KR102514574B1 - Apparatus for detecting passenger in a vehicle and method thereof - Google Patents

Abstract

The present invention relates to an apparatus and method for detecting passengers in a vehicle capable of controlling the operation of a vehicle by determining whether or not there are occupants in the vehicle based on an image, wherein the apparatus for detecting passengers in the vehicle is installed on the upper front portion of the vehicle, and a camera that photographs both the inside and the rear of the vehicle; and receiving images of the inside and rear of the vehicle frame by frame from the camera, analyzing the input frames to determine whether there are unoccupied passengers, and setting a Region of Interest (ROI) in the frames; and a control unit that detects an object candidate within the ROI and then detects the number of weighted pixels for the object candidate to determine whether or not there is an unseated passenger.

Description

Passenger detection device and method in a vehicle {APPARATUS FOR DETECTING PASSENGER IN A VEHICLE AND METHOD THEREOF}

The present invention relates to an apparatus and method for detecting a passenger in a vehicle, and more particularly, to an apparatus and method for detecting an occupant in a vehicle capable of controlling vehicle operation by determining whether or not there is an unoccupied passenger in the vehicle based on an image.

Autonomous Vehicle refers to a vehicle that can operate on its own without driver or passenger operation, and Automated Vehicle & Highway Systems refers to a system that monitors and controls autonomous vehicles so that they can operate on their own. .

Recently, autonomous vehicles have been used as vehicles used as public transportation for transporting a large number of people, for example, subways, buses, taxis, and the like.

For example, in the case of an autonomous bus, autonomous driving is performed with a number of passengers on board, but the bus may shake or the driving stability may be lowered depending on the autonomous driving route and driving environment.

At this time, when passengers on the bus move inside the bus to move seats or get off the bus, there is a risk of safety accidents in which passengers fall or collide with structures inside the bus.

Therefore, when an autonomous vehicle transports a large number of people, there is a need for technology and research capable of ensuring the safety of passengers on board during autonomous driving.

KR 1020200060630 (published date: 2020.06.01)

The present invention has been made to improve the above problems, and an object of the present invention according to one aspect is to set a region of interest in a frame of an image captured from a camera installed in a vehicle, and detect the number of pixels of an object in the region of interest An object of the present invention is to provide an in-vehicle detection device and method capable of detecting an unseated passenger in a vehicle based on an image by detecting a standing passenger by applying a weighted value for each distance to .

In addition, an object of the present invention according to an aspect is an in-vehicle detection device capable of preventing safety accidents in the vehicle due to unseated passengers by controlling vehicle operation by determining whether there are unseated passengers in the vehicle before departure of the vehicle. and to provide a method thereof.

The technical problems to be achieved by the present invention are not limited to the above-mentioned technical problems, and other technical problems not mentioned are clear to those skilled in the art from the detailed description of the invention below. will be understandable.

An apparatus for detecting a passenger in a vehicle according to an embodiment of the present invention includes a camera installed at an upper front portion of a vehicle and photographing both the inside and the rear of the vehicle; and receiving images of the inside and rear of the vehicle frame by frame from the camera, analyzing the input frames to determine whether there are unoccupied passengers, and setting a Region of Interest (ROI) in the frames; After detecting object candidates within the ROI, a control unit that determines whether or not there are unseated passengers by detecting the number of weighted pixels for the object candidates.

The control unit sets the ROI by normalizing a passage inside the vehicle to a set size in the frame, and detects a weighted number of pixels by applying a weight to the number of detection pixels in a vertical line of an object candidate in the ROI; If the number of detected weighted pixels of the object candidate is greater than or equal to a set threshold, the object candidate is determined to be an unseated passenger, and if the number of detected weighted pixels of the object candidate is less than the set threshold, it is determined that the object candidate is not a passenger.

The controller may detect the weighted number of pixels (D _w ) through Equation 1 below.

[Equation 1]

으로 구해지는 값이고, 상기 H는 관심영역(ROI)의 수직라인을 구성하는 총 화소 개수를 각각 의미함)(Where, d _i is the number of detected pixels in the region of interest (ROI) in the i (1≤i≤H)-th vertical line of the frame, and r _i is in the i (1≤i≤H)-th vertical line of the frame. The number of pixels in the region of interest (ROI), w _i is the weight in the i (1≤i≤H)th vertical line of the frame

It is a value obtained by , and H denotes the total number of pixels constituting the vertical line of the region of interest (ROI))

The relationship between the number of detected pixels (d _i ) and the number of weighted pixels (D _w ) of the object candidate is that the vertical line (i) of the frame is d _i < w _i d _i in the range of 1≤i≤H/2, It is characterized in that the vertical line (i) of the frame is d _i > w _i d _i in the range H/2<i≤H.

The device for detecting occupants in the vehicle further includes an alarm unit configured to output an announcement to alert a risk of unseated passengers or guide seating and departure when an unseated passenger is detected in the region of interest, wherein the control unit , When an unseated passenger within the region of interest is detected, an alarm and announcement of the alarm unit are controlled, and when no unseated passenger within the region of interest is detected, a control signal instructing departure of the vehicle is output. .

According to an embodiment of the present invention, it is possible to improve the detection performance of unseated passengers by determining whether or not there is an unseated passenger by applying a weight according to the separation distance from the camera based on the image. And as the accuracy is improved, it is possible to prevent safety accidents caused by vehicle operation due to non-detection of unseated passengers.

In addition, according to an embodiment of the present invention, low-cost installation is possible by accurately detecting an unseated passenger based on an image without having a separate sensor for detecting an unseated passenger, and an alarm is provided to the detected unseated passenger before departure. By doing so, you can promote safe vehicle operation.

The effects obtainable in the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below. .

1 is a block diagram illustrating a detailed configuration of an in-vehicle passenger detecting device according to an embodiment of the present invention.
2 is an exemplary diagram for explaining an example in which a camera is installed inside a vehicle in the device for detecting a passenger in a vehicle according to an embodiment of the present invention.
3 is a diagram illustrating a region of interest for detecting a passenger and a difference in the number of pixels for each distance in the device for detecting a passenger in a vehicle according to an embodiment of the present invention.
4 is a diagram for explaining object detection using a weighted number of pixels according to a distance from a camera in the device for detecting a passenger in a vehicle according to an embodiment of the present invention.
5 to 7 are diagrams illustrating examples of results of detecting a passenger in a region of interest in the apparatus for detecting a passenger in a vehicle according to an embodiment of the present invention.
8 is a flowchart illustrating a method for detecting a passenger in a vehicle according to an embodiment of the present invention.

Like reference numbers designate like elements throughout the specification. This specification does not describe all elements of the embodiments, and general content or overlapping content between the embodiments in the technical field to which the present invention belongs is omitted. The term 'unit, module, member, or block' used in the specification may be implemented as software or hardware, and according to embodiments, a plurality of 'units, modules, members, or blocks' may be implemented as one component, It is also possible that one 'part, module, member, block' includes a plurality of components.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case of being directly connected but also the case of being indirectly connected, and indirect connection includes being connected through a wireless communication network. do.

In addition, when a certain component is said to "include", this means that it may further include other components without excluding other components unless otherwise stated.

Terms such as first and second are used to distinguish one component from another, and the components are not limited by the aforementioned terms.

Expressions in the singular number include plural expressions unless the context clearly dictates otherwise.

In each step, the identification code is used for convenience of description, and the identification code does not explain the order of each step, and each step may be performed in a different order from the specified order unless a specific order is clearly described in context. there is.

Hereinafter, the working principle and embodiments of the present invention will be described with reference to the accompanying drawings.

1 is a block diagram illustrating a detailed configuration of an in-vehicle passenger detecting device according to an embodiment of the present invention.

2 is an exemplary diagram for explaining an example in which a camera is installed inside a vehicle in the device for detecting a passenger in a vehicle according to an embodiment of the present invention.

A vehicle according to an embodiment of the present invention is an autonomous vehicle, and a bus is described as an example, but is not limited thereto and may be applied to other types of vehicles.

First, referring to FIG. 1 , an in-vehicle passenger detecting apparatus 10 according to an embodiment of the present invention includes a camera 11 that photographs the rear of the vehicle and outputs image information, and a camera 11 photographed from the camera 11. It may be configured to include at least a control unit 12 that receives image information and determines whether there are unseated passengers.

The in-vehicle passenger detecting device 10 according to an embodiment of the present invention may be electrically connected to the vehicle driving system 100 or connected through a wireless network. Through this, the in-vehicle detecting device 10 according to the present invention transmits passenger detection information to the vehicle drive system 100 so that driving related to departure can be performed.

Here, the vehicle may be an autonomous bus that drives to a destination by itself without manipulation by a driver, and passengers inside the bus may include passengers sitting in seats and unseated passengers standing without sitting in seats.

In addition, the camera 11 may acquire image information by photographing the inside of the vehicle, that is, the rear, and may transmit the acquired image information to the control unit 12 in units of frames.

As shown in FIG. 2 , the camera 11 may be installed on the front upper part of the vehicle interior, that is, on the ceiling, so as to photograph the front and rear of the vehicle 1, but the present invention is not limited thereto. The installation location of (11) can be installed anywhere in the vehicle where a field of view (FOV) can be secured in order to acquire front and rear image information to detect unoccupied passengers by taking pictures of the inside of the vehicle. .

In addition, the camera 11 may be implemented as a camera having a field of view (FOV) capable of capturing all of the central aisle or corridor of the vehicle from the installation location of the camera 11 to the rear, for example, It may be a wide-angle camera supporting more than 130 degrees, or a camera using a fish-eye lens for a high angle of view.

In addition, when the image information photographed by the camera 11 is transmitted in frame units, the controller 12 performs image processing by applying an image processing technique to the transmitted frame, and the distance from the location point of the camera 11 By detecting the number of pixels of an object in a frame using a weight according to , it is possible to determine whether or not there is an unseated passenger.

In one embodiment of the present invention, determining whether or not there is an unseated passenger is essential for the vehicle driving system 100 to improve the safety of unseated passengers immediately before departure.

Specifically, the controller 12 performs subsampling on the input frame. Subsampling may use various methods such as bilinear interpolation, nearest nearest value method, and cubic convolutional interpolation, but is not limited thereto, and various known techniques may be adopted and applied. That is, if the frame image input through sampling is a color image, it is converted into a black-and-white image, and if it is a black-and-white image, it can be used as it is. That is, by performing human detection using a black-and-white image, it is possible to reduce the effect of lighting as well as the processing time.

Then, the control unit 12 performs object candidate detection from the subsampling frame, sets a Region of Interest (ROI) in the subsampling frame, and moves within the ROI. Detect object candidates according to the number of pixels.

Here, object detection is a computer vision technology for accurately finding the type and location (bounding box) of a meaningful object such as a person, animal, or vehicle in an image or video. In order to recognize a target of interest, a candidate region for a detection target is generally found and the type and location of an object in the candidate region is predicted through a learned model. As a method for detecting such an object, a machine learning-based approach or a deep learning-based approach object detection technique is generally used, and these are all well-known techniques, and detailed descriptions thereof are omitted in the present invention.

In addition, the region of interest (ROI), in one embodiment of the present invention, is an area including a central aisle or corridor in the vehicle to detect unseated passengers, or designates a specific area to detect unseated passengers. may be set by

The region of interest (ROI) may be set by normalizing the central passage of the vehicle in the frame to a set size with respect to the inside of the vehicle captured by the camera 11 .

In addition, the control unit 12 performs noise cancellation on the object candidate and determines whether or not there is an unseated passenger on the object candidate within the region of interest (ROI).

At this time, the control unit 12 may consider the distance from the installation position of the camera 11 to the object in order to accurately determine an unseated passenger. For example, the same object based on the installation position of the camera 11 has a large number of pixels in the near area, but has a small number of pixels in the far area. In particular, the number of pixels in the far area is too small. In this case, object detection may not be performed because detection is not possible.

In this way, based on the camera 11, in the case of a short-distance passenger, the detection is good due to the large number of pixels, but in the case of a long-distance passenger, the detection is deteriorated due to the small number of pixels.

Therefore, according to an embodiment of the present invention, the control unit 12 applies a weight to the number of detected pixels of object candidates in the region of interest in order to compensate for the difference in the number of pixels according to the position to the object, and thus detects the distance from the camera 11. The number of weighted pixels according to the distance of is obtained, and the presence or absence of unseated passengers is determined using the obtained number of weighted pixels. Through this, the detection efficiency of distant passengers can be improved.

In other words, the controller 12 applies the weighted number of pixels for object determination to the number of detected pixels in the vertical line (i, 1≤i≤H) in the region of interest (ROI) to obtain Equation 1 and can be obtained together.

[Equation 1]

d _i is the number of detected pixels in the region of interest (ROI) in the i (1≤i≤H)-th vertical line of the frame, and r _i is the region of interest (ROI) in the i (1≤i≤H)-th vertical line of the frame ( ROI), w _i is the weight in the i (1≤i≤H) th vertical line of the frame, and H is the total number of pixels constituting the vertical line of the region of interest (ROI), which is calculated by the following equation It can be obtained using Equation 2.

[Equation 2]

, 1≤i≤H

, 1≤i≤H

Accordingly, the control unit 12 compares the detection result of the number of weighted pixels (DR _w ) of the object candidate with the threshold value (Thd), but determines the object according to Equation 3 below. For example, if the number of weighted pixels (D _w ) obtained using Equation 1 is greater than or equal to a preset threshold value (Thd), the control unit 12 determines that the object is an object, and the number of weighted pixels obtained is the preset threshold value. If it is less than, it is determined that it is not an object (Non-Object).

[Equation 3]

In addition, the control unit 12 may control the operation of the vehicle driving system 100 and the alarm unit 13 by using information on the presence of unseated passengers.

For example, if it is determined that there are unseated passengers before the departure of the vehicle, the control unit 12 requests the vehicle drive system 100 to delay the departure of the vehicle, and at the same time sends a request to the alarm unit 13 for unseated passengers. An announcement may be made to induce seating or hold on to the handrail for departure. Here, the alarm unit 13 may include a speaker capable of outputting voice guidance information.

For another example, if it is determined that there are no occupants before departure of the vehicle, the control unit 12 may notify the vehicle drive system 100 so that the vehicle departs and travels.

As described above, the device for detecting passengers in a vehicle according to an embodiment of the present invention determines whether or not there are unoccupied passengers using the weighted number of pixels obtained by applying weights to the number of detected pixels of object candidates in the region of interest, thereby detecting short-distance passengers within the frame. In addition, the detection efficiency of distant passengers can be improved.

In addition, the device for detecting occupants in a vehicle according to an embodiment of the present invention detects the presence or absence of an unseated passenger before the vehicle departs, thereby inducing the unseated passenger to sit or recognizing the departure of the unseated passenger, thereby preventing shaking caused by departure. It enables you to be prepared to prevent safety accidents in the vehicle and enable safe driving.

3 is a diagram illustrating a region of interest for detecting a passenger and a difference in the number of pixels for each distance in the device for detecting a passenger in a vehicle according to an embodiment of the present invention.

Referring to FIG. 3, in order to detect unoccupied seats, a region of interest (ROI) 22 is designated as a location matched to the central aisle of the vehicle in the frame 20, and the same object in the region of interest (ROI) 22 ( 21) can be seen that there is a difference in the number of pixels in the image according to the positions (P1, P2, P3).

4 is a diagram for explaining object detection using a weighted number of pixels according to a distance from a camera in the device for detecting a passenger in a vehicle according to an embodiment of the present invention.

Referring to FIG. 4 , the control unit 12 of the device for detecting passengers in a vehicle according to an embodiment of the present invention detects each object candidate 31, 33, 35 and 37) by applying a weight (w _i ) to the number of detected pixels (d _i ) to obtain the number of weighted pixels (D _w ), it is possible to determine whether an object exists for each object candidate (31, 33, 35, 37). can

That is, the control unit 12 applies the weighted number (D _w ) of the vertical line (i) in the region of interest (ROI) 32 to the number of weighted pixels (D _w ) for determining the presence or absence of an object to obtain Equation 1 below and can be obtained together.

[Equation 1]

d _i is the number of detected pixels in the region of interest (ROI) in the i (1≤i≤H)-th vertical line of the frame 30, and r _i is the i (1≤i≤H)-th vertical line of the frame 30 The number of pixels in the region of interest (ROI) in a line, and w _i is a weight in the i (1≤i≤H) th vertical line of the frame 30, which can be obtained using Equation 2 below.

[Equation 2]

, 1≤i≤H

, 1≤i≤H

Here, according to an embodiment of the present invention, the relationship between the number of detected pixels (d _i ) and the number of weighted pixels (D _w ) of the object candidate is such that the vertical line (i) of the frame is in the range of 1≤i≤H/2 d _i < w _i d _i , and the vertical line (i) of the frame satisfies d _i > w _i d _i in the range of H/2<i≤H. Due to this relationship, when detecting a distant object within a frame, the number of pixels in which the number of weighted pixels (D _w ) of the object candidate is greater than the number of detection pixels (d _i ) of the object candidate is detected, so that the performance of detecting a distant passenger can be improved.

Therefore, the control unit 12 compares the weighted pixel number detection result (DR _w ) and the threshold value (Thd) for the object candidates 31, 33, 35, and 37, but performs object determination according to Equation 3 below do. For example, if the number of weighted pixels (D _w ) obtained using Equation 1 is greater than or equal to a preset threshold value (Thd), the control unit 12 determines that the object is an object, and the number of weighted pixels obtained is the preset threshold value. If it is less than, it is determined that it is not an object (Non-Object).

[Equation 3]

On the other hand, (Comparative Example) In the case of object determination using only the number of detection pixels (d _i ) of object candidates within the ROI in the vertical line of the frame, the region of interest (ROI) 32 of the frame 30 ), the number of detected pixels (d _i ), and object determination may be performed by comparing the object detection result (DR) with a threshold value as shown in Comparative Equation 2.

[Comparative Formula 1]

[Comparative Formula 2]

That is, when comparing the comparative example and the present invention, as shown in Table 1, in the case of the comparative example, the object candidate 37 located the farthest among the object candidates is determined as a non-object, whereas the present invention determines the farthest among the object candidates. The located object candidate 37 is determined as an object.

object candidate Threshold (Thd) DR DRw 37
35 Non-Object Object 33 Object Object 31 Object Object

5 to 7 are diagrams illustrating examples of results of detecting a passenger in a region of interest of a frame in an apparatus for detecting a passenger in a vehicle according to an embodiment of the present invention.

5(a) is an image frame in which a camera captures a situation in which passengers P2 and P3 in a vehicle pass through a central aisle. As shown in (a) of FIG. 5, in the frame input from the camera, a passenger P1 seated in a seat and passengers P2 and P3 standing in the aisle exist, and a region of interest (ROI) is set in the frame. Set up.

FIG. 5(b) is an image frame of a passenger detection result within a region of interest (ROI) of the image frame shown in FIG. 5(a). As shown in (b) of FIG. 5 , regions other than the region of interest (ROI) are excluded from object detection, and passengers P2 and P3 standing in the region of interest (ROI) are determined as objects.

FIG. 5(c) is an image frame in which a camera captures a situation in which passengers P1, P2, and P3 are all sitting in their seats. Here, THD in a frame denotes a threshold value. As shown in (c) of FIG. 5 , only passengers P1 , P2 , and P3 previously seated in seats exist in the frame input from the camera, and a region of interest (ROI) is set in the frame.

(d) of FIG. 5 is an image frame of a passenger detection result within a region of interest (ROI) of the image frame shown in (c) of FIG. 5 . As shown in (d) of FIG. 5 , it can be seen that there is no effective number of detection pixels in the region of interest (ROI).

And, as shown in (a) to (c) of FIG. 6, for a motion object candidate (marked with a red line) having a smaller number of pixels than the threshold value (Thd) in the region of interest (marked with a yellow line) of the frame. It can be seen that passenger detection is possible by obtaining the number of weighted pixels.

And, as shown in (a) of FIG. 7, by calculating the number of weighted pixels for moving object candidates (marked by a red line) in which the passenger moves backward along the central aisle in the region of interest (marked by a yellow line) of the frame, the passenger can be detected.

Similarly, as shown in (b) of FIG. 7, by calculating the number of weighted pixels for motion object candidates (marked with a red line) in which a passenger finds his/her seat and sits in the region of interest (marked with a yellow line) of the frame, A passenger not sitting in a seat can be detected. After that, the object is not detected after the passenger completely sits down.

8 is a flowchart illustrating a method for detecting a passenger in a vehicle according to an embodiment of the present invention.

Referring to FIG. 8 , in the device for detecting passengers in a vehicle according to an embodiment of the present invention, first, a camera installed on the front upper portion (eg, ceiling) of the vehicle captures the entire inside of the vehicle, that is, the entire front and rear sides of the vehicle at once. And, the image captured by the camera may be input to the controller in units of frames (S41). Here, the camera may preferably be a fixed camera, but is not limited thereto, and may be implemented as a pan-tilt-zoom camera capable of adjusting a photographing direction to a desired position.

Next, a region of interest (ROI) for detecting unseated passengers is set in the frame (S42). In one embodiment of the present invention, since a bus is described as a vehicle, a central aisle of the bus may be set as a region of interest (ROI). Here, the region of interest (ROI) is set by normalizing the passage inside the vehicle to a set size in the frame, and its shape may be set according to the shape of the passage.

Subsequently, subsampling is performed on the input frame (S43), and a motion object candidate within the region of interest (ROI) is detected (S44).

Then, noise cancellation is performed on the detected motion object candidate, and the number of weighted pixels according to the distance from the camera is detected for the noise-cancelled motion object candidate (S45).

In one embodiment of the present invention, the number of weighted pixels is determined when the number of pixels of motion object candidates in the region of interest (ROI) is detected. Passenger detection efficiency in a remote area can be improved by calculating the number of moving pixels of an object by applying a weight to compensate for this.

That is, it is easy to detect a short-distance passenger based on the camera because it has a large number of pixels, whereas it is difficult to detect a long-distance passenger because it has a small number of pixels. Accordingly, in one embodiment of the present invention, the accuracy of object detection in a far area can be increased by applying a weight to the number of detected pixels in a corresponding vertical line.

Specifically, the weighted number of pixels (D _w ) for object determination can be obtained as shown in Equation 1 below by applying the weight (w _i ) of the vertical line (i) in the region of interest (ROI).

Here, d _i is the number of detected pixels in the region of interest (ROI) in the i (1≤i≤H)-th vertical line of the image, and r _i is the interest in the i (1≤i≤H)-th vertical line of the image The number of pixels in the region (ROI), and w _i is a weight in the i (1≤i≤H)-th vertical line of the image, which can be obtained using Equation 2 below.

Therefore, in the present invention, as shown in Equation 3, if the number of weighted pixels (D _w ), which is the detection result for the object candidate, is equal to or greater than the predetermined threshold value (Thd) (S46, Y), it is determined as an object (Object) (S47 ), if the number of weighted pixels obtained is less than a preset threshold value (S46, N), it is determined that the object is not (Non-Object) (S48).

As described above, according to the method for detecting occupants in a vehicle according to an embodiment of the present invention, safety accidents caused by unseated passengers in the vehicle are prevented by determining the departure signal of the vehicle after determining whether or not there is an unseated passenger based on an image. It can be prevented.

In addition, according to the present invention, the number of moving pixels of an object is detected from a frame of a captured image, but the number of weighted pixels obtained by applying a weight for each distance based on the position of a camera that captures the image is obtained, thereby detecting not only the near object but also the distant object. Detection accuracy and efficiency can be improved.

Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium storing instructions executable by a computer. Instructions may be stored in the form of program codes, and when executed by a processor, create program modules to perform operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

Computer-readable recording media include all types of recording media in which instructions that can be decoded by a computer are stored. For example, there may be read only memory (ROM), random access memory (RAM), magnetic tape, magnetic disk, flash memory, optical data storage device, and the like.

As above, the disclosed embodiments have been described with reference to the accompanying drawings. Those skilled in the art to which the present invention pertains will understand that the present invention can be implemented in a form different from the disclosed embodiments without changing the technical spirit or essential features of the present invention. The disclosed embodiments are illustrative and should not be construed as limiting.

10. In-vehicle detection device 11. Camera
12. Control unit 13. Alarm unit
100. Vehicle drive system

Claims (5)

a camera installed on the upper front portion of the vehicle and taking pictures of both the inside and the rear of the vehicle; and
An image captured from the inside and rear of the vehicle is input from the camera frame by frame, the input frame is analyzed to determine whether or not there are unoccupied passengers, and a region of interest (ROI) is set in the frame. After detecting an object candidate in the region of interest, a control unit that determines whether or not there is an unseated passenger by detecting the number of weighted pixels for the object candidate;
The control unit,
In the frame, a passage inside the vehicle is normalized to a set size to set the ROI, and a weighted number of pixels is detected by applying a weight to the number of detection pixels in a vertical line of an object candidate in the ROI;
If the number of detected weighted pixels of the object candidate is greater than or equal to a set threshold, the object candidate is determined to be an unseated passenger, and if the number of detected weighted pixels of the object candidate is less than the set threshold, it is determined that the object candidate is not a passenger. My passenger detection device. delete According to claim 1,
The control unit detects the weighted pixel number (D _w ) through Equation 1 below
[Equation 1]

(Where, d _i is the number of detected pixels in the region of interest (ROI) in the i (1≤i≤H)-th vertical line of the frame, and r _i is in the i (1≤i≤H)-th vertical line of the frame. The number of pixels in the region of interest (ROI), w _i is the weight in the i (1≤i≤H)th vertical line of the frame

A value obtained by , where H denotes the total number of pixels constituting a vertical line of the region of interest (ROI). According to claim 3,
The relationship between the number of detected pixels (d _i ) and the number of weighted pixels (D _w ) of the object candidate is that the vertical line (i) of the frame is d _i < w _i d _i in the range of 1≤i≤H/2, In-vehicle detection device, characterized in that the vertical line (i) of the frame is d _i > w _i d _i in the range H / 2 < i ≤ H. According to claim 4,
The device for detecting occupants in the vehicle further includes an alarm unit that, when an unseated passenger within the region of interest is detected, outputs an announcement to alert a risk of unseated seats or guide seating and departure,
The control unit controls an alarm and announcement of the alarm unit when an unseated passenger is detected in the region of interest, and outputs a control signal instructing departure of the vehicle when no unseated passenger is detected in the region of interest. In-vehicle detection device characterized by.

Images