resources of the monitoring point are obtained at the time, and the information resources obtained by two adjacent acquisitions are compared to determine whether the situation in the unmanned store corresponding to the monitoring point has changed, including:

Obtain static images of monitoring points at all times, and use CNN convolutional neural network to identify static images;

Determine whether the static images of two adjacent frames have changed;

If there is a change, the relevant recorded video of the monitoring point is intercepted for submission; if there is no change, the image and recorded video of the monitoring point in the period of time when there is no change are discarded.

Further, the use of CNN convolutional neural network to identify static images includes:

Determine the convolution kernel suitable for the static image by continuously setting different convolution kernels, use the adapted convolution kernel to initially extract the features of the image, and determine the output value of the image in the convolution layer;

Send the output value of the convolutional layer to the pooling layer for pooling to remove image noise and retain the main features of the image;

Through the function of the fully connected layer, the features of each part of the image are summarized to generate a classifier for predictive recognition.

Further, the extracted change features of the recorded video submitted to the monitoring point include facial structure change features, body language change features, and behavior track change features.

Further, the extracted change features are analyzed and judged, including:

Perform big data analysis on the face change features and the face features in the face database to determine whether it is similar to the shape of the thief, including at least the position of the eyes, the curvature of the mouth corners, and the curvature of the eyebrows;

Perform big data analysis on the characteristics of body language changes and body language changes in the body language database to determine whether they are similar to the body language changes of the perpetrator of the theft, including at least whether the perpetrator's head is looking around and whether the limbs are curled up;

Judging whether the behavior track is similar to that of the thief, including at least whether the actor wanders within a certain range and stays in the commodity for more than a predetermined time;

If the characteristics of structural changes, body language changes, and behavioral trajectory changes are consistent with theft, it will be marked as abnormal changes.

Further, the locking of the behavior object and starting the anti-theft early warning processing include:

Set a lock frame for the behavior object that changes abnormally in the recorded video, and make the lock frame move with the behavior object;

Broadcast the reminder voice to the unmanned store through the voice system;

Send alert information to the mobile terminal of the behavior object.

Further, after the anti-theft early warning process is started, the anti-theft tracking and monitoring is performed on the abnormally changing behavior objects, specifically:

Identify whether the behavior object in the recorded video overlaps with the product area, and judge whether the product area changes after the behavior object leaves the product area;

If the product area changes, disconnect the mobile terminal of the behavior object from the unmanned store, and re-connect until it is recognized that the behavior object completes the payment.

The present invention also provides an anti-theft monitoring system for an unmanned store, including a server;

Servers include processors and storage devices;

a processor adapted to execute program instructions;

The storage device is suitable for storing program instructions, and the program instructions are suitable for being loaded and executed by the processor to realize the above-mentioned anti-theft monitoring method for unmanned stores.

The present invention also provides a mobile terminal, including:

a processor adapted to execute program instructions;

The storage device is suitable for storing program instructions, and the program instructions are suitable for being loaded and executed by the processor to realize the above-mentioned anti-theft monitoring method for unmanned stores.

Beneficial effects of the present invention: provide a method and system for anti-theft monitoring of unmanned stores, by presetting the monitoring points in the unmanned store space and interconnecting with the unmanned store management system, the monitoring points can be obtained at all times during monitoring For static information resources, compare the information resources obtained twice adjacently to determine whether the situation in the unmanned store corresponding to the monitoring point has changed; if there is a change, submit the recorded video of the monitoring point to extract the change features, and perform the extracted change features Analysis and judgment; if it is an abnormal change, lock the behavior object and start the anti-theft early warning process, so that through the static analysis of the image, it can accurately identify whether the situation in the unmanned store has changed, and if there is a change, the corresponding time period will be Dynamic analysis of recorded video submissions eliminates the need for full-process video analysis, which not only saves resources, but also conducts dynamic identification and analysis of changing time periods in a targeted manner, which can greatly improve identification efficiency and prevent abnormal Changing behavior objects are locked to facilitate background monitoring, and anti-theft warning processing can be performed on behavior objects, which can prevent problems before they happen.

Description of drawings

FIG. 1 is a schematic flowchart of an anti-theft monitoring method according to an embodiment of the present invention.

Detailed ways

In order to facilitate the understanding of those skilled in the art, the present invention will be further described below in conjunction with the embodiments and accompanying drawings, and the contents mentioned in the implementation modes are not intended to limit the present invention.

A method for anti-theft monitoring of unmanned stores, which reduces the process of dynamic analysis through static image analysis and dynamic video analysis, so that targeted analysis can be carried out, and the efficiency of analysis and identification can be improved, so as to achieve better anti-theft monitoring effects , the specific process is shown in Figure 1, including:

In order to fully monitor the space in the unmanned store, it is necessary to pre-set the monitoring points in the unmanned store space and interconnect with the unmanned store management system. Preferably, the preset processing includes at least dividing the space in the unmanned store into several areas, and setting several monitoring points in each area. Each monitoring point can be aimed at different positions of the area plane, or each monitoring point can be Aim at different heights at the same position, so as to realize the monitoring of horizontal space and vertical space, and realize the comprehensive monitoring of the unmanned store. Then number each area, establish a list of monitoring points in each numbered area, and associate the list of monitoring points with the monitoring point resources, so as to facilitate the management of each monitoring point, and also facilitate manual resource information of a certain monitoring point Make an inquiry. After completing the setting of monitoring points, connect each monitoring point to the unmanned store management system through relevant protocols to realize real-time preview, video playback, voice intercom and other functions.

When monitoring an unmanned store, if the video is identified and analyzed throughout the entire process, the workload will be large, and it will occupy analysis resources, and then it is very likely that the analysis monitoring point has not changed during this period of time, resulting in a waste of resources. Wasteful and inefficient. Therefore, the present invention adopts a combination of static analysis and dynamic analysis. Specifically, the static information resources of the monitoring point are obtained at all times. The static information resources are mainly static images. The CNN convolutional neural network is used to identify the static images. Two frames of static images adjacent to each other) to judge whether the situation in the unmanned store corresponding to the monitoring point has changed, and the image recognized by the CNN convolutional neural network can quickly judge whether there is a change. If there is a change, it means that the unmanned store area corresponding to the monitoring point has moved. Therefore, the recorded video of this time period can be submitted for analysis to obtain the change characteristics. Before submitting, it is necessary to intercept the relevant recorded video of the monitoring point for submission preparation, such as Format conversion, marking the area number, monitoring point list, etc., to facilitate subsequent feature extraction analysis. If there is no change, it means that there is no flow of people at this monitoring point during this time period. Since there is no change, there is no need to analyze it. Therefore, the images and recorded videos of this monitoring point during the time period when there is no change do not need to be retained. The discarding process is performed, thereby releasing the storage space of the system, which is beneficial to storing more changed video recordings.

More specifically, the use of CNN convolutional neural networks to identify static images includes continuously setting different convolution kernels to determine the convolution kernels that are suitable for static images, using the adapted convolution kernels to initially extract the features of the image, and determining The output value of the image in the convolutional layer; the output value of the convolutional layer is sent to the pooling layer for pooling to remove image noise and retain the main features of the image; the features of each part of the image are processed through the function of the fully connected layer Summarize, generate a classifier, and perform prediction and recognition.

Then it is necessary to analyze and judge the extracted change features. The video extraction and change features submitted to the monitoring point include, but are not limited to, facial structure change features, body language change features, and behavior track change features. Here is not the traditional way of identifying the face and other characteristics of the behavior object to judge whether it is a repeat offender to determine the suspect, but to judge whether the changed characteristics are similar to the shape of the thief through big data analysis, so as to better deal with the crime. Behavior objects are controlled.

Specifically, big data analysis is performed on facial change features and face features in the face database to determine whether it is similar to the shape of the thief, including at least eye position, mouth curvature, and eyebrow curvature; body language change features Carry out big data analysis with the characteristics of body language changes in the body language database to determine whether it is similar to the body language changes of the thief, including at least whether the person’s head is looking around and whether the body is curled up; judge whether the behavior trajectory is consistent with the theft The behavior track changes of people are similar, including at least whether the perpetrator wanders within a certain range and stays in the commodity for more than a predetermined time; for abnormal changes.

Of course, even if the behavior object conforms to the form of theft behavior, it can be used as the key object of attention, but the behavior object has not carried out the stealing behavior. The frame moves with the behavior object, so that it is convenient for the background personnel to know that the behavior object is in an abnormal state, and it is convenient to remind the background personnel to focus on it. At the same time, start the anti-theft early warning processing, such as broadcasting a reminder voice to the unmanned store through the voice system, which can remind other personnel in the unmanned store to pay attention to keeping their belongings. In addition, because the unmanned store adopts the method of scanning to enter the door, the unmanned store management system has been connected to the mobile terminal of the behavior object when entering the door, and can send an alert message to the mobile terminal of the behavior object, which can be used without causing panic among other store personnel. Alerting the behavior object can make the behavior object voluntarily give up the stealing behavior.

In order to prevent the behavior object from still carrying out theft after being alerted, after the anti-theft early warning process is started, anti-theft tracking and monitoring is carried out on the abnormally changing behavior object. Specifically, identify whether the behavior object in the recorded video overlaps with the commodity area, and judge whether the commodity area changes after the behavior object leaves the commodity area; if the commodity area changes, disconnect the behavior object’s mobile terminal from the unmanned store door , that is, the behavior object cannot open the door of the unmanned store through scanning, and re-connects until the behavior object is recognized and completes the payment. After re-connecting, the behavior object can open the door of the unmanned store and leave. In this way, the object of behavior can be prevented from escaping from the unmanned store after stealing, which ensures the safety of the goods in the unmanned store and avoids the loss of the owner. It can also be added to identify the type of product and its corresponding price, so as to prevent the behavior object from entering a price after stealing and then opening the door of the unmanned store to escape.

The method for anti-theft monitoring of an unmanned store provided by the present invention can accurately identify whether the situation in the unmanned store has changed through the static analysis of the image, and if there is a change, the recorded video of the corresponding time period will be submitted for dynamic analysis, thus eliminating the need for The analysis of the whole process of video and video can not only save resources, but also carry out dynamic identification and analysis on the changing time period in a targeted manner, which can greatly improve the identification efficiency, and lock the abnormally changing behavior objects to facilitate background monitoring , It can also carry out anti-theft early warning processing on the behavior object, which can prevent problems before they happen.

The present invention also provides an unmanned shop anti-theft monitoring system, including a server;

Servers include processors and storage devices;

a processor adapted to execute program instructions;

The storage device is suitable for storing program instructions, and the program instructions are suitable for being loaded and executed by the processor to realize the above-mentioned anti-theft monitoring method for unmanned stores.

The present invention provides a mobile terminal, including:

a processor adapted to execute program instructions;

The storage device is suitable for storing program instructions, and the program instructions are suitable for being loaded and executed by the processor to realize the above-mentioned anti-theft monitoring method for unmanned stores.

The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention is disclosed as above with preferred embodiments, it is not intended to limit the present invention. Anyone familiar with this field , without departing from the scope of the technical solution of the present invention, when using the technical content disclosed above to make some changes or modifications to equivalent embodiments with equivalent changes, but as long as it does not depart from the technical solution of the present invention, the technology of the present invention refers to the above Any simple modifications, equivalent changes and modifications made in the embodiments all belong to the scope of the technical solution of the present invention.

Claims (10)

1. A method for unmanned store anti-theft monitoring, characterized in that, comprising: Preset the monitoring points in the unmanned store space, and interconnect with the unmanned store management system; Obtain the static information resources of the monitoring point at all times, and compare the information resources obtained twice adjacent to determine whether the situation in the unmanned store corresponding to the monitoring point has changed; If there is a change, submit the recorded video of the monitoring point to extract the change feature, and analyze and judge the extracted change feature; If it is an abnormal change, the behavior object is locked, and the anti-theft warning processing is started. 2. The method for anti-theft monitoring of an unmanned store according to claim 1, wherein the preset processing of monitoring points in the unmanned store space and interconnection with the unmanned store management system include: Divide the space in the unmanned store into several areas, and set up several monitoring points in each area; Number each area, establish a list of monitoring points in each numbered area, and associate the list of monitoring points with monitoring point resources; Connect each monitoring point to the unmanned store management system through relevant protocols. 3. The method for anti-theft monitoring of a kind of unmanned store according to claim 1, characterized in that, the static information resources of the monitoring point are obtained at the time, and the information resources obtained twice adjacently are compared to determine the unmanned location corresponding to the monitoring point. Whether the situation in the store has changed, including: Obtain static images of monitoring points at all times, and use CNN convolutional neural network to identify static images; Determine whether the static images of two adjacent frames have changed; If there is a change, the relevant recorded video of the monitoring point is intercepted for submission; if there is no change, the image and recorded video of the monitoring point in the period of time when there is no change are discarded. 4. The method for anti-theft monitoring of a kind of unmanned shop according to claim 3, is characterized in that, described use CNN convolutional neural network to identify static image, comprising: Determine the convolution kernel suitable for the static image by continuously setting different convolution kernels, use the adapted convolution kernel to initially extract the features of the image, and determine the output value of the image in the convolution layer; Send the output value of the convolutional layer to the pooling layer for pooling to remove image noise and retain the main features of the image; Through the function of the fully connected layer, the features of each part of the image are summarized to generate a classifier for predictive recognition. 5. The method for anti-theft monitoring of a kind of unmanned shop according to claim 1, is characterized in that, the recording video extraction change feature of described submission monitoring point comprises facial structure change feature, body language change feature and behavior track change feature . 6. The method for anti-theft monitoring of a kind of unmanned store according to claim 5, characterized in that, analyzing and judging the extracted change characteristics, including: Perform big data analysis on the face change features and the face features in the face database to determine whether it is similar to the shape of the thief, including at least the position of the eyes, the curvature of the mouth corners, and the curvature of the eyebrows; Perform big data analysis on the characteristics of body language changes and body language changes in the body language database to determine whether they are similar to the body language changes of the perpetrator of the theft, including at least whether the perpetrator's head is looking around and whether the limbs are curled up; Judging whether the behavior track is similar to that of the thief, including at least whether the actor wanders within a certain range and stays in the commodity for more than a predetermined time; If the characteristics of structural changes, body language changes, and behavioral trajectory changes are consistent with theft, it will be marked as abnormal changes. 7. The method for anti-theft monitoring of a kind of unmanned store according to claim 1, wherein the locking of the behavior object and starting anti-theft early warning processing include: Set a lock frame for the behavior object that changes abnormally in the recorded video, and make the lock frame move with the behavior object; Broadcast the reminder voice to the unmanned store through the voice system; Send alert information to the mobile terminal of the behavior object. 8. The method for anti-theft monitoring of a kind of unmanned shop according to claim 1, is characterized in that, after starting the anti-theft early warning processing, carry out anti-theft tracking monitoring to the abnormally changing behavior object, specifically: Identify whether the behavior object in the recorded video overlaps with the product area, and judge whether the product area changes after the behavior object leaves the product area; If the product area changes, disconnect the mobile terminal of the behavior object from the unmanned store, and re-connect until it is recognized that the behavior object completes the payment. 9. A system for anti-theft monitoring in an unmanned store, characterized in that it includes a server; Servers include processors and storage devices; a processor adapted to execute program instructions; The storage device is suitable for storing program instructions, and the program instructions are suitable for being loaded and executed by the processor to realize the anti-theft monitoring method for unmanned stores according to any one of claims 1 to 8. 10. A mobile terminal, characterized in that, comprising: a processor adapted to execute program instructions; The storage device is suitable for storing program instructions, and the program instructions are suitable for being loaded and executed by the processor to realize the anti-theft monitoring method for unmanned stores according to any one of claims 1 to 8.