Device and method for determining when a safety belt is fastened

Abstract

Device for determining the fastening of a safety belt, comprising: an image sensor unit (11) adapted to capture a front image of a vehicle occupant; a section detection unit (12) adapted to detect a chest section and a pelvic section of the occupant using the frontal image; an image segmentation unit (13) adapted to obtain a first target area relating to a seat belt by segmenting a partial image of the occupant's chest section into a plurality of areas and to obtain a second target area relating to the seat belt by segmenting a partial image of the pelvic section into a plurality of areas; and a control unit (14) adapted to determine that the seat belt is fitted normally when a first area value of the first target area is greater than a preset first reference value and a second area value of the second target area is greater than a preset second reference value, where the first area value of the first target area is an area of ​​a section in which a shoulder strap overlaps a shoulder and chest area of ​​the occupant, and the second area value of the second target area is an area of ​​a section in which a lap belt overlaps the occupant's pelvic area, the control unit is adapted to determine that the seat belt is fitted normally when the area of ​​the section in which the shoulder belt rests against the shoulder and the chest area of ​​the occupant is larger than the preset first reference value and at the same time the area of ​​the section in which the lap belt overlaps the occupant's pelvic area is larger than the preset second reference value, and where the control unit is adapted to determine that the seat belt is abnormally fitted if at least one of the surfaces of the section where the shoulder belt overlaps the occupant's shoulder and chest area is smaller or is equal to the first reference value, or the area of ​​the section in which the lap belt overlaps the occupant's pelvic section is less than or equal to the second reference value.

Description

Field of the invention

[0001] The present invention relates to a device and a method for determining the fastening of a safety belt, in particular a device and a method for determining the fastening of a safety belt, which is used with a 3-point safety belt in a vehicle. Discussion of the state of the art

[0002] In general, a vehicle seatbelt warning system is a system that determines whether an occupant is in a vehicle and wearing a seatbelt, and transmits this information to an electronic control unit. The electronic control unit then warns the occupant by sounding an alarm, using the information provided by the vehicle seatbelt warning system, if the occupant is in the vehicle and not wearing a seatbelt.

[0003] For example, a vehicle seatbelt warning system mat equipped with a sensor is installed on a seat in a vehicle. It determines whether an occupant is in the vehicle and whether a seatbelt is fastened using a buckle switch, and transmits this result to an electronic control unit. If the electronic control unit determines, based on the transmitted result, that the occupant is in the vehicle but not wearing a seatbelt, the electronic control unit warns the occupant by means of a warning light, an audible signal, or similar means.

[0004] Alternatively, a mat-shaped structure is installed on the back frame of the upper section of a vehicle seat. When an occupant is present in the vehicle, the structure determines whether the occupant is an adult, a child, or not present and whether a seat belt is fastened using a belt buckle switch, and transmits this determination to an electronic control unit. If the electronic control unit determines that the occupant is in the vehicle but not wearing a seat belt, it issues a warning signal, such as a light, an audible alarm, or similar indicator.

[0005] However, since drivers who drive vehicles for extended periods of time in the passenger transport industry, such as in the bus or taxi industry, often feel uncomfortable wearing a seatbelt, there are instances where a driver fastens a seatbelt incorrectly, for example by using the wrong buckle or by fastening the seatbelt behind their back.

[0006] The method described above is problematic insofar as, even if the wrong lock is used or the belt is not fitted correctly, a state in which an occupant is in the vehicle is detected, and a lock sensor of a belt component is also recognized as being in an "on" state.

[0007] Furthermore, a method is used that employs deep learning to determine whether an occupant is wearing a seatbelt based on an area of ​​exposed seatbelt in a captured image of the occupant. However, such a method can be problematic because even when a seatbelt is in a non-fastened state, such as when it is parked, a large area of ​​exposed belt can lead to the misconception that the seatbelt is fastened. Additionally, if a diagonal section of a 3-point seatbelt is worn behind the back, there is the problem that an entire section of the lap belt is exposed, potentially creating the impression that the belt is fastened.

[0008] US Patent 2021 / 0206 344 A1 concerns a method for detecting whether a seat belt is being used (or worn or fastened) in a vehicle, whereby a multitude of partial images are determined from an image taken inside a vehicle, and it is determined whether a part of a seat belt is present in each partial image. Furthermore, the position of at least one body key point (of a vehicle occupant who is supposed to fasten the seat belt) is determined, and the multitude of partial images is determined based on the position of each body key point.

[0009] AT 523 045 B1 relates to a method for seatbelt detection, whereby a spatial path of a seatbelt is extracted from an image of the seat and the person, and the correctness of the seatbelt's position is assessed by comparing a calculated body data model with a seatbelt data model. A data processing unit combines the seatbelt data model and the body data model into a seat data pattern, which is compared with correct reference data patterns stored in a database to determine whether the seatbelt is correctly positioned. SUMMARY OF THE INVENTION

[0010] The present invention is directed to provide a device and a method for determining whether a seat belt is fastened, which is capable of significantly improving the accuracy and reliability of determining whether a seat belt is fastened.

[0011] Furthermore, the present invention is directed to provide a device and a method for determining the fastening of a seat belt, which is capable of preventing misunderstandings about the fastening of the seat belt due to a faulty buckle sensor or improper fastening of the belt.

[0012] These tasks are solved by a device according to claim 1 and a method according to claim 10. Preferred embodiments are specified in the dependent claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above and other tasks, features and advantages of the present invention will become even clearer to the person skilled in the art from the detailed description of exemplary embodiments with reference to the accompanying drawings, in which: Fig. 1 is a block diagram showing the structure of a device for determining the fastening of a seat belt; Fig. 2 to 5 views describing the operation of a device for determining the fastening of a safety belt according to one embodiment are; and Fig. Figures 6 to 9 are flowcharts that show procedures for determining the fastening of a safety belt according to one embodiment. DETAILED DESCRIPTION OF EXAMPLE EXECUTIONS

[0014] Exemplary embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0015] However, the present invention is not limited to some of the embodiments described.

[0016] Furthermore, all terms used herein (including technical and scientific terms), unless clearly and specifically defined differently by the context, may be interpreted as having a meaning generally understood by experts, and the meanings of commonly used terms, such as those defined in commonly used dictionaries, are interpreted taking into account the context-dependent meanings of the related state of the art.

[0017] Furthermore, the terms used in the embodiments of the present invention are to be understood in a descriptive sense only and are not intended to limit the present invention.

[0018] In the present description, unless the context clearly indicates otherwise, singular forms include the plural forms thereof, and in a case where "at least one (or one or more) of A, B and C" is described, this may include at least one combination of all possible combinations of A, B and C.

[0019] Furthermore, terms such as “first”, “second”, “A”, “B”, “(a)” and “(b)” may be used when describing components of the present invention.

[0020] Furthermore, when an element is described as being “connected or coupled” to another element, such a description can include both a case in which the element is directly connected or coupled to another element, and a case in which the element is connected or coupled to another element with another element placed in between.

[0021] When an element is described as being formed or arranged "on or under" another element, such a description includes both the case where the two elements are formed or arranged in direct contact with each other, and the case where one or more other elements are arranged between the two elements. When an element is described as being formed "on or under" another element, such a description may also include the case where one element is formed on a top or bottom surface with respect to another element.

[0022] In the following detailed description of the exemplary embodiments of the invention, with reference to the accompanying drawings, components that are the same or correspond to each other are designated with the same reference numerals, irrespective of the figure numbers, and redundant descriptions are omitted.

[0023] Fig. Figure 1 is a block diagram showing the structure of a device for determining the fastening of a seat belt, and Fig. Figures 2 to 5 are views describing the operation of a device for determining the fastening of a safety belt according to the embodiment.

[0024] According to Fig. 1 to 5, a device 10 for determining the fastening of a safety belt according to the embodiment comprises an image sensor unit 11, a section detection unit 12, an image segmentation unit 13 and a control unit 14 and may further comprise a notification unit 15 and a database 16.

[0025] In this embodiment, a safety belt can be a 3-point safety belt that encloses one shoulder, the pelvis and the abdomen of a vehicle occupant.

[0026] The image sensor unit 11 can capture a front image of the vehicle occupant.

[0027] The image sensor unit 11 can capture the frontal image of the vehicle occupant according to a preset period and transmit the frontal image to the section detection unit 12. In this embodiment, the frontal image of the occupant can be captured in such a way that it encompasses the entire upper body of the vehicle occupant, as shown in Fig. 2 shown.

[0028] The image sensor unit 11 can be an image sensor that captures an image of an object using a complementary metal-oxide semiconductor module, a charge-coupled device, or the like. In this case, an input image frame is fed to the complementary metal-oxide semiconductor module or the charge-coupled device module in the image sensor unit 11 through a lens, and the complementary metal-oxide semiconductor module or the charge-coupled device module converts an optical signal for the object passing through the lens into an electrical signal (image data) and outputs the converted electrical signal (image data).

[0029] The image sensor unit 11 can include a fisheye lens or a wide-angle lens with a large viewing angle. Accordingly, the image sensor unit 11 can capture the entire upper body of the occupant.

[0030] Alternatively, the image sensor unit 11 can comprise a first image acquisition unit 111 for capturing an image of a shoulder belt section and a second image acquisition unit 112 for capturing an image of a lap belt section. The first image acquisition unit 111 and the second image acquisition unit 112 can be arranged in a central panel of the vehicle. A lens of the first image acquisition unit 111 can be adjusted to face the shoulder belt section, and a lens of the second image acquisition unit 112 can be adjusted to face the lap belt section. The direction in which the lens of the first image acquisition unit 111 points and the direction in which the lens of the second image acquisition unit 112 points can be adjusted by controlling the control unit 14.

[0031] The first image acquisition unit 111 and the second image acquisition unit 112 can generate a multitude of images using a multitude of sequentially acquired images. For example, the first image acquisition unit 111 can acquire a first image encompassing the occupant's shoulder belt section, and the second image acquisition unit 112 can acquire a second image encompassing the lap belt section. The image acquisition units can transmit first and second images, acquired sequentially according to the set period, to the section acquisition unit 12.

[0032] In this embodiment, the image can be image data with pixels of N×M (N and M are natural numbers).

[0033] The section detection unit 12 detects a chest section and a pelvic section of the occupant using an image.

[0034] According to Fig. 3 and Fig. 4. The section detection unit 12 can detect the thoracic and pelvic sections of the occupant using a method for estimating human posture.

[0035] For example, the section acquisition unit 12 can acquire the occupant's chest section by cutting out an image of an outboard shoulder, an image of an inboard pelvis, and an image of a central area connecting the outboard shoulder and the inboard pelvis from an image acquired by the first image acquisition part 111 using the human posture estimation method.

[0036] In addition, the section acquisition unit 12 can, for example, acquire the occupant's pelvic section by extracting an image of an inboard pelvis, an image of an outboard pelvis, and an image of a central area connecting the inboard and outboard pelvises from an image acquired by the second image acquisition unit 112 using the human posture estimation method.

[0037] In this embodiment, the method for estimating human posture is an algorithm for capturing the position and orientation of an object using a computer vision method. It can locate and estimate the positions of key points, which are joints of the human body, in order to determine how these key points are composed. Key points of the human body can include the head, neck, shoulders, elbows, wrists, hips, pelvis, knees, ankles, and the like. The section capture unit 12 can capture a person from an image, estimate a posture within a drawing box, and extract an image of the occupant's chest section and an image of the pelvic section as key point sections.

[0038] In describing an example of a driver's seat with reference to Fig. 4. When the section detection unit 12 detects a driver's seat shoulder belt, the section detection unit 12 can extract an image in which the belt connects coordinates of sections 17-6-1, and the image segmentation unit 13 can obtain a first target area for detecting the shoulder belt using the extracted image. Additionally, the section detection unit 12 can extract an image in which the belt connects coordinates of sections 1-3-2, or an image in which the belt connects coordinates of sections 1-0-2, and the image segmentation unit 13 can obtain a second target area for detecting a lap belt using the extracted image.

[0039] In the case of a passenger seat, the section detection unit 12 can obtain a first target area by cutting out an image in which a belt connects coordinates of sections 16-6-2; additionally, the section detection unit 12 can cut out an image in which the belt connects coordinates of sections 1-3-2, or an image in which the belt connects coordinates of sections 1-0-2; and the image segmentation unit 13 can obtain a second target area for detecting a lap belt using the cut-out image.

[0040] Furthermore, as another example, an image of a shoulder girdle segment can be cut out by estimating the coordinates of a central segment, such as the coordinates of segments 12, 13, or the like, instead of using the coordinates of segments 16 and 17, which are positioned at either end of the shoulders. In this case, the segment acquisition unit 12 can generate virtual coordinates of segments 16 and 17 by projecting an X-coordinate contained in the coordinates of segments 1 and 2 and a Y-coordinate contained in the coordinates of a central segment of the shoulder. Similarly, an image of the pelvic girdle segment can be cut out by estimating a central segment of the abdomen, such as the coordinates of segment 0, instead of the coordinates of segments 1 and 2.In this case, the section acquisition unit can generate 12 virtual coordinates of sections 1 and 2 by projecting an X-coordinate contained in the coordinates of sections 16 and 17 and a Y-coordinate contained in the coordinates of the central section of the belly.

[0041] Alternatively, the section detection unit 12 can detect an outline of an object in an image, compare the detected outline with a chest section and a pelvic section of a person pre-stored in the database 16, and detect an area with an outline matching the chest section and pelvic section of the person as a chest section and pelvis of an occupant.

[0042] Furthermore, the section detection unit 12 can, for example, extract feature points of an object in an image, and if the extracted feature points match the feature points of a chest and pelvic section of a person pre-stored in database 16 with a similarity greater than or equal to a threshold, the corresponding area can be recognized as the chest and pelvic section of an occupant. In this case, the section detection unit 12 can extract feature points from images of two objects that are comparison targets and use a scale-invariant feature transformation or accelerated robust feature algorithm that matches feature point descriptors of the two extracted objects.

[0043] Furthermore, the section capture unit 12 can, for example, capture a chest section and a pelvic section of an occupant based on the outlines of objects in an image. More precisely, the section capture unit 12 can generate an edge image by capturing the outlines of the objects from the image, generate a background edge image by capturing the outline from foreground image data, which is an image of the interior of a vehicle prestored in database 16, and capture the chest section and pelvic section of the occupant from another image obtained by subtracting the background edge image from the edge image. In this case, the section capture unit 12 generates the edge image by capturing the outline of the object captured in a frame as an edge using gradient information from an image data frame.In this case, the gradient information is a value generated from a difference in value between neighboring pixels among predetermined pixels in the frame and is a sum of absolute values ​​of differences, and the edge is a boundary line between the objects obtained using the gradient information.

[0044] Furthermore, the section detection unit 12 can detect a chest section and a pelvic section of an inmate from an image using an object detection classifier. In this case, the object detection classifier is a classifier trained using a training database pre-constructed using captured images of various inmate body postures or different external environments. The object detection classifier generates a database of chest and pelvic sections of inmates through various learning algorithms, including a support vector machine algorithm, a neural network algorithm, an AdaBoost algorithm, and the like.In particular, the section detection unit 12 can detect an edge of an object corresponding to a foreground from a background image captured in advance from the interior of a vehicle, apply the edge of the foreground object captured from image data, and apply the object detection classifier to an area of ​​the image data to which the edge of the foreground object is applied in order to detect the chest section and pelvic section of the occupant.

[0045] Furthermore, the section detection unit 12 can reduce noise in an image captured by the image sensor unit 11 and perform image signal processing, such as gamma correction, color filter array interpolation, color matrix, color correction, and color enhancement, to improve image quality. In addition, the section detection unit 12 can also perform color processing, blur processing, edge enhancement processing, image analysis processing, image recognition processing, image effects processing, and the like.

[0046] The image segmentation unit 13 obtains a first target area, relating to a seatbelt, by segmenting a partial image of the occupant's chest into a multitude of areas, and a second target area by segmenting a partial image of the pelvic area into a multitude of areas. Each of the target areas can be formed from at least one pixel, and the respective sizes of the areas can be the same or different.

[0047] In the embodiment, the first target area can denote a set of pixels representing the seat belt on the occupant's chest section, i.e., on a section extending from the occupant's outboard shoulder to the occupant's inboard pelvis.

[0048] Furthermore, the second target area can denote a set of pixels representing the seatbelt on the occupant's pelvic section, i.e., a section extending from the inboard pelvis to the outboard pelvis of the occupant.

[0049] In this embodiment, the control unit 14 can be an electronic control unit. If a first area value of the first target area is greater than a preset first reference value and a second area value of the second target area is greater than a preset second reference value, the control unit 14 can determine that the seat belt is fitted normally. If an area of ​​the first target area is greater than a preset specific area and an area of ​​the second target area is greater than a preset specific area, the control unit 14 can determine that the seat belt is fitted normally.

[0050] In this case, the control unit 14 determines that the seat belt is abnormally fitted if at least one of the first area value of the first target area and of the second area value is less than or equal to the corresponding reference value.

[0051] This means that in a recorded image, the control unit 14 determines that the seat belt is fitted normally only if an area of ​​a section in which the shoulder belt overlaps the shoulder and chest section of the occupant and an area of ​​a section in which the lap belt overlaps the pelvic section of the occupant simultaneously meet corresponding reference values.

[0052] Additionally, the control unit 14 compares the first area value and the second area value with the first reference value and the second reference value, respectively, for each preset period. If the time during which the area values ​​are held greater than the corresponding reference values ​​exceeds a preset time, the control unit 14 determines that the seat belt is properly fastened. As described in Fig. As shown in Figure 5, the control unit 14 repeatedly determines whether the seat belt is fastened by using images of the occupant taken sequentially according to the preset period. In this case, the control unit 14 can compare the first and second area values ​​with the first and second reference values, respectively, using a multitude of consecutive images. Only if the first and second area values ​​of the multiple consecutive images are greater than the first and second reference values, respectively—that is, if the time during which the area value is greater than the reference value is greater than or equal to the preset time—does the control unit 14 determine that the seat belt is properly fastened.

[0053] In this case, i.e., if the time during which the area value is kept greater than the reference value is less than the preset time, the control unit 14 determines that the seat belt is abnormally fastened.

[0054] If the seat belt is incorrectly fastened, the notification unit 15 can issue a warning signal according to the control unit 14. The notification unit 15 can be configured as a voice output device and can emit a voice warning signal into the vehicle interior for a predetermined time according to the control unit 14.

[0055] Database 16 can include at least one storage medium consisting of a flash memory type storage medium, a hard disk type storage medium, a multimedia card micro type storage medium, a card type storage medium (e.g., a secure digital (SD) or extreme digital (XD) storage medium), a magnetic storage medium, a magnetic disk, an optical disk, a random access memory (RAM), a static RAM (SRAM), a read-only memory (ROM), an electrically erasable programmable ROM (EEPROM), and a programmable ROM (PROM).

[0056] Database 16 can store an image captured by the image sensor unit 11 and store images for a predetermined period in the past.

[0057] In addition, the database can store 16 data items, a program, and the like, which are necessary for the operation of the device for determining when to fasten a seatbelt.

[0058] Fig. Figure 6 is a flowchart showing a method for determining the fastening of a safety belt according to one embodiment.

[0059] According to Fig. 6. First, an image sensor unit captures a frontal image of a vehicle occupant. In this case, the frontal image of the occupant can be captured in such a way that it contains an image of the entire upper body of the vehicle occupant (S601).

[0060] Next, a section capture unit acquires a chest and a pelvic section of the occupant using the frontal image of the occupant acquired by the image sensor unit. In this case, the section capture unit can acquire the chest and pelvic sections of the occupant using a method for estimating human posture. The section capture unit extracts an image of the chest and an image of the occupant's pelvic section from the image and transmits the extracted images to an image segmentation unit (S602).

[0061] Next, the image segmentation unit obtains a first target area relating to a seatbelt by segmenting the partial image of the occupant's chest area into a plurality of areas. Simultaneously, the image segmentation unit obtains a second target area relating to a seatbelt by segmenting the partial image of the pelvic area into a plurality of areas. In the embodiment, the first target area can denote a set of pixels representing the seatbelt on the occupant's chest area, i.e., on a section extending from an outboard shoulder to an inboard pelvis of the occupant. Furthermore, the second target area can denote a set of pixels representing the seatbelt on the occupant's pelvic area.on a section extending from the inboard pelvis to the occupant's outboard pelvis (S603 and S604).

[0062] Next, a control unit compares a first area value of the first target area with a preset first reference value and simultaneously compares a second area value of the second target area with a preset second reference value (S605 and S606).

[0063] If the first area value of the first target area is greater than the preset first reference value and the second area value of the second target area is greater than the preset second reference value, the control unit determines that the seat belt is fitted normally (S607 and S608).

[0064] Alternatively, the control unit determines that the seat belt is abnormally fastened if at least one of the first area values ​​of the first target area and the second area value is less than or equal to the corresponding reference value. If the control unit determines that the seat belt is abnormally fastened, it issues a warning signal by controlling a notification unit (S609).

[0065] Fig. Figure 7 is a flowchart showing a method for determining the fastening of the safety belt according to another embodiment.

[0066] According to Fig. 7. First, an image sensor unit captures a frontal image of a vehicle occupant according to a preset period. In this case, the frontal image of the occupant can be captured in such a way that it contains an image of the entire upper body of the vehicle occupant (S701).

[0067] Next, a section capture unit acquires a chest and a pelvic section of the occupant using the frontal image of the occupant acquired by the image sensor unit. In this case, the section capture unit can acquire the chest and pelvic sections of the occupant using a method for estimating human posture. The section capture unit extracts an image of the chest and an image of the occupant's pelvic section from the image and transmits the extracted images to an image segmentation unit (S702).

[0068] Next, the image segmentation unit obtains a first target area relating to a seatbelt by segmenting the partial image of the occupant's chest into a plurality of areas. Simultaneously, the image segmentation unit obtains a second target area relating to the seatbelt by segmenting the partial image of the pelvic area into a plurality of areas. In the embodiment, the first target area can designate a set of pixels representing the seatbelt on the occupant's chest, i.e., on a section extending from an outboard shoulder to an inboard pelvis of the occupant. Furthermore, the second target area can designate a set of pixels representing the seatbelt on the occupant's pelvic area, i.e., on a section extending from an inboard pelvis to an outboard pelvis of the occupant (S703 and S704).

[0069] Next, a control unit compares the first area value and the second area value with a first reference value and a second reference value, respectively, for each preset period (S705 and S706), and if a time during which the area values ​​are held greater than the corresponding reference values ​​is greater than or equal to a preset time, the control unit determines that the seat belt is normally fitted (S707 and S708).

[0070] The control unit repeatedly determines whether the seat belt is fastened by using occupant images taken sequentially according to a preset period. In this case, the control unit compares the first and second area values ​​with the first and second reference values, respectively, using a large number of consecutive images, and only if the time during which the area values ​​remain greater than the corresponding reference values ​​exceeds a preset time does the control unit determine that the seat belt is properly fastened (S709).

[0071] Alternatively, the control unit determines that the seat belt is abnormally fastened if at least one of the first area values ​​of the first target area and the second area value is less than or equal to the corresponding reference value, or if the time during which the area values ​​are held greater than the corresponding reference values ​​is less than the preset time. If the control unit determines that the seat belt is abnormally fastened, it issues a warning signal by controlling a notification unit (S710).

[0072] Fig. Figure 8 is a flowchart showing a method for determining the fastening of the safety belt according to a further embodiment.

[0073] According to Fig. 8. First, an image sensor unit captures a frontal image of a vehicle occupant. The image sensor unit comprises a first image capture section for capturing an image of a shoulder belt section and a second image capture section for capturing an image of a lap belt section. A lens of the first image capture section can be adjusted to face the shoulder belt section, and a lens of the second image capture section can be adjusted to face the lap belt section. Accordingly, the first image capture section produces a first image, obtained by capturing an image of the shoulder belt section (S801), and the second image capture section produces a second image, obtained by capturing an image of the lap belt section (S802).

[0074] Next, a section acquisition unit captures a section of the occupant's chest using the first image. In this case, the section acquisition unit can capture the occupant's chest using a human posture estimation method. The section acquisition unit can extract an image of the occupant's chest from the image and transfer the extracted image to an image segmentation unit (S803).

[0075] Simultaneously, the section detection unit captures the occupant's pelvic section from the second image. In this case, the section detection unit can capture the occupant's pelvic section using the human posture estimation method. The section detection unit extracts an image of the occupant's pelvic section from the image and transfers the extracted image to the image segmentation unit (S804).

[0076] Next, the image segmentation unit is given an initial target area relating to a seatbelt by segmenting the partial image of the occupant's chest area into a multitude of areas (S805).

[0077] At the same time, the image segmentation unit receives a second target area relating to the seat belt by segmenting the partial image of the pelvic section into a multitude of areas (S806).

[0078] In this embodiment, the first target area can represent a set of pixels depicting the seat belt on the occupant's chest section, i.e., on a section extending from an outboard shoulder to an inboard pelvis of the occupant. Additionally, the second target area can represent a set of pixels depicting the seat belt on the occupant's pelvic section, i.e., on a section extending from an inboard pelvis to an outboard pelvis of the occupant.

[0079] Next, a control unit compares the first area value of the first target area with a preset first reference value and simultaneously compares the second area value of the second target area with a preset second reference value (S807 and S808).

[0080] If the first area value of the first target area is greater than the preset first reference value and the second area value of the second target area is greater than the preset second reference value, the control unit determines that the seat belt is fitted normally (S809 and S810).

[0081] Alternatively, the control unit determines that the seat belt is abnormally fastened if at least one of the first area values ​​of the first target area and the second area value is less than or equal to the corresponding reference value. If the control unit determines that the seat belt is abnormally fastened, it issues a warning signal by controlling a notification unit (S811).

[0082] Fig. Figure 9 is a flowchart showing a method for determining the fastening of the safety belt according to a further embodiment.

[0083] According to Fig.9. First, an image sensor unit captures a frontal image of a vehicle occupant. The image sensor unit comprises a first image capture section for capturing an image of a shoulder belt section and a second image capture section for capturing an image of a lap belt section. A lens of the first image capture section can be adjusted to face the shoulder belt section, and a lens of the second image capture section can be adjusted to face the lap belt section. Accordingly, the first image capture section produces a first image obtained by capturing an image of the shoulder belt section (S901), and the second image capture section produces a second image obtained by capturing an image of the lap belt section (S902).

[0084] Next, a section acquisition unit captures a section of the occupant's chest from the first image. In this case, the section acquisition unit can capture the occupant's chest using a human posture estimation method. The section acquisition unit extracts an image of the occupant's chest from the image and transfers the extracted image to an image segmentation unit (S903).

[0085] Simultaneously, the section detection unit captures a pelvic section of the occupant using the second image. In this case, the section detection unit can capture the occupant's pelvic section using the human posture estimation method. The section detection unit extracts an image of the occupant's pelvic section from the image and transfers the extracted image to the image segmentation unit (S904).

[0086] Next, the image segmentation unit is given an initial target area relating to a seatbelt by segmenting the partial image of the occupant's chest area into a multitude of areas (S905).

[0087] At the same time, the image segmentation unit receives a second target area relating to the seat belt by segmenting the partial image of the pelvic section into a multitude of areas (S906).

[0088] In this embodiment, the first target area can represent a set of pixels depicting the seat belt on the occupant's chest section, i.e., on a section extending from an outboard shoulder to an inboard pelvis of the occupant. Additionally, the second target area can represent a set of pixels depicting the seat belt on the occupant's pelvic section, i.e., on a section extending from an inboard pelvis to an outboard pelvis of the occupant.

[0089] Next, a control unit compares a first area value and a second area value with a first reference value and a second reference value, respectively, for each preset period (S907 and S908). If a time during which the area values ​​are held higher than the corresponding reference values ​​is greater than or equal to a preset time, the control unit determines that the seat belt is properly fastened (S909 and S910). The control unit repeatedly determines whether the seat belt is fastened by using occupant images taken sequentially according to the preset period. In this case, the control unit compares the first area value and the second area value with the first reference value and the second reference value, respectively.the second reference value using a large number of successive images, and only if a time during which the area values ​​are kept greater than the corresponding reference values ​​is greater than or equal to the preset time, does the control unit determine that the seat belt is fitted normally (S911).

[0090] Alternatively, the control unit determines that the seat belt is abnormally fastened if at least one of the first area values ​​of the first target area and the second area value is less than or equal to the corresponding reference value, or if the time during which the area values ​​are held greater than the corresponding reference values ​​is less than the preset time. If the control unit determines that the seat belt is abnormally fastened, it issues a warning signal by controlling a notification unit (S912).

[0091] According to the embodiments, a device and a method for determining whether a seat belt is fastened according to the present invention can significantly improve the accuracy and reliability of determining whether a seat belt is fastened.

[0092] Furthermore, misunderstandings about fastening the seatbelt due to a faulty buckle sensor or improper fastening of the belt can be prevented.

[0093] In this embodiment, terms such as "unit" refer to software or a hardware component such as an FPGA or an ASIC, and the term "unit" fulfills certain functions. However, the term "unit" is not limited to software or hardware. The term "unit" can be configured to reside on an addressable storage medium or to represent one or more processors. Thus, in one example, the term "unit" includes components such as software components, object-oriented software components, class components, task components, processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables.The functions provided by these components and the term "unit" can be grouped into a smaller number of components and "units" or subdivided into additional components and "units." Furthermore, the components and "units" can also be implemented to replicate one or more central processing units (CPUs) within a device or security multimedia card.

Claims

[1] Device for determining the fastening of a safety belt, comprising: an image sensor unit (11) adapted to capture a front image of a vehicle occupant; a portion detection unit (12) adapted to detect a chest portion and a pelvic portion of the occupant using the front image; an image segmentation unit (13) adapted to obtain a first target region related to a seat belt by segmenting a partial image of the chest portion of the occupant into a plurality of regions and to obtain a second target region related to the seat belt by segmenting a partial image of the pelvic portion into a plurality of regions; and a control unit (14) adapted to determine that the seat belt is normally fastened when a first area value of the first target area is greater than a preset first reference value and a second area value of the second target area is greater than a preset second reference value, wherein the first area value of the first target area is an area of ​​a portion in which a shoulder belt overlaps a shoulder and chest area of ​​the occupant, and the second area value of the second target area is an area of ​​a section in which a lap belt overlaps the occupant's lap area, wherein the control unit is adapted to determine that the safety belt is normally worn when the area of ​​the portion in which the shoulder belt contacts the shoulder and overlaps the chest area of ​​the occupant is larger than the preset first reference value and at the same time the area of ​​the section in which the lap belt overlaps the occupant's pelvic area is larger than the preset second reference value, and wherein the control unit is adapted to determine that the safety belt is abnormally fastened if at least one of the areas of the portion in which the shoulder belt overlaps the shoulder and chest areas of the occupant is smaller or is equal to the first reference value, or the area of ​​the section in which the lap belt overlaps the occupant's lap section is less than or equal to the second reference value. [2] A seat belt fastening determining device according to claim 1, wherein the portion detecting unit (12) is adapted to detect the chest portion and the pelvic portion of the occupant using a human posture estimation method. [3] The seat belt fastening determining device according to claim 1, wherein the image sensor unit (11) is adapted to capture the front image of the vehicle occupant and transmit the captured front image to the section detecting unit according to a preset period (S701). [4] The seat belt fastening determining device according to claim 3, wherein the control unit is adapted to compare the first area value with the first reference value and the second area value with the second reference value for each preset period (S705 and S706) and to determine that the seat belt is normally fastened when a time during which the first area value is maintained greater than the first reference value and the second area value is maintained greater than the second reference value is greater than or equal to a preset time (S707 and S708). [5] A seat belt fastening determining device according to claim 1, further comprising a notification unit (15) adapted to output a warning signal according to the control of the control unit when the seat belt is abnormally fastened. [6] A seat belt fastening determining device according to claim 1, wherein the seat belt comprises a 3-point seat belt. [7] A seat belt fastening detecting device according to claim 1, wherein the image sensor unit (11) comprises: a first image pickup part (111) configured to pick up an image of a shoulder strap portion; and a second image pickup part (112) configured to pick up an image of a lap belt portion. [8] The seat belt fastening determining device according to claim 7, wherein the portion detecting unit (12) is adapted to detect the chest portion of the occupant by cutting out an image of an outboard shoulder, an image of an inboard pelvis, and an image of a central portion connecting the outboard shoulder and the inboard pelvis from the image picked up by the first image picking up part (111) using a human posture estimation method. [9] The seat belt fastening determining device according to claim 7, wherein the portion detecting unit (12) is adapted to detect the pelvic portion of the occupant by cutting out an image of an inboard pelvis, an image of an outboard pelvis, and an image of a central portion connecting the inboard pelvis and the outboard pelvis from the image picked up by the second image picking up part (112) using a human posture estimation method. [10] A method for determining the fastening of a safety belt, comprising: Capturing, by an image sensor unit (11), a front image of a vehicle occupant (S601)); detecting, by a portion detecting unit (12), a chest portion and a pelvic portion of the occupant using the front image (S803, S804); Obtaining, by an image segmentation unit (13), a first target region relating to a seat belt by segmenting a partial image of the chest portion of the occupant into a plurality of regions (S603) and a second target region relating to the seat belt by segmenting a partial image of the pelvic portion into a plurality of regions (S604); and Determining, by a control unit, that the seat belt is normally fastened when a first area value of the first target area is larger than a preset first reference value and a second area value of the second target area is larger than a preset second reference value (S607, S608). wherein the first area value of the first target area is an area of ​​a portion in which a shoulder belt overlaps a shoulder and chest area of ​​the occupant, and the second area value of the second target area is an area of ​​a section in which a lap belt overlaps the occupant's lap area, wherein the control unit determines that the seat belt is normally fastened when the area of ​​the section in which the shoulder belt overlaps the shoulder and chest areas of the occupant is larger than the preset first reference value and, at the same time, the area of ​​the section in which the lap belt overlaps the lap area of ​​the occupant is larger than the preset second reference value, and wherein, if at least one of the areas of the portion in which the shoulder belt overlaps the shoulder and chest areas of the occupant is less than or equal to the first reference value, or the area of ​​the portion in which the lap belt overlaps the lap portion of the occupant is less than or equal to the second reference value, the control unit determines that the seat belt is abnormally fastened. [11] The method of claim 10, wherein determining that the seat belt is normally worn further comprises: Comparing, by the control unit (14), the first area value and the second area value with the first reference value and the second reference value, respectively, using the image taken for each preset period; and Determining that the seat belt is normally fastened when a time during which the first area value is maintained greater than the first reference value and the second area value is maintained greater than the second reference value is greater than or equal to a preset time according to a comparison result (S709). [12] The method of claim 10, further comprising: Determining, by the control unit (14), that the seat belt is abnormally fastened, if at least one of the first area value of the first target area is less than or equal to the first reference value and the second area value is less than or equal to the second reference value. [13] The method of claim 12, further comprising: Outputting, by a notification unit (15), a warning signal according to the control of the control unit (14) when the seat belt is abnormally fastened (S710). [14] The method of claim 10, wherein the seat belt has a 3-point seat belt; and the image sensor unit (11) has a first image pickup part (111) configured to capture an image of a shoulder belt portion (S801) and a second image pickup part (112) configured to capture an image of a lap belt portion (S802). [15] The method of claim 14, wherein detecting the chest portion and the pelvic portion of the occupant comprises: Detecting the chest portion by cutting out an image of an outboard shoulder, an image of an inboard pelvis, and an image of a central portion connecting the outboard shoulder and the inboard pelvis from the image captured by the first image capturing part (111) using a human posture estimation method (S803). [16] The method of claim 14, wherein detecting the chest portion and the pelvic portion of the occupant comprises: Detecting the pelvis portion by cutting out an image of an inboard pelvis, an image of an outboard pelvis, and an image of a central region connecting the inboard pelvis and the outboard pelvis from the image captured by the second image capturing part (112) using a human posture estimation method (S804).

Images