TW202123075A - Image analyzing method and image analyzing device - Google Patents

Abstract

An image analyzing method is used for performing image analysis for a gate. A movable door is disposed at the gate and the door has a mark. The image analyzing method includes steps of capturing an image sequence related to the gate; detecting the mark from the image sequence; and determining that the door opens or closes according to a moving state of the mark.

Description

Image analysis method and image analysis device

The present invention relates to an image analysis method and an image analysis device, in particular to an image analysis method and an image analysis device that can determine whether a door is opened or closed according to the movement state of a sign on the door. In addition, the image analysis method and image analysis device of the present invention can further determine the time point for counting the object according to the moving direction of the object.

With the advent of the era of big data, the use of objective data analysis to review or improve the service quality and operating efficiency of a place has become an increasingly important trend. Many places also use cameras to track customers to count the number of customers entering and leaving. In addition, some places will be equipped with movable doors to count the number of customers entering and exiting according to the time interval of door opening and closing. At present, the prior art uses digital input to transmit the control signal of the door switch to the camera to determine the time interval of the door switch. However, for different places, there is no uniform standard for control circuits and control signals. When installing the camera, the installer needs to spend a lot of time researching the design of the control circuit. If the voltage range does not match, the camera may not operate or even be burned. In addition, for different places, the time points calculated by the customers entering and leaving from the entrance and exit should also be different, so as to increase the statistical accuracy of the number of customers entering and leaving the different places.

One of the objectives of the present invention is to provide an image analysis method and an image analysis device that can determine whether the door is opened or closed according to the movement state of the mark on the door, so as to solve the above-mentioned problems. In addition, the image analysis method and image analysis device of the present invention can further determine the time point for counting the object according to the moving direction of the object.

According to an embodiment, the image analysis method of the present invention is used to perform image analysis on an entrance and exit. A movable door is arranged at the entrance and exit, and the door has a sign. The image analysis method includes the following steps: capturing an image sequence about the entrance and exit; detecting the sign from the image sequence; and judging whether the door is opened or closed according to the movement state of the sign.

In another embodiment, the image analysis method may further include the following steps: capturing an image sequence about entrances and exits within a field of view; when an object in the image sequence passes through an entry and exit reference of the corresponding entrance, it is determined that one of the objects is moving Direction; when judging that the object moves from the first side of one of the entrances and exits to the second side of one of the entrances, the object is counted into a first cumulative total; when the object leaves the field of view, the exit position of one of the objects is recorded; and according to the exit position Decide whether to count the object in a second cumulative total.

According to another embodiment, the image analysis device of the present invention is used to perform image analysis on an entrance and exit. A movable door is arranged at the entrance and exit, and the door has a sign. The image analysis device includes an image capturing unit and a processing unit, wherein the processing unit is electrically connected to the image capturing unit. The image capturing unit is used for capturing an image sequence about the entrance and exit. The processing unit is used to execute the image analysis method described above.

In summary, the present invention detects the sign on the door by image analysis, and then judges whether the door is opened or closed according to the movement state of the sign on the door. Therefore, the image analysis device of the present invention does not need to receive additional digital input signals, and can effectively solve the problem of compatibility between the image analysis device and the control circuit of a specific place (for example, shops, public transportation vehicles, etc.). In addition, when the object in the image sequence passes the entry and exit criteria of the corresponding entrance, the present invention first determines the moving direction of the object, and then determines the time point for counting the object according to the moving direction of the object. In this way, the statistical accuracy of the number of customers entering and leaving a specific place can be effectively increased.

The advantages and spirit of the present invention can be further understood from the following detailed description of the invention and the accompanying drawings.

Please refer to Figures 1 to 4. Figure 1 is a functional block diagram of the image analysis device 10 according to an embodiment of the present invention, and Figure 2 is the image analysis device 10 in Figure 1 installed at the entrance 300 of the place 30 Fig. 3 is a schematic diagram of the door 301 in Fig. 2 after opening, and Fig. 4 is a flowchart of an image analysis method according to an embodiment of the present invention. The image analysis method in Figure 4 is applicable to the image analysis device 10 in Figure 1.

As shown in FIG. 1, the image analysis device 10 includes an image capturing unit 100 and a processing unit 102, wherein the processing unit 102 is electrically connected to the image capturing unit 100. In this embodiment, the image analysis device 10 may be a camera. In practical applications, the image capturing unit 100 may be a Charge-coupled Device (CCD) sensor or a Complementary Metal-Oxide Semiconductor (CMOS) sensor; the processing unit 102 may be Processor or controller with data processing/arithmetic functions. Generally speaking, the image analysis device 10 will also be provided with necessary software and hardware components during operation, such as circuit boards, power supplies, application programs, communication modules, lenses, etc., depending on the actual application.

As shown in FIG. 2, the image analysis device 10 is used to perform image analysis on one of the entrances and exits 300 of the place 30. In this embodiment, the image analysis device 10 can be installed at the entrance 300 of the site 30 to capture an image sequence about the entrance 300 within a FOV of the image analysis device 10. In practical applications, the place 30 can be a store, a public transportation vehicle, or other places where the total number of in and out objects needs to be counted.

In addition, the entrance 300 is provided with a movable door 301, and the door 301 has a sign 3010. In this embodiment, the door 301 can be opened (as shown in Figure 3) or closed (as shown in Figure 2) relative to the entrance and exit 300 in a rotating manner. In another embodiment, the door 301 can also be opened or closed relative to the entrance 300 by linear movement or other ways, depending on the actual application.

When the image analysis method is executed by the image analysis device 10, first, the image capturing unit 100 captures the image sequence about the entrance and exit 300 within the field of view FOV (step S10 in FIG. 4). In this embodiment, after capturing the image sequence, the processing unit 102 analyzes the image sequence by image analysis technology to detect the mark 3010 from the image sequence (step S12 in FIG. 4). Then, the processing unit 102 determines whether the door 301 is opened or closed according to the movement state of the flag 3010.

In this embodiment, the processing unit 102 can determine whether the door 301 is opened or closed according to the position change of the mark 3010 in the image sequence. As shown in Figure 2, when the door 301 is closed, the flag 3010 is located within the FOV of the field of view. Therefore, when the processing unit 102 detects the mark 3010 in the image sequence, the processing unit 102 can determine that the door 301 is closed. As shown in Figure 3, when the door 301 is opened, the flag 3010 is located outside the FOV of the field of view. Therefore, when the processing unit 102 cannot detect the mark 3010 in the image sequence, the processing unit 102 can determine that the door 301 is open.

In another embodiment, when the door 301 is closed, the mark 3010 may be located outside the FOV; when the door 301 is open, the mark 3010 may be located within the FOV of the field of view. Therefore, when the processing unit 102 cannot detect the sign 3010 in the image sequence, the processing unit 102 can determine that the door 301 is closed; when the processing unit 102 detects the sign 3010 in the image sequence, the processing unit 102 can determine that the door 301 is open .

In another embodiment, when the door 301 is closed, the mark 3010 may be located at a first position within the FOV; when the door 301 is opened, the mark 3010 may be located at a second position within the FOV. Therefore, when the processing unit 102 detects that the mark 3010 is located at the first position in the image sequence, the processing unit 102 can determine that the door 301 is closed; when the processing unit 102 detects that the mark 3010 is located at the second position in the image sequence, it processes The unit 102 can determine that the door 301 is open.

In another embodiment, the processing unit 102 can determine whether the door 301 is opened or closed according to the moving distance of the mark 3010 in the image sequence. Furthermore, the processing unit 102 can compare the moving distance of the marker 3010 in the image sequence with a predetermined distance threshold. When the moving distance of the mark 3010 is greater than the preset distance threshold, the processing unit 102 may determine that the door 301 is open; when the moving distance of the mark 3010 is less than or equal to the preset distance threshold, the processing unit 102 may judge that the door 301 is closed. The above-mentioned preset distance threshold can be set according to actual applications.

In this embodiment, the sign 3010 can be a sticker, a color block or other signs, and the sign 3010 can be manually set at a position on the door 301 that is not easily covered. In another embodiment, the image analysis device 10 may communicate with a user interface (for example, a computer), so that the user can circle a part of the door 301 in the image sequence through the user interface as the mark 3010. In addition, the present invention may use an object re-identification method to compare the image feature vector of the area where the sign 3010 is located to confirm whether the position of the door 301 has changed.

When determining whether the door 301 is open or closed, the present invention can further count objects (for example, people, animals, or other objects) in the image sequence. In this embodiment, when the processing unit 102 determines that the door 301 is open, the processing unit 102 can start counting the objects in the image sequence. When the processing unit 102 determines that the door 301 is closed, the processing unit 102 can stop counting the objects in the image sequence. In this way, the present invention can count the objects in the image sequence according to the time interval when the door 301 is opened and closed.

It should be noted that some objects will pass through the entrance and exit 300 during the opening process of the door 301. Therefore, in another embodiment, after the processing unit 102 determines that the door 301 is opened a predetermined time (for example, 1 second, 2 seconds, etc.), the processing unit 102 can start counting the objects in the image sequence. For example, the present invention can count the objects in the image sequence and temporarily store the counting result when the image sequence is captured. When the processing unit 102 determines that the door 301 is open, the processing unit 102 may include the counting result of the predetermined time before the door 301 is opened into the cumulative total. In this way, it is possible to avoid missing items that pass through the entrance and exit 300 during the opening process of the door 301. When the processing unit 102 determines that the door 301 is closed, the processing unit 102 can stop counting the objects in the image sequence.

Please refer to FIGS. 5-9. FIG. 5 is a flowchart of an image analysis method according to another embodiment of the present invention. FIG. 6 is a schematic diagram of the movement of the object 5 relative to the entrance and exit 300, and Figure 7 is the movement of the object 5 relative to the entrance and exit. Another schematic diagram of the movement of 300, FIG. 8 is another schematic diagram of the movement of the object 5 relative to the entrance 300, and FIG. 9 is another schematic diagram of the movement of the object 5 relative to the entrance 300. The image analysis method in Figure 5 is also applicable to the image analysis device 10 in Figure 1. The following uses the embodiments shown in FIGS. 6 to 9 to illustrate how the present invention counts objects in an image sequence. It should be noted that, for the convenience of description, the entrance and exit 300 in FIGS. 6 to 9 are enlarged and drawn, and the aforementioned door 301 is omitted.

As shown in Figures 6-9, an entry and exit datum 302 can be defined at the entrance and exit 300, where the entry and exit datum 302 can be a dividing line, but it is not limited to this. In addition, the two opposite sides of the entrance and exit 300 can be defined as a first side S1 and a second side S2. In this embodiment, the first side S1 of the entrance 300 may be the outside side of the place 30, and the second side S2 of the entrance 300 may be the inner side of the place 30. However, in another embodiment, the first side S1 of the entrance 300 can also be the inside of the place 30, and the second side S2 of the entrance 300 can also be the outside of the place 30, depending on the actual application.

When the image analysis device 10 is used to count the objects in the image sequence, first, the image capturing unit 100 captures the image sequence about the entrance and exit 300 within the field of view FOV (step S20 in FIG. 5). In this embodiment, after capturing the image sequence, the processing unit 102 analyzes the image sequence by image analysis technology to determine whether there are objects (for example, people, animals, or other objects) in the image sequence. If there is an object in the image sequence, the processing unit 102 can further analyze the movement track and position of the object in the image sequence by using image analysis technology. As shown in FIGS. 6-9, when the object 5 enters the field of view FOV, the processing unit 102 will record the entry position P1 of one of the objects 5 (step S22 in FIG. 5).

When the object 5 in the image sequence passes the entry and exit reference 302 corresponding to the entrance 300 for the first time within the FOV of the field of view, the processing unit 102 will determine one of the moving directions of the object 5 (step S24 in FIG. 5). As shown in Figures 6 and 7, when the object 5 in the image sequence passes the entry and exit reference 302 corresponding to the entrance 300 for the first time within the FOV of the field of view, the processing unit 102 determines that the movement direction of the object 5 is from the entrance 300 The first side S1 moves toward the second side S2 of the entrance 300. As shown in Figures 8 and 9, when the object 5 in the image sequence passes the entry and exit reference 302 corresponding to the entrance 300 for the first time within the FOV of the field of view, the processing unit 102 determines that the movement direction of the object 5 is from the entrance 300 The second side S2 moves toward the first side S1 of the entrance 300. It should be noted that since the entry-exit standard 302 is a dividing line, when the object 5 crosses the entry-exit standard 302, it means that the object 5 passes the entry-exit standard 302.

As shown in Figures 6 and 7, when the processing unit 102 determines that the object 5 is moving from the first side S1 of the entrance 300 to the second side S2 of the entrance 300, the processing unit 102 will count the object 5 as a first Accumulate the total and record the current position P2 of the object 5 on the second side S2 (step S26 in Fig. 5). In this embodiment, since the first side S1 of the entrance 300 is the outside side of the place 30, and the second side S2 of the entrance 300 is the inside of the place 30, it means that the object 5 passes the access reference 302 for the first time within the FOV of the field of view , From the outside of the place 30 into the inside of the place 30. Therefore, the above-mentioned first cumulative total can be the cumulative total number of places 30 entered.

In addition, when the processing unit 102 counts the object 5 in the first cumulative total, the processing unit 102 may mark that the object 5 has been included in the first cumulative total. Therefore, when the object 5 passes the entry and exit criterion 302 back and forth within the FOV of the field of view, the present invention can prevent the object 5 from being repeatedly counted into the first cumulative total.

When the object 5 leaves the field of view FOV, the processing unit 102 will record the leaving position P3 of one of the objects 5 (step S28 in FIG. 5). Next, the processing unit 102 determines whether the current position P2 and the leaving position P3 are located on two opposite sides of the entry and exit reference 302 (step S30 in FIG. 5). When the processing unit 102 determines that the current position P2 and the leaving position P3 are on two opposite sides of the entry and exit reference 302, the processing unit 102 will count the object 5 into a second cumulative total (step S32 in Figure 5). When the processing unit 102 determines that the current position P2 and the leaving position P3 are on the same side of the entry and exit reference 302, the processing unit 102 does not count the object 5 in the second cumulative total (step S34 in Fig. 5). Since the first side S1 of the entrance and exit 300 is outside the place 30, and the second side S2 of the entrance and exit 300 is the inner side of the place 30, it means that when the object 5 leaves the field of view FOV shown in FIG. 6, it is left from the place 30. Therefore, the above-mentioned second cumulative total can be the cumulative total number of departures of the place 30.

As shown in Figure 6, since the current position P2 and the leaving position P3 are located on opposite sides of the entry and exit reference 302 (that is, the object 5 has left the location 30), the processing unit 102 will count the object 5 in the second cumulative total . On the other hand, as shown in Figure 7, since the current position P2 and the leaving position P3 are on the same side of the entry and exit reference 302 (that is, the object 5 has not left the location 30), the processing unit 102 will not count the object 5 in The second cumulative total.

As shown in Figures 8 and 9, when the processing unit 102 determines that the object 5 moves from the second side S2 of the entrance 300 to the first side S1 of the entrance 300, the processing unit 102 will determine whether the entry position P1 and the exit position P3 are It is located on two opposite sides of the entry-exit reference 302 (step S36 in Fig. 5). When the processing unit 102 determines that the entry position P1 and the exit position P3 are on two opposite sides of the entry and exit reference 302, the processing unit 102 will count the object 5 into the second cumulative total (step S38 in Figure 5). When the processing unit 102 determines that the entry position P1 and the exit position P3 are on the same side of the entry and exit reference 302, the processing unit 102 does not count the object 5 in the second cumulative total (step S40 in Fig. 5).

As shown in Figure 8, since the entry position P1 and the exit position P3 are located on opposite sides of the entry and exit reference 302 (that is, the object 5 has left the location 30), the processing unit 102 will count the object 5 in the second cumulative total . On the other hand, as shown in Figure 9, since the entry position P1 and the exit position P3 are located on the same side of the entry and exit reference 302 (that is, the object 5 has not left the location 30), the processing unit 102 will not count the object 5 in The second cumulative total.

Please refer to FIG. 10, which is a flowchart of an image analysis method according to another embodiment of the present invention. The image analysis method in Figure 10 is also applicable to the image analysis device 10 in Figure 1. Except for the differences between steps S26' and S30' in Figure 10 and steps S26 and S30 in Figure 5, the other steps in Figure 10 and Figure 5 are substantially the same. Therefore, only the steps S26' and S30' in Figure 10 will be described in conjunction with Figures 6 and 7.

As shown in Figures 6 and 7, when the processing unit 102 determines that the object 5 is moving from the first side S1 of the entrance 300 to the second side S2 of the entrance 300, the processing unit 102 will count the object 5 as a first Cumulative total (step S26' in Figure 10). In other words, the difference between step S26' in Figure 10 and step S26 in Figure 5 is that step S26' in Figure 10 does not need to record the current position P2 of the object 5 on the second side S2.

When the object 5 leaves the field of view FOV, the processing unit 102 will record the leaving position P3 of the object 5 (step S28 in Fig. 10). Next, the processing unit 102 determines whether the exit position P3 is located on the first side S1 of the entry and exit reference 302 (step S30' in FIG. 10). When the processing unit 102 determines that the leaving position P3 is located at the first side S1 of the entry and exit reference 302, the processing unit 102 will count the object 5 into the second cumulative total (step S32 in Figure 10). When the processing unit 102 determines that the leaving position P3 is not located on the first side S1 of the entry and exit reference 302, the processing unit 102 does not count the object 5 in the second cumulative total (step S34 in Figure 10).

As shown in FIG. 6, because the leaving position P3 is on the first side S1 of the entry and exit reference 302 (that is, the object 5 has left the location 30), the processing unit 102 counts the object 5 into the second cumulative total. On the other hand, as shown in Figure 7, since the exit position P3 is not located on the first side S1 of the entry and exit datum 302 (that is, the exit position P3 is on the second side S2 of the entry and exit datum 302, and the object 5 has not yet left the place 30), Therefore, the processing unit 102 does not count the object 5 in the second cumulative total.

Therefore, according to the image analysis methods in FIGS. 5 and 10, the processing unit 102 can determine whether to include the object 5 in the second cumulative total according to the departure position P3.

Please refer to FIG. 11, which is a schematic diagram of an access datum 302' according to another embodiment of the present invention. As shown in FIG. 11, the access reference 302' is a counting channel, that is, the access reference 302' covers a specific range. The specific range covered by the access standard 302' can be set according to actual applications. In this embodiment, the access reference 302' may include two boundaries L1, L2, and a first boundary E1 and a second boundary E2 connecting the boundaries L1, L2. When the object 5 enters the entry and exit reference 302 ′ through any one of the boundaries L1, L2, the first boundary E1 and the second boundary E2, the processing unit 102 may set the entry position as the starting point Pa. When the object 5 leaves the entry and exit reference 302' through any one of the boundaries L1, L2, the first boundary E1, and the second boundary E2, the processing unit 102 may set the exit position as the end point Pb. In this embodiment, the object 5 enters the access datum 302' through the first boundary E1, and leaves the access datum 302' via the second boundary E2.

Then, the processing unit 102 can use the boundaries L1 and L2 to define the reference axis Ax. In this embodiment, since the two boundaries L1 and L2 are parallel and not interlaced, the reference axis Ax can be defined as any virtual line that intersects the end point Pb and is substantially parallel to the boundaries L1 and L2. Next, the processing unit 102 calculates the projection length Lp of the movement track of the object 5 in the entry and exit reference 302' relative to the reference axis Ax. In addition, the processing unit 102 may use a preset ratio length of the access reference 302' as the threshold value, for example, the threshold may be 90% of the length of the access reference 302'. Next, the processing unit 102 compares the projection length Lp with the threshold value. When the projection length Lp is greater than the threshold value, the processing unit 102 determines that the object 5 has passed the entry and exit criterion 302'. Conversely, when the projection length Lp is not greater than the threshold value, the processing unit 102 determines that the object 5 does not pass the entry and exit criterion 302'.

Please refer to Figures 12 and 13. Figure 12 is a schematic diagram of the object 5 moving relative to the entrance 300 when the door 301 is opened, and Figure 13 is a schematic diagram of the object 5 entering the place 30 and located near the door 301. In the embodiments shown in FIGS. 12 and 13, the place 30 may be a public transportation vehicle (for example, a bus). When the door 301 is opened, the object 5 can enter the place 30. As shown in Figure 12, if the object 5 enters the place 30 during the opening of the door 301 and then immediately (when the door 301 is not closed) and then leaves the place 30, the present invention may not count the object 5. As shown in FIG. 13, after the object 5 enters the place 30, it can stay near the entrance 300 inside the place 30. When the door 301 is closed, the object 5 is located inside the place 30 near the door 301. At this time, the present invention can count the object 5 into the above-mentioned first cumulative total. If the object 5 leaves the place 30 after the door 301 is opened, the present invention can include the object 5 in the above-mentioned second cumulative total. In another embodiment, if the object 5 enters the area 30 after leaving the place 30 during the opening of the door 301 (while the door 301 is not closed), the present invention may not count the object 5. After the object 5 leaves the place 30, it can stay near the entrance and exit 300 outside the place 30. When the door 301 is closed, the object 5 is located outside the place 30 near the door 301. At this time, the present invention can count the object 5 into the above-mentioned second cumulative total. In this way, it is possible to effectively increase the statistical accuracy of the number of inbound and outbound customers for public transportation vehicles.

In summary, the present invention detects the sign on the door by image analysis, and then judges whether the door is opened or closed according to the movement state of the sign on the door. Therefore, the image analysis device of the present invention does not need to receive additional digital input signals, and can effectively solve the problem of compatibility between the image analysis device and the control circuit of a specific place (for example, shops, public transportation vehicles, etc.). In addition, when the object in the image sequence passes the entry and exit criteria of the corresponding entrance, the present invention first determines the moving direction of the object, and then determines the time point for counting the object according to the moving direction of the object. In this way, the statistical accuracy of the number of customers entering and leaving a specific place can be effectively increased. The foregoing descriptions are only preferred embodiments of the present invention, and all equivalent changes and modifications made in accordance with the scope of the patent application of the present invention shall fall within the scope of the present invention.

5: Object 10: Image analysis device 30: place 100: Image capture unit 102: Processing Unit 300: entrance and exit 301: door 302, 302': Access benchmark 3010: Sign Ax: reference axis E1: The first boundary E2: Second Demarcation FOV: Field of view L1, L2: boundary Lp: projection length P1: Enter position P2: current position P3: leave position Pa: starting point Pb: End point S1: First side S2: second side S10-S14, S20-S40, S26', S30': steps

FIG. 1 is a functional block diagram of an image analysis device according to an embodiment of the invention. Figure 2 is a schematic diagram of the image analysis device in Figure 1 installed at the entrance and exit of the venue. Figure 3 is a schematic diagram of the door in Figure 2 after it is opened. Figure 4 is a flowchart of an image analysis method according to an embodiment of the invention. FIG. 5 is a flowchart of an image analysis method according to another embodiment of the invention. Figure 6 is a schematic diagram of objects moving relative to the entrance and exit. Figure 7 is another schematic diagram of objects moving relative to the entrance and exit. Figure 8 is another schematic diagram of objects moving relative to the entrance and exit. Figure 9 is another schematic diagram of objects moving relative to the entrance and exit. FIG. 10 is a flowchart of an image analysis method according to another embodiment of the invention. Figure 11 is a schematic diagram of an access standard according to another embodiment of the present invention. Figure 12 is a schematic diagram of objects moving relative to the entrance and exit when the door is opened. Figure 13 is a schematic diagram of objects entering the place and located near the door.

S10-S14: steps

Claims (12)

An image analysis method for image analysis of an entrance and exit, the entrance and exit is provided with a movable door, the door has a mark, the image analysis method includes the following steps: Capture an image sequence about the entrance and exit; Detect the mark from the image sequence; and Determine whether the door is open or closed according to the movement state of the sign. The image analysis method described in claim 1 further includes the following steps: A user interface is used to mark a part of the door in the image sequence as the mark. The image analysis method described in claim 1 further includes the following steps: Determine whether the door is opened or closed according to the position change or the moving distance of the mark in the image sequence. The image analysis method described in claim 1 further includes the following steps: When determining that the door is open, start counting an object in the image sequence; and When it is determined that the door is closed, stop counting the objects in the image sequence. The image analysis method described in claim 1 further includes the following steps: Starting to count an object in the image sequence at a predetermined time before determining that the door is opened; and When it is determined that the door is closed, stop counting the objects in the image sequence. The image analysis method described in claim 1 further includes the following steps: Capture the image sequence about the entrance and exit within a field of view; When an object in the image sequence passes through an entry and exit reference corresponding to one of the entrances and exits, judging a moving direction of the object; When it is determined that the object moves from a first side of one of the entrances and exits to a second side of the entrance and exits, the object is counted into a first cumulative total; When the object leaves the field of view, record the exit position of one of the objects; and Decide whether to count the object into a second cumulative total according to the departure position. The image analysis method according to claim 6, wherein the entry and exit criterion is a dividing line or a counting channel. The image analysis method described in claim 6 further includes the following steps: When the object is included in the first cumulative total, it is marked that the object has been included in the first cumulative total. The image analysis method described in claim 6 further includes the following steps: Determine whether the departure position is on the first side; and When it is determined that the leaving position is on the first side, the object is included in the second cumulative total. The image analysis method described in claim 6 further includes the following steps: When it is determined that the object moves from the first side of the doorway to the second side of the doorway, record the current position of the object on the second side; Determine whether the current position and the leaving position are on two opposite sides of the entry and exit datum; and When it is determined that the current position and the leaving position are on two opposite sides of the entry and exit criterion, the object is counted into the second cumulative total. The image analysis method described in claim 6 further includes the following steps: When the object enters the field of view, record the entry position of one of the objects; When judging that the object moves from the second side of the doorway to the first side of the doorway, judging whether the entry position and the exit position are on two opposite sides of the entry and exit reference; and When it is determined that the entry position and the exit position are located on two opposite sides of the entry and exit reference, the object is counted into the second cumulative total. An image analysis device for performing image analysis on an entrance and exit, the entrance and exit is provided with a movable door, the door has a sign, and the image analysis device includes: An image capturing unit for capturing an image sequence about the entrance and exit; and A processing unit, electrically connected to the image capturing unit, is used to execute the image analysis method described in one of request item 1 to request item 11 or a combination thereof.

Images