CN114898302A

CN114898302A - Legacy monitoring method, device, computer equipment and storage medium

Info

Publication number: CN114898302A
Application number: CN202210647246.2A
Authority: CN
Inventors: 刘海龙; 刘西彦; 刘彤; 韩耘
Original assignee: Industrial and Commercial Bank of China Ltd ICBC
Current assignee: Industrial and Commercial Bank of China Ltd ICBC
Priority date: 2022-06-09
Filing date: 2022-06-09
Publication date: 2022-08-12

Abstract

The application relates to a method and a device for monitoring a remnant, computer equipment, a storage medium and a computer program product, relates to the technical field of artificial intelligence, and can be used in the field of financial technology or other fields. The method comprises the following steps: acquiring a first shot image of a target environment; according to the background model, carrying out moving foreground target detection on the first shot image to obtain connected domain blocks of all moving foreground targets; carrying out pedestrian detection on the image of each connected domain block to obtain a pedestrian connected domain block of the target pedestrian; in the next frame of shot image, identifying whether the preset area range of the target pedestrian contains the object connected domain block mass, and if so, correspondingly storing the face image of the target pedestrian and the object image of the object connected domain block mass; and if the position information of the article connected domain block mass in the subsequent shot image is not changed, determining the article corresponding to the article connected domain block mass as the remnant of the target pedestrian. The method can improve the timeliness of finding the remnant.

Description

Legacy monitoring method, device, computer equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to a method, an apparatus, a computer device, a storage medium, and a computer program product for monitoring a legacy.

Background

The lobby of bank branch is at the first line position of service client, and it features large daily flow of people and frequent client access. In the process of frequent trips to a bank lobby location, customers typically carry items with them, such as handbags, backpacks, suitcases, and the like. Through the statistical investigation of the bank outlets, the event that a customer forgets to carry articles with him occurs when the customer leaves a hall of the bank outlet after handling the business.

At present, the discovery process and the processing process of a bank to a client left article are generally as follows: the bank staff regularly or irregularly checks the articles in the branch hall, if the articles left by the customers are picked up, the articles are safely checked, the dangerous articles are removed and then are temporarily stored by the customers, and then the owner of the bank is waited to claim the left articles. However, this method has a problem that the discovery of the carryover is not timely.

Disclosure of Invention

In view of the above, there is a need to provide a method, an apparatus, a computer device, a computer readable storage medium, and a computer program product for detecting a legacy capable of improving the timeliness of discovery of the legacy.

In a first aspect, the present application provides a carryover monitoring method. The method comprises the following steps:

acquiring a first shot image of a target environment;

according to a pre-established background model, carrying out moving foreground target detection on a collected first shot image by adopting a moving target detection algorithm to obtain connected domain lumps corresponding to all moving foreground targets in the first shot image;

carrying out pedestrian detection on the image corresponding to each connected domain block to obtain a pedestrian connected domain block corresponding to a target pedestrian;

identifying whether an article connected domain block is contained in a preset region range corresponding to the target pedestrian in a next frame of shooting image of the first shooting image;

under the condition that the object connected domain cluster is included, acquiring a face image corresponding to the target pedestrian and an object image corresponding to the object connected domain cluster, and storing the face image and the object image correspondingly;

and if the position information of the object connected domain block in the shot image after the next frame of shot image is not changed, determining the object corresponding to the object connected domain block as the remnant of the target pedestrian.

In one embodiment, the process of establishing the background model includes:

acquiring a second shot image of the target environment as a sample image;

determining a target sample image according to the sample image;

for each pixel point in each target sample image, randomly selecting a pixel point value of one pixel point from pixel points in a preset neighborhood of the pixel point in the target sample image as a sample pixel value of the pixel point in the target sample image;

performing weighted average calculation according to the sample pixel values of the pixel points in each target sample image to obtain a background sample library pixel value of the pixel points;

and returning to the step of determining the target sample image according to the sample image until the background sample library pixel values of the preset number of each pixel point in the sample image are obtained, and establishing the background model of the target environment based on the background sample library pixel values of the preset number of each pixel point in the sample image.

In one embodiment, the performing weighted average calculation according to the sample pixel values of the pixel points in each of the target sample images to obtain a background sample library pixel value of the pixel point includes:

carrying out mean value calculation on the sample pixel values of the pixel points in each target sample image to obtain a mean value;

calculating the weight of the pixel points in each target sample image by adopting a one-dimensional Gaussian distribution function according to the sample pixel values of the pixel points in each target sample image and the average value, and determining the weight of the average value according to the weight of the pixel points in each target sample image;

and performing weighted average calculation according to the sample pixel value of the pixel point in each target sample image, the mean value, the weight of the pixel point in each target sample image and the weight of the mean value to obtain a background sample library pixel value of the pixel point.

In one embodiment, if the position information of the article connected domain block in the captured image after the next captured image is not changed, determining the article corresponding to the article connected domain block as the carry-over object of the target pedestrian includes:

determining position information of the article connected domain block in the next frame of shot image and in shot images after the next frame of shot image;

calculating the position distance of the article connected domain block mass in the two adjacent frames of shot images according to the position information of the article connected domain block mass in the two adjacent frames of shot images;

if the position distance is smaller than or equal to a preset threshold value, recording that the number of times of stillness of the article connected domain block is increased once;

and when the number of times of stillness of the article connected domain block mass is greater than the preset number of times, determining the article corresponding to the article connected domain block mass as the remnant of the target pedestrian.

In one embodiment, the method further comprises:

and according to the face image corresponding to the target pedestrian, identifying the identity of the target pedestrian, determining contact information of the target pedestrian, and sending a carry-over object getting prompt message to the target pedestrian according to the contact information.

In one embodiment, the method further comprises:

acquiring a face image of a person to be picked aiming at a target remnant;

and matching and identifying the face image of the person to be picked with the face image of the pedestrian corresponding to the target object, and outputting prompt information that the identity check is passed if the matching is successful.

In one embodiment, the method further comprises:

receiving a human face image of a retriever aiming at a target legacy sent by a legacy retriever;

and matching and identifying the face image of the retriever and the face image of the pedestrian corresponding to the target remnant, and sending a throwing instruction aiming at the target remnant to the remnant retriever under the condition of successful matching so that the target remnant is thrown to the retriever by the remnant retriever according to the throwing instruction.

In a second aspect, the present application further provides a carryover monitoring apparatus. The device comprises:

the first acquisition module is used for acquiring a first shot image of a target environment;

the first detection module is used for detecting a moving foreground target of the collected first shot image by adopting a moving target detection algorithm according to a pre-established background model to obtain a connected domain block corresponding to each moving foreground target in the first shot image;

the second detection module is used for carrying out pedestrian detection on the image corresponding to each connected domain block mass to obtain a pedestrian connected domain block mass corresponding to the target pedestrian;

the first identification module is used for identifying whether the object pedestrian contains an article connected domain block in a preset area range corresponding to the target pedestrian in a next frame of shooting image of the first shooting image;

the storage module is used for acquiring a face image corresponding to the target pedestrian and an article image corresponding to the article connected domain cluster under the condition that the article connected domain cluster is included, and storing the face image and the article image correspondingly;

a first determining module, configured to determine, if the position information of the article connected domain blob in the captured image after the next frame of captured image is not changed, an article corresponding to the article connected domain blob as a carry-over of the target pedestrian.

In one embodiment, the apparatus further comprises:

the second acquisition module is used for acquiring a second shot image of the target environment as a sample image;

a second determining module, configured to determine a target sample image according to the sample image;

a selecting module, configured to randomly select, for each pixel point in each target sample image, a pixel value of a pixel point from pixel points in a preset neighborhood of the pixel point in the target sample image, where the pixel value is used as a sample pixel value of the pixel point in the target sample image;

the calculation module is used for performing weighted average calculation according to the sample pixel values of the pixel points in each target sample image to obtain a background sample library pixel value of the pixel points;

and the establishing module is used for returning and executing the step of determining the target sample image according to the sample image until the background sample library pixel values of the preset number of each pixel point in the sample image are obtained, and establishing the background model of the target environment based on the background sample library pixel values of the preset number of each pixel point in the sample image.

In one embodiment, the calculation module is specifically configured to:

carrying out mean value calculation on the sample pixel values of the pixel points in each target sample image to obtain a mean value; calculating the weight of the pixel points in each target sample image by adopting a one-dimensional Gaussian distribution function according to the sample pixel values of the pixel points in each target sample image and the mean value, and determining the weight of the mean value according to the weight of the pixel points in each target sample image; and performing weighted average calculation according to the sample pixel value of the pixel point in each target sample image, the mean value, the weight of the pixel point in each target sample image and the weight of the mean value to obtain a background sample library pixel value of the pixel point.

In one embodiment, the first determining module is specifically configured to:

determining position information of the article connected domain block in the next frame of shot image and in shot images after the next frame of shot image; calculating the position distance of the article connected domain block mass in the two adjacent frames of shot images according to the position information of the article connected domain block mass in the two adjacent frames of shot images; if the position distance is smaller than or equal to a preset threshold value, recording that the number of times of stillness of the article connected domain block is increased once; and when the number of times of stillness of the article connected domain block mass is greater than the preset number of times, determining the article corresponding to the article connected domain block mass as the remnant of the target pedestrian.

In one embodiment, the apparatus further comprises a second identification module configured to:

In one embodiment, the apparatus further comprises:

the third acquisition module is used for acquiring a face image of a person to be picked aiming at the target legacy;

and the matching module is used for matching and identifying the face image of the person to be picked and the face image of the pedestrian corresponding to the target remnant, and outputting prompt information for passing identity check if matching is successful.

In one embodiment, the apparatus further comprises:

the receiving module is used for receiving the human face image of the retriever aiming at the target legacy sent by the legacy retriever;

and the sending module is used for matching and identifying the face image of the retriever with the face image of the pedestrian corresponding to the target remnant, and sending a throwing instruction aiming at the target remnant to the remnant retriever under the condition of successful matching, so that the target remnant is thrown to the retriever by the remnant retriever according to the throwing instruction.

In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the steps of the method of the first aspect when executing the computer program.

In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method of the first aspect.

In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program that, when executed by a processor, performs the steps of the method of the first aspect.

According to the method, the device, the computer equipment, the storage medium and the computer program product for monitoring the carry-over object, the moving foreground object detection is carried out on the first shot image of the target environment by adopting a moving object detection algorithm according to a pre-established background model to obtain the connected domain blocks corresponding to all the moving foreground objects, then the pedestrian detection is carried out on all the connected domain blocks to determine the pedestrian connected domain blocks corresponding to the target pedestrians, namely to determine that the moving pedestrians exist in the target environment, and further to identify whether the connected domain blocks (namely the object connected domain blocks) of the non-pedestrians are included in the preset region range of the pedestrians in the next frame shot image, if so, the pedestrians place the carry-on object aside object between the moment of the first shot image and the moment of the next frame shot image, so that the object connected domain blocks appear around the pedestrians (the preset region range), therefore, the corresponding relation between the face image of the pedestrian and the article image corresponding to the article connected domain block can be established. Then, whether the position information of the object connected domain block is changed or not is monitored in a subsequent shot image, if the position information of the object connected domain block is not changed, the object corresponding to the object connected domain block is determined as a left object, and a face image of a pedestrian corresponding to the object image is determined as a face image of a loser of the left object. Therefore, whether the object is left over or not in the target environment can be monitored in real time and automatically, and the timeliness of finding the object is improved.

Drawings

FIG. 1 is a schematic flow chart diagram of a carryover monitoring method in one embodiment;

FIG. 2 is a flow diagram illustrating the process of building a background model in one embodiment;

FIG. 3 is a flow diagram illustrating a process for computing background sample library pixel values in one embodiment;

FIG. 4 is a schematic flow chart of a carryover monitoring method in another embodiment;

FIG. 5 is a schematic diagram of a legacy retriever in one example;

FIG. 6 is a flow diagram of a method of carryover pickup in one example;

FIG. 7 is a block diagram of a carryover monitoring apparatus in one embodiment;

FIG. 8 is a diagram illustrating an internal structure of a computer device according to an embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.

First, before specifically describing the technical solution of the embodiment of the present application, a technical background or a technical evolution context on which the embodiment of the present application is based is described. When the client leaves a hall of a bank outlet after handling the business, the event of forgetting to carry articles with the client occurs. At present, the discovery process and the processing process of a bank to a client left article are generally as follows: the bank staff regularly or irregularly checks the articles in the branch hall, if the articles left by the customers are picked up, the articles are safely checked, the dangerous articles are removed and then are temporarily stored by the customers, and then the owner of the bank is waited to claim the left articles. However, this method has a problem that the discovery of the carryover is not timely. On one hand, for normal remnants of non-dangerous goods, if the normal remnants are not found in time, the remnants may be lost; on the other hand, if dangerous goods left over cannot be found in time, alarm processing cannot be performed in time, and sudden safety incidents are easy to happen. Moreover, the method needs to consume a large amount of labor cost, and the labor processing efficiency is low, so that the customer experience is poor. Based on the background, the applicant provides the method for monitoring the abandoned object through long-term research and development and experimental verification, whether the abandoned object exists in the target environment can be monitored in real time and automatically, the timeliness of finding the abandoned object is improved, the labor consumption can be effectively saved, the processing efficiency of the abandoned object is improved, and the customer experience is improved. In addition, it should be noted that the applicant has paid a lot of creative efforts in finding the technical problems of the present application and the technical solutions described in the following embodiments.

In one embodiment, as shown in fig. 1, a method for monitoring a legacy is provided, and this embodiment is illustrated by applying the method to a server, and it is to be understood that the method may also be applied to a server, and may also be applied to a system including a terminal and a server, and is implemented by interaction between the terminal and the server.

In this embodiment, the method includes the steps of:

step 101, a first captured image of a target environment is acquired.

The target environment refers to an environment where the legacy monitoring is required, and may be an indoor environment (such as a bank branch hall, etc.) or an outdoor environment (such as an outdoor park, etc.), and the present application is not limited. The first captured image may be a captured image acquired at any time during the process of performing the carryover monitoring on the target environment. It can be understood that the first captured image may be a captured image of a whole area of the target environment, or may be a captured image of a local area where only the left-over object needs to be monitored in a focused manner, and may be specifically set according to an actual situation.

In an implementation, the server may capture a first captured image of the target environment, for example, the captured image of the target environment may be captured by a camera installed in the target environment, and the captured image may be transmitted to the server in real time.

And 102, detecting moving foreground targets of the collected first shot image by adopting a moving target detection algorithm according to a pre-established background model to obtain connected domain blocks corresponding to all the moving foreground targets in the first shot image.

The background model is established for the target environment and can be used for describing background image information of relatively static articles, facilities and the like in the target environment. The background model may be established before the moving foreground object detection is performed on the first captured image (i.e., the current frame captured image), and the specific establishment method may be a method of establishing a background model in a moving object detection algorithm (e.g., a vibe algorithm), or may be an improved method, which will be described in detail later, and will not be described herein again.

In implementation, the server may perform moving foreground object detection on the acquired first captured image by using a moving object detection algorithm according to a pre-established background model. The moving foreground target can be a pedestrian, a cart, a trunk or the like which moves relative to the background of the target environment. The server may perform moving foreground object detection on the first captured image by using a foreground segmentation method in a moving object detection algorithm (e.g., a vibe algorithm) based on the background model to obtain a foreground image. The foreground image is a binary image corresponding to the first captured image, wherein a region corresponding to the moving foreground object may be white, and a background portion except for the moving foreground object may be black. If the first shot image contains a plurality of moving foreground objects, the foreground image contains a plurality of connected domain blobs (namely white blobs), and each connected domain blob corresponds to one moving foreground object. Therefore, connected domain blobs corresponding to all moving foreground objects in the first shot image can be obtained.

And 103, carrying out pedestrian detection on the image corresponding to each connected domain block to obtain a target connected domain block corresponding to the target pedestrian.

The image corresponding to each connected domain blob is an image (RGB image) of a region corresponding to each connected domain blob in the first captured image.

In an implementation, the server may detect pedestrians in the images corresponding to the connected domain blobs, for example, the server may input the images corresponding to the connected domain blobs into an MB-CLBP-ADABOOST pedestrian detector (a pedestrian detector based on a plurality of Local Binary features and a machine learning classifier) to determine whether the images are pedestrians. If it is detected that the image corresponding to a certain connected domain block contains a pedestrian, the connected domain block can be determined as a pedestrian connected domain block, and the pedestrian in the image is the target pedestrian, so that the pedestrian connected domain block corresponding to the target pedestrian is obtained.

And 104, identifying whether the preset area range corresponding to the target pedestrian contains the article connected domain block mass or not in the next frame of shot image of the first shot image.

The article connected domain block mass refers to a connected domain block mass corresponding to a non-pedestrian.

In implementation, the server may identify whether the object connected domain blob is included in the preset region range corresponding to the target pedestrian in the captured image of the next frame of the first captured image. For example, the server may perform motion foreground object detection on the next captured image to obtain a foreground image of the next captured image, and then perform pedestrian detection on images corresponding to connected domain blobs included in the foreground image to determine a pedestrian connected domain blob and an article connected domain blob (i.e., a connected domain blob corresponding to a non-pedestrian). Then, the server may determine, according to the target position information (e.g., the centroid position) of the pedestrian connected domain blob corresponding to the target pedestrian in the first captured image, the pedestrian connected domain blob closest to the target position information from the pedestrian connected domain blobs in the next captured image, that is, determine the connected domain blob corresponding to the target pedestrian in the foreground image of the next captured image. Then, the server may determine whether there is an article connected domain blob in the foreground image of the next frame of captured image within the preset region range of the target pedestrian, for example, may set a range centered on the connected domain blob corresponding to the target pedestrian in the foreground image of the next frame of captured image, and determine whether the article connected domain blob is included in the foreground image within the range. The specific range size may be set empirically or experimentally and in practice, for example, a circular range including the connected domain blob may be set around the centroid of the connected domain blob, a rectangular range including the connected domain blob may be set, or a range of the connected domain blob expanded by a predetermined multiple may be set.

It can be understood that if a plurality of moving pedestrians are included in the first captured image, then pedestrian connected domain masses corresponding to a plurality of target pedestrians can be obtained, and whether an article connected domain mass is included in the preset region range of each target pedestrian is identified in the next captured image.

And 105, acquiring a face image corresponding to the target pedestrian and an article image corresponding to the article connected domain cluster under the condition that the article connected domain cluster is included, and correspondingly storing the face image and the article image.

In implementation, if the server identifies that the preset area range corresponding to the target pedestrian includes the article connected domain blob in the next captured image in step 104, the server may obtain the face image corresponding to the target pedestrian and the article image corresponding to the article connected domain blob, and store the face image and the article image correspondingly. For example, the server may intercept the face image of the target pedestrian in the corresponding region of the target pedestrian (i.e., the image region corresponding to the pedestrian connected domain blob of the target pedestrian) in the next captured image, and intercept the article image in the corresponding region of the article connected domain blob, and store the two correspondingly. It can be understood that the face image of the target pedestrian can be intercepted in the first shot image, and the face image of the target pedestrian can be intercepted.

It can be understood that, if the server recognizes that the preset region range corresponding to the target pedestrian does not contain the object connected domain blob in the next frame of captured image, the server may continue to detect and recognize the captured image after the next frame of captured image until the preset region range corresponding to the target pedestrian is recognized in a certain frame of captured image contains the object connected domain blob, and may perform step 105 on the object connected domain blob; or stopping the object pedestrian carrying out the object connected domain block monitoring until the object pedestrian is not included in a certain frame of shot image, which indicates that the object pedestrian leaves the object environment at the moment. So as to realize real-time and continuous monitoring on each target pedestrian in the target environment until the pedestrian leaves the target environment.

And step 106, if the position information of the article connected domain block mass in the shot image after the next frame of shot image is not changed, determining the article corresponding to the article connected domain block mass as the remnant of the target pedestrian.

In an implementation, the server may monitor whether the position information of the article connected domain block changes in the captured image after the next captured image, and for convenience of subsequent description, the article corresponding to the article connected domain block may be referred to as a target article. In one example, the method for determining whether the location information changes is specifically that the server may record the location information (e.g., the centroid location, which may be represented as (x) of the connected object domain blob in the next captured image (which may be denoted as the t-th captured image) of the connected object domain blob _t ，y _t ) Then, the server may perform moving foreground object detection on a next frame captured image (i.e., the t +1 th frame captured image) of the next frame captured image, determine an article connected domain blob in the foreground image of the t +1 th frame captured image, and then find an article connected domain blob corresponding to the target article in the foreground image of the t +1 th frame captured image (which may be determined by a position, a shape, or other methods), that is, may determine position information (e.g., a centroid position, which may be expressed as (x) of the article connected domain blob in the t +1 th frame captured image _t+1 ，y _t+1 )). Then, the server can obtain the position information (x) of the object connected domain block in the t frame shooting image according to the object _t ，y _t ) And position information (x) of the connected article region blob in the t +1 th frame captured image _t+1 ，y _t+1 ) The distance (which may be referred to as dist) between the two position information is calculated as follows:

where dist (t +1) represents a distance between the position information of the article connected domain blob corresponding to the target article in the t +1 th frame captured image and the position information in the t th frame captured image. The server may then compare the distance dist (T +1) to a preset threshold (which may be denoted as T) _dist ) Comparing if dist (T +1) is less than or equal to the preset threshold T _dist And if so, determining that the position information of the object connected domain block corresponding to the target object in the t +1 th frame of the shot image is unchanged. If the server determines that the object connected domain cluster block corresponding to the target object has no position change in the subsequent preset number of captured images or has no position change in the subsequent captured images within a preset time period, the target object (i.e., the object corresponding to the object connected domain cluster block) may be determined as the carry-over object of the target pedestrian.

Further, after the server determines the left-over object of the target pedestrian, prompt information can be sent to the staff, so that the staff can safely check the left-over object in time, the left-over object is removed to be a dangerous article, the left-over object is stored or a left-over object extractor is put in, and the owner (namely the target pedestrian) is waited for and is extracted before.

Further, the server can also identify the face image of the target pedestrian corresponding to the remnant, and determine the identity information and the contact information of the owner (namely the target pedestrian) so as to timely inform the owner of picking up the remnant before.

It can be understood that the server may continuously acquire the shot image of the target environment in real time, and thus for a current frame (for example, a t-th frame) shot image, the server may use the t-th frame shot image as a first shot image and execute the processing procedures of steps 101 to 106, may also use the t-th frame shot image as a next frame shot image of a t-1-th frame shot image and execute the processing procedures of steps 104 to 105, and may also use the t-th frame shot image as a shot image after the next frame shot image of the t-2-th frame (and before) shot image and execute the processing procedure of step 106, which are not described herein again. The above-described processing procedure of the first captured image, the processing procedure of the next frame captured image as the t-1 th frame captured image, and the processing procedure of the captured image after the next frame captured image as the t-2 th frame (and before) captured image do not differ in order.

In the method for monitoring the remnant, the moving foreground object detection is carried out on the first shot image of the object environment by adopting a moving object detection algorithm according to a pre-established background model to obtain connected domain blocks corresponding to all the moving foreground objects, then carrying out pedestrian detection on the connected domain masses, determining pedestrian connected domain masses corresponding to the target pedestrians, namely, the moving pedestrian in the target environment is determined, whether the pedestrian in the preset area range contains the object connected domain block is further identified in the next frame of shot image, if yes, the object connected domain block is explained between the moment of the first shot image and the moment of the next frame of shot image, the pedestrian places the personal belongings aside, so that an object connected domain block appears around the pedestrian (in a preset area range), and a corresponding relation can be established between the face image of the pedestrian and the object image corresponding to the object connected domain block. Then, whether the position information of the object connected domain block is changed or not is monitored in a subsequent shot image, if the position information of the object connected domain block is not changed, the object corresponding to the object connected domain block is determined as a left object, and a face image of a pedestrian corresponding to the object is determined as a face image of a loser of the left object. Therefore, whether the object is left over or not in the target environment can be monitored in real time and automatically, and the timeliness of finding the object is improved. And can also confirm the identity information of the owner of the lost according to the pedestrian's face image that this legacy corresponds fast to in time contact and get before the owner of the lost, can also carry out the identity according to the face image pair person of getting fast simultaneously and check, in order to improve the legacy and return or the treatment effeciency, promote customer experience, still can use manpower sparingly and consume.

In one embodiment, as shown in fig. 2, the process of establishing the background model used in step 102 specifically includes the following steps:

step 201, a second shot image of the target environment is acquired as a sample image.

In an implementation, the server may acquire a second captured image of the target environment as a sample image for establishing the background model, where the sample image includes a plurality of sample images as a sample image library. For example, a second captured image of the target environment may be captured by a camera installed in the target environment, such as a captured image captured within a preset time period (e.g., within 2 minutes) before or after the business hours of the day in a bank outlet hall as a sample image. Wherein the shooting angle and the shooting range of the second shot image are consistent with those of the first shot image. If M frames of sample images are collected as the sample image library, the sample image library s (x) can be expressed as:

S(x)＝{I ₁ (k)，I ₂ (k)，...I _t (k)，...，I _M (k)}

wherein, I _t (k) And expressing the pixel value of a pixel point k of the t frame sample image in the sample image library, wherein if the number of the pixel points in the sample image is P, the value of k is 1, 2.

Step 202, determining a target sample image according to the sample image.

In implementation, the server may determine a plurality of target sample images according to the sample images, where the target sample images may be all the sample images acquired in step 201 or may be partial sample images. For example, the server may randomly select a preset number (which may be denoted as N) of sample images from a plurality of sample images (i.e., a sample image library) as the target sample image, where the specific number may be set according to the situation, such as 4.

Step 203, for each pixel point in each target sample image, randomly selecting a pixel value of a pixel point from the pixel points in the preset neighborhood of the pixel point in the target sample image as a sample pixel value of the pixel point in the target sample image.

In implementation, for each target sample image determined in step 202, for each pixel point in the target sample image, a pixel value of a pixel point is randomly selected from pixel points in a preset neighborhood of the pixel point in the target sample image, and the pixel value is used as a sample pixel value of the pixel point in the target sample image, so that a plurality of sample pixel values of the same pixel point in each target sample image can be obtained. For example, if N sample images are randomly selected from the sample image library as the target sample images, for a pixel point k with a position coordinate of (I, j) in the ith target sample image, the pixel value of a pixel point is randomly selected from a preset neighborhood of the pixel point k in the target sample image as the sample pixel value (which may be denoted as I) of the pixel point k in the target sample image _rand-i (k) 1, 2.., N). The preset neighborhood may be set empirically or experimentally, and may be, for example, an 8 neighborhood, a 5 × 5 neighborhood, or the like.

And 204, performing weighted average calculation according to the sample pixel values of the pixel points in each target sample image to obtain a background sample library pixel value of the pixel points.

The background sample library pixel values are sample values corresponding to the pixel points in the background model, and the pixel points in the background model correspond to the pixel points in the sample image and also correspond to the pixel points in the first shot image. Each pixel point in the background model is provided with a plurality of sample values, and the plurality of sample values form a background sample library of one pixel point in the background model. For example, the background model m (k) may be expressed as:

M(k)＝{v ₁ ，v ₂ ，...，v _j ，...，v _Q }

wherein v is _j One sample value (i.e., a background sample library pixel value) of a pixel point k in the background model m (k), and Q is the number of samples. The Q sample values form a background sample library of the pixel points, and the background sample library of all the pixel points (corresponding to the pixel points in the sample image) in the background model forms a background model.

In practice, the server may be able to determine the target sample maps from the target sample maps calculated in step 203And carrying out weighted summation on the sample pixel values of all the pixel points in the image to obtain a background sample library pixel value of each pixel point. The weight of each sample pixel value may be preset, or may be calculated by using other methods. In one example, the sample pixel values may be set to be equally weighted, i.e., for each sample pixel value I _rand-i (k) Performing mean value calculation to obtain a background sample library pixel value v of the pixel point k _j The calculation formula of (a) is as follows:

where N is the total number of target sample images, I _rand-i (k) And (i ═ 1, 2., N) is the sample pixel value of the pixel point k in the ith target sample image.

Step 205, the step of determining the target sample image according to the sample image is executed again until the background sample library pixel values of the preset number of each pixel point in the sample image are obtained, and a background model of the target environment is established based on the background sample library pixel values of the preset number of each pixel point in the sample image.

In practice, the server calculates a background sample library pixel value v of the pixel point k based on the target sample image determined in step 202 _j Then, the step 202 may be executed again, the target sample image is re-determined (for example, N sample images are randomly selected as the target sample image), and based on the new target sample image, another background sample library pixel value of the pixel point k is determined until a preset number (marked as Q) of background sample library pixel values corresponding to each pixel point are obtained, that is, the background sample library { v) corresponding to each pixel point is obtained ₁ ，v ₂ ，...，v _j ，...，v _Q }. Then, a background model of the target environment is established based on the preset number of background sample library pixel values of each pixel point, namely, the background sample library { v } of each pixel point ₁ ，v ₂ ，...，v _j ，...，v _Q Constitute a background model m (k) ═ v ₁ ，v ₂ ，...，v _j ，...，v _Q }。

In this embodiment, a second captured image is collected as a sample image, a target sample image is determined from the sample image, a neighborhood pixel value of each pixel point is collected from each target sample image as a sample pixel value of each target sample image, then, weighted averaging is performed on the sample pixel values of each target sample image to obtain a background sample library pixel value of each pixel point, and then, the step of determining the target sample image is returned until a preset number of background sample library pixel values are obtained, so that a background model is established. In the method, the background model established according to the plurality of shot images contains time domain information of a target environment, so that the influence of a Ghost phenomenon (or a Ghost phenomenon) can be reduced, the moving foreground target detection is carried out based on the background model, the connected domain lumps corresponding to each moving foreground target can be more accurately detected, and the detection accuracy of the remnant is further improved.

In one embodiment, as shown in fig. 3, the process of calculating the pixel value of the background sample library in step 204 specifically includes the following steps:

step 301, performing mean value calculation on sample pixel values of pixel points in each target sample image to obtain a mean value.

In implementation, after the server selects the sample pixel value of each pixel point in each target sample image in step 203, for each pixel point, the server may calculate the sample pixel value I of the pixel point (e.g., pixel point k) in each target sample image _rand-i (k) Mean value of (can be denoted as I) _mean (k) ). If the number N of the target sample images is larger than the number N of the target sample images, the average value I corresponding to the pixel point k _mean (k) The calculation formula of (a) is as follows:

step 302, calculating the weight of the pixel points in each target sample image by adopting a one-dimensional Gaussian distribution function according to the sample pixel values and the average values of the pixel points in each target sample image, and determining the weight of the average value according to the weight of the pixel points in each target sample image.

In implementation, the server may determine the sample pixel value I of the pixel point k in each target sample image _rand-i (k) And the mean value I corresponding to the pixel point calculated in step 301 _mean (k) Calculating the weight (which can be recorded as alpha) of the pixel point in each target sample image by adopting a one-dimensional Gaussian distribution function _i ). In one example, the weight α of the pixel point k in each target sample image _i The calculation formula of (a) is as follows:

wherein alpha is _i And (i 1, 2.., N) represents the weight of the pixel point k in the ith target sample image, and N is the number of the target sample images. σ is a standard deviation of the one-dimensional gaussian distribution function, and can be preset according to experiments or experience.

Then, the server can obtain the weight alpha of the pixel point k in each target sample image _i Calculating the mean value I corresponding to the pixel point k _mean (k) The calculation formula of the weight β is as follows:

step 303, performing weighted average calculation according to the sample pixel value and the mean value of the pixel points in each target sample image, the weight of the pixel points in each target sample image, and the weight of the mean value, to obtain a background sample library pixel value of the pixel points.

In implementation, the server may determine the sample pixel value I of the pixel point k in each target sample image _rand-i (k) Mean value I corresponding to pixel point k _mean (k) And the weight alpha of the pixel point k in each target sample image _i And the mean value I _mean (k) The weighted average calculation is carried out to obtain a background sample library pixel value v of the pixel point k _j Background sample library pixel value v _j Is as follows：

v _j ＝∑α _i I _rand-i (k)+βI _mean (k)

In this embodiment, a mean value of sample pixel values of each pixel point in each target sample image is calculated, and then a weight of the pixel point in each target sample image is calculated by using a one-dimensional gaussian distribution function according to each sample pixel value and the corresponding mean value, and then a weight average calculation is performed according to each weight, each sample pixel value, and the mean value, so as to obtain a background sample library pixel value of the pixel point. The larger the difference value between the sample pixel value and the mean value of the pixel point is, the more the pixel point is not in accordance with the principle of homogeneity and consistency of pixels, so that the weight of the pixel point in each target sample image can be calculated by adopting a one-dimensional Gaussian distribution function, the weight of the pixel point in each target sample image can be more reasonably distributed, and the accuracy of foreground detection can be improved by the established background model, so that the accuracy of the detection of the left-over object is improved.

In one embodiment, the process of determining the carry-over in step 106 specifically includes the following steps: determining the position information of the article connected domain block in the next frame of shot image and the shot image after the next frame of shot image; calculating the position distance of the article connected domain block mass in the two adjacent frames of shot images according to the position information of the article connected domain block mass in the two adjacent frames of shot images; if the position distance is smaller than or equal to a preset threshold value, recording that the number of times of stillness of the object connected domain block is increased once; and when the number of times of stillness of the article connected domain block masses is greater than the preset number of times, determining the articles corresponding to the article connected domain block masses as the remnants of the target pedestrian.

In implementation, after the server identifies that the preset area range corresponding to the target pedestrian includes the article connected domain blob in the next frame of captured image of the first captured image, the server may further determine the position information of the article connected domain blob in the next frame of captured image (which may be referred to as the t-th frame of captured image), and may record the position information, for example, the centroid position information of the article connected domain blob may be recorded in a position array linked list, which may be expressed as follows,

Track_list[m]＝＜pos_center(x，y)[m]，boundry_box[m]，flag，count＞

wherein, the Track _ list [ m ] is a position array linked list corresponding to the article connected domain block (which can be marked as m, if a plurality of article connected domain blocks exist, the article connected domain blocks can be distinguished according to the mark); pos _ center (x, y) [ m ] is the centroid position information of the commodity connected domain blob m; bounding _ box [ m ] is the external minimum rectangular frame information of the object connected domain blob m; the flag is a label obtained by detecting the pedestrian of the object connected domain block, wherein the label of the object connected domain block is 0, and the label of the pedestrian connected domain block is 1; count corresponds to the number of stillness, starting with 0.

Then, the server may determine the location information of the object connected domain block m in the captured image after the next captured image (t-th captured image), and record the location information in the location array linked list Track _ list [ m ] corresponding to the object connected domain block m to form a series of location information parameters. Then, the server can obtain the position information of the object connected domain block in the two adjacent frames of shot images from the position array linked list Track _ list [ m ], calculate the position distance of the two position information, and compare the position distance with a preset threshold value. If the position distance is smaller than or equal to the preset threshold, the number of times of stillness of the object connected domain block m is recorded in the position array chain table Track _ list [ m ] and is increased once, namely the count in the chain table is increased by 1. And when the number of times of stillness of the article connected domain block m is greater than the preset number of times, determining the article corresponding to the article connected domain block as the remnant of the target pedestrian.

In this embodiment, the position distance of the article connected domain blob in the two adjacent frames of the shot images is calculated according to the position information of the article connected domain blob in the two adjacent frames of the shot images, if the position distance is smaller than or equal to a preset threshold, the number of times of stillness of the article connected domain blob is recorded to be increased once, and when the number of times of stillness of the article connected domain blob is greater than the preset number of times, the article corresponding to the article connected domain blob is determined as the remnant of the target pedestrian. Therefore, whether the article corresponding to the article connected domain block is a remnant can be judged quickly, in real time and accurately.

In one embodiment, the method further comprises a step of sending a carry-over pick-up prompt message to the target pedestrian, specifically comprising: and according to the face image corresponding to the target pedestrian, identifying the identity of the target pedestrian, determining the contact information of the target pedestrian, and sending the remnant getting prompt information to the target pedestrian according to the contact information.

In implementation, the server may call the face recognition interface to perform identity recognition on a face image corresponding to the target pedestrian, determine identity information and contact information of the target pedestrian, and send a legacy pickup prompt message to the target pedestrian according to the contact information, for example, send a short message to the target pedestrian to notify the target pedestrian (i.e., the owner) of picking up the legacy before. The left object pick-up prompt message can contain a pick-up code corresponding to the left object, so that the owner can pick up the left object conveniently according to the pick-up code.

In the embodiment, the contact information is determined by identifying the face image of the target pedestrian, and the obtaining prompt information is sent to the target pedestrian so as to timely inform the owner of obtaining the abandoned object before the owner is lost, so that the returning efficiency of the abandoned object can be improved, and the customer experience can be improved.

In one embodiment, the method further comprises the step of identity verification of the retriever, which specifically comprises: collecting a face image of a person to be picked aiming at a target legacy object; and matching and identifying the face image of the person to be picked and the face image of the pedestrian corresponding to the target object, and outputting prompt information that the identity check is passed if the matching is successful.

In implementation, when the retriever receives the legacy from the front, the operator may be presented with a pickup code corresponding to the legacy to be received (i.e., the target legacy), or select the legacy to be received on a legacy image display interface of the service terminal, so as to determine the target legacy. Then, the service terminal may send an identifier of the target leave-behind object (e.g., a pickup code or other number corresponding to the leave-behind object) to the server, and the server may obtain a pre-stored face image of the pedestrian corresponding to the target leave-behind object according to the identifier of the target leave-behind object (see the description of step 105 for details of the storage process). Then, the server may collect the face image of the retriever, for example, the face image data of the retriever may be collected by a shooting device of the service terminal, and then the face image data is sent to the server. And then, the server can match and identify the face image data of the person to be picked up and the face image of the pedestrian corresponding to the target object to be left, and if the matching is successful, the server can output prompt information that the identity check is passed. The staff can return the target remnant to the retriever according to the prompt message.

In this embodiment, can match the discernment through the facial image of the person of getting of collection to the target legacy, with the pedestrian's that the target legacy that stores in advance corresponds facial image, if match successfully, then the identity is checked and is passed through to make the staff return the target legacy and give the person of getting, can save the manpower consumption that the manual work checked the identity from this, and can improve the treatment effeciency of returning the legacy, promote customer experience.

In one embodiment, as shown in fig. 4, the method further includes the step of releasing the legacy to the retriever through the legacy retriever, which specifically includes:

step 401, receiving a human face image of a retriever for a target legacy sent by a legacy retriever.

In the implementation, bank's staff can put into the legacy machine of drawing with the legacy, makes the person of drawing can draw the legacy by oneself through the legacy machine of drawing. The person of drawing can be at the person of drawing of remnant get the machine input target object leave-over's code of getting, perhaps choose the image of well target leave-over in the display screen of the person of drawing, then, the person of drawing can gather the people's face image through shooting device (like the camera) by the person of drawing, and will draw the people's face image and send for the server, the person of drawing can also send the sign that the target left-over corresponds for the server to the person of drawing.

And 402, matching and identifying the face image of the retriever and the face image of the pedestrian corresponding to the target remnant, and sending a throwing instruction aiming at the target remnant to the remnant retriever under the condition of successful matching so that the target remnant is thrown to the retriever by the remnant retriever according to the throwing instruction.

In implementation, after receiving the face image of the retriever, the server may perform matching identification on the face image of the retriever and the face image of the pedestrian corresponding to the target remnant, and send a delivery instruction for the target remnant to the remnant retriever if matching is successful. And after the remnant getting machine receives the throwing instruction, the target remnant can be thrown to the taker.

In this embodiment, can receive the machine through leaving over and keep and put in the leave-over thing, realize leaving over the thing and receive by oneself, use manpower sparingly consumption, improve the leave-over thing efficiency of returning and promote customer experience.

In one example, the legacy extractor is schematically illustrated in fig. 5, and specifically includes a housing 110, a touch display screen 120, an image capturing device 130, a legacy throwing device 140, a sliding rail 151, a tray assembly 152, a driving assembly 153, a controller 160, a transceiver 170, a disinfecting device 180, a ventilating device 190, and the like. The carry-over object putting device 140 includes a carry-over object accommodating part 141 and a putting part 142, a movable end of the putting part 142 is provided with an electromagnetic part 1421, and the carry-over object accommodating part 141 is provided with a magnetic buckle 1411. The releasing member 142 further includes a rotating member 1422, and the rotating member 1422 may specifically include a driving motor 1423, a driving belt 1424, and a rotating shaft 1425. The lower end of the tray component 152 is provided with a roller 1521 matched with the slide rail 151, and the transmission end of the driving component 153 is connected with the roller 1521. The tray assembly 152 further includes a tray 1522, a telescoping member 1523, and a gravity sensor 1524. The drive assembly 153 may specifically include a drive motor 1531 and a drive rod 1532.

After monitoring the object left in the target environment by using the object left monitoring method according to the above embodiment, the worker may place the object left into the object left accommodating part 141 of the object left getting machine shown in fig. 5, and then the server may generate a corresponding pickup code for the object left, and determine contact information of the owner by recognizing a face image of a pedestrian corresponding to the object left, and then send the pickup code to the owner, and notify the owner to get the object left according to the pickup code. Based on this, in an example, as shown in fig. 6, there is also provided a legacy retriever-based legacy retriever method, which can be applied to the legacy retriever shown in fig. 5, and specifically includes the following steps:

and 601, acquiring a pickup code input by a retriever.

Specifically, the retriever may input the pickup code through the touch display screen 120 of the legacy pickup machine, and the touch display screen 120 sends a signal corresponding to the pickup code to the controller 160, so that the controller 160 determines the corresponding legacy accommodation part 141 according to the pickup code.

Step 602, verifying the pickup code, and determining whether the verification is passed.

Specifically, the controller 160 may perform step 603 if the pickup code is verified, and perform step 606 if the pickup code does not pass.

And 603, acquiring a face image of the person to be picked, and sending the face image and the pickup code to a server so as to enable the server to perform face matching identification.

Specifically, the controller 160 may control the image capturing device 130 to capture a face image of the person to be picked up, and send the face image and the pickup code to the server through the transceiver 170. The server can determine the target object according to the pickup code, perform matching identification according to the face image of the pickup person and the stored face image of the pedestrian corresponding to the target object, send matching success information to the transceiver 170 if matching is successful, and send matching failure information to the transceiver 170 if matching is failed.

And step 604, releasing the target legacy object under the condition of receiving the matching success information sent by the server.

Specifically, if the transceiver 170 receives the matching success information sent by the server, the controller 160 may control the telescopic component 1523 of the tray assembly 152 to extend, and control the rotating component 1422 of the releasing component 142 to rotate, so that the object carry-over accommodating component 141 storing the object carry-over object rotates to the upper side of the tray assembly 152, and then control the power-off of the power-on magnet 1421 connected to the object carry-over accommodating component 141, so that the object carry-over accommodating component 141 falls into the tray 1522 of the tray assembly 152, and further the controller 160 may control the driving component 153 to drive the tray assembly 152 to move to the picking door 111 along the sliding rail 151, after the picking person takes the object carry-over object at the picking door 111, the controller 160 controls the driving component 153 to drive the tray assembly 152 to move back to the initial position along the sliding rail 151, so as to complete releasing the object carry-over object.

Step 605, in case of receiving the matching failure information sent by the server, outputting a prompt message of the legacy pickup failure.

Specifically, if the transceiver 170 receives the matching failure information sent by the server, the controller 160 may control the touch display screen 120 to display a prompt that the legacy pickup fails.

And step 606, outputting prompt information of the error of the pickup code.

Specifically, if the controller 160 does not verify the pickup code, the controller 160 may control the touch display 120 to display a prompt message indicating that the pickup code is incorrect.

It should be understood that, although the steps in the flowcharts related to the embodiments described above are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be rotated or alternated with other steps or at least a part of the steps or stages in other steps.

Based on the same inventive concept, the embodiment of the present application further provides a carryover monitoring apparatus for implementing the above-mentioned carryover monitoring method. The solution to the problem provided by the device is similar to the solution described in the above method, so the specific limitations in one or more embodiments of the legacy monitoring device provided below can be referred to the limitations in the legacy monitoring method above, and are not described herein again.

In one embodiment, as shown in fig. 7, there is provided a carryover monitoring apparatus 700 comprising: a first acquisition module 701, a first detection module 702, a second detection module 703, a first identification module 704, a storage module 705, and a first determination module 706, wherein:

the first acquisition module 701 is configured to acquire a first captured image of a target environment.

The first detection module 702 is configured to perform moving foreground object detection on the acquired first captured image by using a moving object detection algorithm according to a pre-established background model, so as to obtain connected domain blobs corresponding to each moving foreground object in the first captured image.

The second detection module 703 is configured to perform pedestrian detection on the image corresponding to each connected domain blob to obtain a pedestrian connected domain blob corresponding to the target pedestrian.

The first identification module 704 is configured to identify whether an article connected domain blob is included in a preset region range corresponding to a target pedestrian in a next frame captured image of the first captured image.

The storage module 705 is configured to, when the object connected domain blob is included, obtain a face image corresponding to the target pedestrian and an object image corresponding to the object connected domain blob, and perform corresponding storage.

The first determining module 706 is configured to determine, if the position information of the blob in the article connected domain in the captured image after the next captured image is not changed, the article corresponding to the blob in the article connected domain as a carry-over object of the target pedestrian.

In one embodiment, the apparatus further comprises a second acquisition module, a second determination module, a selection module, a calculation module, and a setup module, wherein:

and the second acquisition module is used for acquiring a second shot image of the target environment as a sample image.

And the second determining module is used for determining the target sample image according to the sample image.

And the selecting module is used for randomly selecting the pixel value of one pixel point from the pixel points of the preset neighborhood of the pixel points in the target sample image as the sample pixel value of the pixel point in the target sample image aiming at each pixel point in each target sample image.

And the computing module is used for carrying out weighted average computation according to the sample pixel values of the pixel points in each target sample image to obtain a background sample library pixel value of the pixel points.

And the establishing module is used for returning to execute the step of determining the target sample image according to the sample image until the background sample library pixel values of the preset number of each pixel point in the sample image are obtained, and establishing the background model of the target environment based on the background sample library pixel values of the preset number of each pixel point in the sample image.

In one embodiment, the calculation module is specifically configured to perform mean calculation on sample pixel values of pixels in each target sample image to obtain a mean value; calculating the weight of the pixel points in each target sample image by adopting a one-dimensional Gaussian distribution function according to the sample pixel value and the average value of the pixel points in each target sample image, and determining the weight of the average value according to the weight of the pixel points in each target sample image; and performing weighted average calculation according to the sample pixel value and the average value of the pixel points in each target sample image, the weight of the pixel points in each target sample image and the weight of the average value to obtain a background sample library pixel value of the pixel points.

In one embodiment, the first determining module 706 is specifically configured to determine the position information of the article connected domain blob in the next captured image and in the captured image after the next captured image; calculating the position distance of the article connected domain block mass in the two adjacent frames of shot images according to the position information of the article connected domain block mass in the two adjacent frames of shot images; if the position distance is smaller than or equal to a preset threshold value, recording that the number of times of stillness of the object connected domain block is increased once; and when the number of times of stillness of the article connected domain block masses is greater than the preset number of times, determining the articles corresponding to the article connected domain block masses as the remnants of the target pedestrian.

In one embodiment, the device further comprises a second recognition module, which is used for performing identity recognition on the target pedestrian according to the face image corresponding to the target pedestrian, determining contact information of the target pedestrian, and sending the carry-over object picking-up prompt information to the target pedestrian according to the contact information.

In one embodiment, the apparatus further comprises a third acquisition module and a matching module, wherein:

and the third acquisition module is used for acquiring the face image of the person to be picked aiming at the target legacy.

And the matching module is used for matching and identifying the face image of the person to be picked and the face image of the pedestrian corresponding to the target object to be left, and outputting prompt information for passing identity check if matching is successful.

In one embodiment, the apparatus further comprises a receiving module and a transmitting module, wherein:

and the receiving module is used for receiving the face image of the retriever aiming at the target legacy sent by the legacy retriever.

And the sending module is used for matching and identifying the face image of the retriever and the face image of the pedestrian corresponding to the target legacy, and sending a throwing instruction for the target legacy to the legacy drawing machine under the condition of successful matching so that the target legacy drawing machine throws the target legacy to the retriever according to the throwing instruction.

The various modules in the above-described legacy monitoring device may be implemented in whole or in part by software, hardware, and combinations thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 8. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store data required or generated to perform the above-described carryover monitoring method. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a carryover monitoring method.

Those skilled in the art will appreciate that the architecture shown in fig. 8 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the above-described method embodiments when executing the computer program.

In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

In an embodiment, a computer program product is provided, comprising a computer program which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

The application provides a method, a device, computer equipment, a storage medium and a computer program product for monitoring the remnant, which relate to the technical field of artificial intelligence and can be used in the field of financial technology or other related fields.

It should be noted that, the user information (including but not limited to user device information, user (or client) personal information, etc.) and data (including but not limited to data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high-density embedded nonvolatile Memory, resistive Random Access Memory (ReRAM), Magnetic Random Access Memory (MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (PCM), graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others. The databases referred to in various embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing based data processing logic devices, etc., without limitation.

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A carryover monitoring method, the method comprising:

acquiring a first shot image of a target environment;

2. The method of claim 1, wherein the background model is established by:

acquiring a second shot image of the target environment as a sample image;

determining a target sample image from the sample image;

and returning to the step of executing the step of determining the target sample image according to the sample image until the background sample library pixel values of the preset number of each pixel point in each sample image are obtained, and establishing the background model of the target environment based on the background sample library pixel values of the preset number of each pixel point in the sample image.

3. The method of claim 2, wherein said performing a weighted average calculation based on the sample pixel values of said pixel points in each of said target sample images to obtain a background sample library pixel value for said pixel points comprises:

4. The method according to claim 1, wherein if the position information of the object connected domain blob in the captured image after the next captured image is not changed, determining the object corresponding to the object connected domain blob as the carry-over object of the target pedestrian comprises:

5. The method of claim 1, further comprising:

6. The method of claim 1, further comprising:

acquiring a face image of a person to be picked aiming at a target remnant;

7. The method of claim 1, further comprising:

8. A carryover monitoring apparatus, the apparatus comprising:

and the determining module is used for determining the article corresponding to the article connected domain block as the remnant of the target pedestrian if the position information of the article connected domain block in the shot image after the next frame of shot image is not changed.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 7 when executed by a processor.