TWI796033B - People flow analysis and identification system - Google Patents

Abstract

The disclosure provides a people flow analysis and identification system, which comprises a storage unit, a space range setting unit, an image capturing unit, a portrait processing unit and a display processing unit. The storage unit may store a 2-D map for a first space, and the space range setting unit may input range information for a region in the first space. The image capturing unit may capture a first image of the region in the first space. The portrait processing unit may capture a target image from the first image according to the range information, and generate at least one mark point on the target image by a counting algorithm. The display processing unit may transform the at least one mark point on the target image into at least one people distribution point on the 2-D map by an affine transformation algorithm, and then show the people distribution point on the 2-D map on a monitor to users.

Description

People Flow Analysis and Identification System

The present invention relates to a human flow analysis and identification system, in particular to an analysis and identification system that accurately calculates the number of people and changes in the flow of people in an area by using image capture and arithmetic processing. .

In the past, in certain venues, such as rally venues, exhibition halls, or New Year’s Eve parties, many people would be attracted to attend. The total number of people at the scene can only be estimated. Sometimes it is calculated by taking a bird’s-eye view of the crowd on the entire site from an aerial photo. Sometimes it is calculated by multiplying the average number of people entering and exiting at the entrance and exit. Sometimes it is calculated by the site Dividing the area by the number of people who can stand within one square meter is used for calculation, which causes a huge gap in the calculation of the number of people in a venue. Over time, the credibility of the number of participants is no longer convincing.

Furthermore, in some specific applications, the recognition of the flow of people in these fields has begun. The purpose is to know the direction of movement of the crowd and the characteristics of the crowd. The more commonly used method can be to use the heat source The way of sensing or identifying mobile devices is used to identify the movement of the crowd and whether it is a large crowd. However, the former cannot accurately calculate the number of people in the crowd due to the isolation of the human body in the crowd, while the latter can only Determine the number of people in a field and their flow However, if it is necessary to further know the number of people and changes in the flow of people in an area of the venue, it cannot be achieved through this method.

In view of the foregoing, the inventor of the present invention conceived and designed a human flow analysis and identification system in order to improve the deficiencies of conventional technologies and further enhance industrial implementation and utilization.

Based on the above purpose, the present invention provides a people flow analysis and identification system, which is suitable for identifying people flow information on a platform host, which includes an image data unit, a field range setting unit, an image capture unit, and a portrait processing unit. unit and a display processing unit. The image data unit is used to store a planar map of the first field. The field range setting unit is used for inputting information of a range of the first field. The image capturing unit is used for capturing an image of an area in the first field. The portrait processing unit extracts a target range image corresponding to the range information from the area image, and uses a crowd counting algorithm to generate at least one marked point in the target range image, wherein each marked point is Represents a human being in this first field. The display processing unit uses a two-dimensional affine transformation algorithm to convert the marked points in the target range image to the planar map, so that at least one crowd distribution point corresponding to the marked points is displayed on the planar map, and Display this plane map and these crowd distribution points on the display screen.

Preferably, the crowd analysis and identification system of the present invention further includes a crowd display unit, which is used to capture and display the at least one crowd distribution point continuously generated by the processing unit within a time interval, and display them in a time sequence The at least one crowd distribution point is used to generate a first crowd movement distribution on the planar map of the display screen.

Preferably, the range information includes coordinates, booth name, address or landmark information.

Preferably, the display processing unit uses different colors on the planar map to present the distribution density of the crowd distribution points at different positions.

Preferably, when the distribution density is greater than a warning threshold, the display processing unit sends a warning notification message.

Preferably, the people flow analysis and recognition system of the present invention further includes a people flow video concatenation unit to connect with a regional people flow movement system, and the regional people flow movement system is used to display the second people flow movement distribution in the second field.

Preferably, the display processing unit serially displays the first movement distribution of the flow of people and the second movement distribution of the flow of people through the connection unit of the flow of people images.

Preferably, the second field includes the first field.

Preferably, this second domain is independent of said first domain.

From the above, it can be seen that the crowd analysis and identification system of the present invention can indeed solve the long-standing problems in the conventional technology. It captures the field image, sets the target range in the image, and recognizes the portraits in the target range And convert the portrait into a marked point to present on the 2D map of the field, so that the user can immediately know the crowd or flow of people in a specific range in the field.

What's more, the people flow analysis and identification system of the present invention can be connected in series with the people flow movement system in another area, so that the people flow in another place can be connected or connected in series to the people flow analysis and identification system of the present invention, so that users can more clearly Understand changes in foot traffic for analysis, therefore, Compared with the prior art, the present invention is indeed novel and progressive, and also has substantial industrial applicability to related fields.

100: People flow analysis and identification system

10: Image and data unit

11: Plane map

20: field range setting unit

21, 211, 212, 213, 214: scope information

30: Image capture unit

31: Regional image

40:Portrait processing unit

41: target range image

42: Dimension point

50: display processing unit

51: crowd distribution point

52: display screen

53: Distribution density

60: People flow display unit

61: The first flow of people mobile distribution

70: Crowd flow video concatenation unit

200: Regional People Flow Mobile System

201: The distribution of the second flow of people

A: The first field

B: the second field

Figure 1 is a block diagram of the people flow analysis and identification system of the present invention.

Figure 2a is the first schematic diagram of the people flow analysis and identification system of the present invention.

Figure 2b is the second schematic diagram of the people flow analysis and identification system of the present invention.

Figure 2c is the third schematic diagram of the people flow analysis and identification system of the present invention.

Figure 2d is the fourth schematic diagram of the people flow analysis and identification system of the present invention.

Fig. 3 is a block diagram of a people flow analysis and identification system according to another embodiment of the present invention.

Fig. 4 is a schematic diagram of the regional people flow moving system of the present invention.

In order for Ligui examiners to understand the inventive features, content and advantages of the present invention and the effects it can achieve, the present invention is hereby combined with the accompanying drawings and described in detail in the form of embodiments as follows, and the drawings used therein, its The subject matter is only for illustration and auxiliary description, not necessarily the true proportion and precise configuration of the present invention after implementation, so the scale and configuration relationship of the attached drawings should not be interpreted to limit the scope of rights of the present invention in actual implementation.

The advantages, features and technical methods achieved by the present invention will be described in more detail with reference to exemplary embodiments and accompanying drawings to make it easier to understand, and the present invention can be implemented in different forms, so it should not be understood as being limited to what is shown here The stated embodiments, on the contrary, are provided to make the present disclosure more thorough and relevant to those having ordinary knowledge in the art. It is intended that the scope of the present invention be fully and completely conveyed, and that the present invention shall be defined only by the appended claims.

Please refer to Fig. 1, which is a block diagram of the crowd analysis and identification system of the present invention. As shown in the figure, the people flow analysis and identification system 100 is applicable to a platform host for identifying people flow information, wherein the platform host can include a desktop computer, a workstation host or a notebook computer. The people flow analysis and identification system 100 may include an image data unit 10, a field range setting unit 20, an image capture unit 30, a portrait processing unit 40, and a display processing unit 50, wherein the image data unit 10, field The range setting unit 20, the portrait processing unit 40, and the display processing unit 50 can be a software application program or a sub-function in the application program.

In this embodiment, the image map unit 10 can be a database application program, which can be used to store a plane map 11 of the first field A in advance. The field range setting unit 20 can be used to input the range information 21 of the first field A, wherein the input method can be input using computer peripherals, such as keyboard, mouse, etc., and the range information 21 can include coordinates , booth name, address or landmark information.

In this embodiment, the image capture unit 30 can be a camera connected to the platform host, which can be used to capture an area image 31 of the first field A. It can be understood that this area image 31 can be composed of multiple The images captured by the cameras are combined, and the purpose is to expand the area covered by the area image 31 as much as possible.

After the area image 31 is generated, the portrait processing unit 40 can extract a target area image 41 corresponding to the area information 21 from the area image 31, and the extraction method can be divided into multiple sub-areas in the area image 31 in advance. , and each sub-area will correspond to a unique range information, such as coordinate position, etc., so that through the set range data Then, the corresponding target range image 41 can be extracted from the area image 31 . Afterwards, the portrait processing unit 40 can use a crowd counting algorithm to generate at least one marked point 42 in the target area image 41 , wherein each marked point 42 represents a human being in the first field A.

It is worth mentioning that the crowd counting algorithm used in the present invention can be based on a detection method or a regression method. The former algorithm includes methods such as SVM, boosting and random forest, which use the detection of the whole or part of the body to The number of people is counted, while the latter algorithm first extracts low-level features, and then uses regression models, such as linear regression, piecewise linear regression, or Gaussian process regression, to learn the mapping relationship between low-level features and the number of people. These algorithms belong to the prior art in the field of crowd counting, and will not be described in detail in the present invention, and related introductions can be obtained from network resources, such as https://www.gushiciku.cn/pl/2xTE/zh-tw .

Then, a two-dimensional affine transformation (affine transformation) algorithm (or called a six-parameter transformation algorithm) can be used by the display processing unit 50 to transform all the marked points 42 in the target range image 41 to the planar map 11, At least one crowd distribution point 51 corresponding to the marked points 42 is displayed on the planar map 11 , and finally the planar map 11 and the crowd distribution point 51 are displayed on a display screen 52 for users to watch.

It is worth mentioning that this two-dimensional affine transformation algorithm refers to the calculation method of performing a linear transformation on a vector space in geometry, followed by a translation, and transforming it into another vector space. In this embodiment, The present invention uses this algorithm to convert the marked points 42 of the target range image 41 in the three-dimensional space to the two-dimensional space, so as to generate the crowd distribution points 51 corresponding to the positions on the planar map 11 . Since the algorithm system is widely used in the field of image processing Common knowledge is well known, so the present invention will not describe it, and related content can refer to network resources, such as Wikipedia or Google.

Through the above-mentioned method, the present invention can capture images of an area, and further identify the distribution of crowds in a target area, and then convert the result of crowd identification into a planar map. In this way, for this The flow of people in the target area can be effectively controlled. What's more, in a preferred embodiment, the display processing unit 50 can use different colors on the planar map 11 to present a distribution density of the crowd distribution points 51 at different positions, for example, when the distribution density is too high, then in The crowd distribution point 51 of this position shows red as warning, and further, when this distribution density 53 is greater than a warning threshold value, this display processing unit 50 just can send out a warning notification message to notify other people, and this application It can be implemented on suspension bridges and elevators with a limited number of people, or in a field with a maximum number of people in groups, which can effectively prevent industrial safety accidents or casualties caused by excessive numbers of people.

Please refer to Figures 2a to 2d, which are schematic diagrams of the people flow analysis and identification system of the present invention, and please also refer to Figure 1. As shown in the figure, Figure 2a shows a planar map 11 of a first field A stored in the image map information unit 10 of the present invention, and the user can use the field range setting unit 20 to input the location of the first field A. The scope information 211, 212, 213, 214 is used to set the size of the scope for observing the flow of people.

Figure 2b shows an area image 31 of the first field A captured by the image capture unit 30, and can be extracted from the area image 31 corresponding to the range information 211, 212, 213, 214 surrounded by A target range image 41 .

Figure 2c shows that the portrait processing unit 40 uses a crowd counting algorithm to generate the marked points 42 in the target range image 41, wherein each marked point 42 represents the presence of One of the portraits in the target range image 41 . Finally, the display processing unit 50 uses a two-dimensional affine transformation algorithm to convert all the marked points 42 in the target range image 41 to the planar map 11 to form crowd distribution points 51 on the planar map for users to view. , and its schematic diagram is shown in Figure 2d.

Please refer to FIG. 3, which is a block diagram of a people flow analysis and identification system according to another embodiment of the present invention, and please refer to FIG. 1. The people flow analysis and recognition system 100 of this embodiment can be applied on a platform host, which includes the image data unit 10, field range setting unit 20, image capture unit 30, portrait processing unit 40 and display processing as described above. In addition to the unit 50, it can further include a people flow display unit 60 and a people flow video concatenation unit 70, wherein the people flow display unit 60 and the people flow video concatenation unit 70 can be a software application or one of the software applications subroutine.

Further explanation, as mentioned above, the display processing unit 50 can convert the marked points 42 in the target range image 41 to the planar map 11, so that all the crowd distribution points 51 corresponding to these marked points 42 are presented on the planar map 11, and The crowd flow display unit 60 can capture the crowd distribution points 51 continuously generated by the display processing unit 50 within a time interval, and display these crowd distribution points 51 in time order, and then display them on the plane map 11 of the display screen 52 A first movement distribution 61 of people flow is generated.

For example, the people flow display unit 60 can capture the planar map 11 and the crowd distribution points 51 every 5 minutes within an hour, so that 12 continuous images can be generated, and the people flow display unit 60 can use 2 The 12 consecutive images are continuously played at an interval of ~3 seconds, so that the user can see the movement of people in the target area image 41 through a dynamic image, thereby deepening the impression of the user when watching.

In addition, this embodiment also includes a people flow image concatenation unit 70, which is used to connect another area people flow moving system 200, wherein this area people flow moving system 200 is used to display people in a second field B. A second crowd movement distribution 201, and this second field B may include the first field A or be independent of the first field A, the former means that the first field A is in the second field B One of the sub-areas, and the latter means the area where the first field A and the second field B are respectively located in different positions, and the technical characteristics of the present invention can be displayed in the two fields through the human flow video concatenation unit 70 Movement of people between domains.

In detail, the regional people flow moving system 200 may include the people flow analysis and identification system of the present invention or a people flow moving system built in other ways. Taking Figure 4 as an example, the human flow movement system 200 in this area represents the movement of people in Hualien County, and its generation method can be calculated by identifying the telecommunication equipment carried by the user, and it can be calculated based on the location in different time periods Knowing the mobile location of the user in Hualien County, and then causing the movement of the flow of people. It is worth mentioning that the position recognition method of the regional people flow moving system 200 is not limited to this, it can also be the people flow analysis and recognition system proposed by the present invention or a system that uses other identification items to generate people flow calculation.

In this embodiment, the second field city B includes the first field A, in which Hualien County can be set as the second field B, and Hualien County’s Guangfu Station can be set as the first field A, and the display processing unit 50 In addition to displaying the second people flow distribution 201 of the regional people flow moving system 200 through the people flow image concatenation unit 70, the user can also further select the Guangfu station on the map, and then the people flow image concatenation unit 70 will display it in tandem The plane map of Guangfu station is used for users to further carry out the above-mentioned flow analysis and recognition program on the first field A, and the The content of the people flow analysis and identification program has been disclosed in the above embodiments, so it will not be repeated here.

From the above, it can be seen that the people flow analysis and identification system of the present invention can indeed solve the long-standing problems in the prior art. It can accurately identify the number of people and their flow of people in an area through image capture and image processing. Changes, thereby producing a more fair and just crowd number, moreover, when it is necessary to strictly regulate the upper limit of the number of people in a specific field, the identification system of the present invention can immediately display the number of people in the specific field, and then immediately take the most appropriate Measures to meet the needs of the specific field. In addition, in addition to analyzing the flow of people in a specific field, the present invention can also be connected to the flow of people moving systems in other fields, so that the observation of flow of people in a more comprehensive manner.

The above-described embodiments are only to illustrate the technical ideas and characteristics of the present invention, and its purpose is to enable those skilled in this art to understand the content of the present invention and implement it accordingly, and should not limit the patent scope of the present invention. That is to say, all equivalent changes or modifications made according to the spirit disclosed in the present invention should still be covered by the patent scope of the present invention.

100: People flow analysis and identification system

10: Image and data unit

11: Plane map

20: field range setting unit

21: Scope information

30: Image capture unit

31: Regional image

40:Portrait processing unit

41: target range image

42: Dimension point

50: display processing unit

51: crowd distribution point

52: display screen

A: The first field

Claims (8)

A people flow analysis and identification system is suitable for identifying people flow information on a platform host, which includes: an image data unit for storing a plane map of a first field; a field range setting unit for to input range information of the first field; an image capture unit is used to capture an area image of the first field; a portrait processing unit is used to extract corresponding range information from the area image A target range image, and using a crowd counting algorithm to generate at least one marked point in the target range image, wherein each of the at least one marked point represents a human being in the first field; a display processing unit , using a two-dimensional affine transformation algorithm to convert the at least one marked point in the target area image to the planar map, so that at least one crowd distribution point corresponding to the at least one marked point is displayed on the planar map, and displayed The planar map and the at least one crowd distribution point are displayed on a display screen; and a crowd flow display unit captures the at least one crowd distribution point continuously generated by the display processing unit within a time interval, and displays them in chronological order Displaying the at least one crowd distribution point to generate a first crowd movement distribution on the planar map of the display screen; wherein, the crowd flow display unit captures the planar map at a first time interval within the time interval and these crowd distribution points to generate a plurality of continuous images, the crowd display unit further plays the plurality of continuous images continuously at a second time interval different from the first time interval to form a dynamic image, thereby through the Dynamic images to see the movement of people within the image of the target range. As for the people flow analysis and identification system described in item 1 of the scope of the patent application, the range information includes coordinates, booth names, addresses or landmark information. The crowd analysis and recognition system described in claim 1 of the patent application, wherein the display processing unit uses different colors on the planar map to present the distribution density of the at least one crowd distribution point at different positions. In the people flow analysis and identification system described in item 3 of the patent application, when the distribution density is greater than a warning threshold, the display processing unit sends a warning notification message. The people flow analysis and identification system described in item 1 of the scope of the patent application further includes a people flow video concatenation unit to be connected to a regional people flow moving system, wherein the regional people flow moving system is used to display one of a second field The distribution of the second flow of people. The people flow analysis and identification system described in claim 5 of the patent application, wherein the display processing unit displays the first people flow movement distribution and the second people flow movement distribution in series through the people flow video concatenation unit. According to the people flow analysis and identification system described in item 5 of the scope of the patent application, wherein the second field includes the first field. In the people flow analysis and identification system described in item 5 of the scope of the patent application, the second field is independent from the first field.

Images