Program, helmet wearing determination system, helmet wearing determination method

Abstract

To provide a technique that can precisely determine whether or not an occupant of a two-wheel vehicle is wearing a helmet.SOLUTION: The present invention relates to a method for determining whether or not an occupant of a two-wheel vehicle is wearing a helmet, the method including: specifying a position of the two-wheel vehicle from a photographed image of a road photographed by an imaging device; estimating an occupant head part region corresponding to the head part of the occupant from an image of a region of an upper position of the specified two-wheel vehicle: comparing image features extracted from the occupant head part region and image features corresponding to the head part at least one of when the occupant is wearing a helmet and when the occupant is not wearing a helmet.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a program, a helmet wearing determination system, and a helmet wearing determination method.

Background Art

[0002] In recent years, technologies for monitoring moving objects such as automobiles traveling on roads have been proposed from the perspective of security. And such technologies are for discriminating and detecting whether a moving object traveling on a road is an automobile (four-wheeled vehicle), a motorcycle (two-wheeled vehicle), a bicycle (two-wheeled vehicle), or a pedestrian.

[0003] In particular, from the viewpoints of safety and the like, it is desired to detect whether a person is wearing a helmet.

[0004] Therefore, technologies for detecting whether a person is wearing a helmet have been proposed (for example, Patent Document 1).

[0005] The invention of this Patent Document 1 processes a captured image input from a camera in an image processing unit to detect a person being imaged. When the image processing unit detects a person, it determines whether the person is wearing a helmet. The determination method first detects a person's face and head, and determines whether the person is wearing a helmet based on the ratio of the width of the face to the width of the head.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0007] However, the invention of Patent Document 1 requires image processing for detecting a person's face and head. However, when wearing a helmet of a certain type, such as a full-face type or a jet type helmet, the face is hidden inside the helmet, making it difficult to detect the face.

[0008] Also, if an attempt is made to determine whether a helmet is being worn based on the ratio of the width of the face to the width of the head, in both the case of not wearing a helmet and the case of wearing a full-face type helmet or the like, the ratio of the width of the person's face to the width of the head both becomes close to 1, making it difficult to determine whether a helmet is being worn.

[0009] Therefore, the present invention has been invented in view of the above problems, and an object thereof is to provide a helmet wearing determination method, a helmet wearing determination device, and a program that can accurately detect whether a person is wearing a helmet.

Means for Solving the Problems

[0010] A first aspect of the present invention is a program for causing a computer to execute a helmet wearing determination process for determining whether a helmet is being worn or not being worn from an image captured by a photographing device, using a learning model obtained by learning the image features of a first image including a head wearing a helmet and the image features of a second image including a head not wearing a helmet. A second aspect of the present invention is a program for causing a computer to execute a process of causing a learning model, which is used to determine whether a helmet is being worn or not being worn from an image captured by a photographing device, to learn the image features of a first image including a head wearing a helmet and the image features of a second image including a head not wearing a helmet.

[0011] A third aspect of the present invention is a helmet wearing determination system including helmet wearing determination means for determining whether a helmet is worn or not worn from an image captured by an imaging device using a learning model that has learned image features of a first image including a head wearing a helmet and image features of a second image including a head not wearing a helmet. A fourth aspect of the present invention is a helmet wearing determination system including means for causing a learning model used to determine whether a helmet is worn or not worn from an image captured by an imaging device to learn image features of a first image including a head wearing a helmet and image features of a second image including a head not wearing a helmet.

[0012] A fifth aspect of the present invention is a helmet wearing determination method including a helmet wearing determination process for determining whether a helmet is worn or not worn from an image captured by an imaging device using a learning model that has learned image features of a first image including a head wearing a helmet and image features of a second image including a head not wearing a helmet. A sixth aspect of the present invention is a helmet wearing determination method including a process for causing a learning model used to determine whether a helmet is worn or not worn from an image captured by an imaging means to learn image features of a first image including a head wearing a helmet and image features of a second image including a head not wearing a helmet.

Advantages of the Invention

[0013] According to the present invention, it is possible to accurately detect whether a person is wearing a helmet from a captured image.

Brief Description of the Drawings

[0014]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Embodiment for Carrying Out the Invention

[0015] An embodiment of the present invention will be described.

[0016] First, the concept of the present invention will be described. FIG. 1 is a block diagram for explaining the concept of the present invention. In FIG. 1, 1 is a photographing device, and 2 is a helmet wearing determination device.

[0017] The photographing device 1 is installed at a predetermined position beside the road on which the two-wheeled vehicle travels, and photographs the two-wheeled vehicle traveling on the road. Here, the two-wheeled vehicle is, for example, a motorcycle such as a bike, a bicycle, or the like.

[0018] The helmet wearing determination device 2 processes the photographed image of the photographing device 1, estimates the rider's head region corresponding to the head of the person riding on the two-wheeled vehicle traveling on the road, and compares the image features of the rider's head region with the image features corresponding to the head when wearing a helmet and / or when not wearing a helmet, and determines whether the rider is wearing a helmet or not.

[0019] In particular, the present invention utilizes the fact that the image features of the head (the overall shape including the head and face) when wearing a helmet and the image features of the head (the overall shape including the head and face) when not wearing a helmet each have specific image features, and is characterized by determining whether the rider of the two-wheeled vehicle is wearing a helmet or not.

[0020] By using such features, the present invention can determine whether a rider of a two-wheeler is wearing a helmet or not, without the need to recognize the face of the rider of the two-wheeler as in the prior art.

[0021] Next, the operation of the helmet wearing determination device 2 will be described with reference to FIG. 2.

[0022] First, as shown in FIG. 2, an image of the area at the upper position of the two-wheeler is extracted from the captured image captured by the imaging device 1. The extraction method is to extract the entire image of the two-wheeler with a driver or the like on board, and then take out the image of the area within the previously determined range of the upper position from the entire image. Alternatively, the vehicle body part of the two-wheeler may be specified from the entire captured image of the two-wheeler with a driver or the like on board, and an image of the area within a predetermined range above the vehicle body part may be extracted.

[0023] Next, a rider's head region corresponding to the head of the person riding on the two-wheeler is estimated from the extracted image of the upper region. The human head is a sphere, and even if wearing a helmet, the contour of the head is somewhat arc-shaped and does not have a shape with a pointed tip like a bump. If the shapes of these human heads and the estimated heads are learned in advance and stored in a database, the rider's head region corresponding to the rider's head can be specified from the extracted region.

[0024] Also, it is possible to specify the shape corresponding to the human head from the shape of the valley part of the contour of the image at the upper position. For example, when two people are riding on a two-wheeler, a unique valley formed by the head of the person riding in front of the two-wheeler, the arm of the person riding behind the two-wheeler from the back, and the line from the chest to the head will be formed between the heads. Therefore, if the shape of this unique valley part is learned in advance and stored in a database, the rider's head region including the head can be specified from the shape of the contour at the upper position. Furthermore, by combining the above two methods, the accuracy of specifying the rider's head region including the head can be further improved.

[0025] In this way, the rider's head area corresponding to the head of the person riding on the two-wheeler is estimated.

[0026] Next, the image features of the rider's head area are compared with the image features corresponding to the head when wearing a helmet or the image features corresponding to the head when not wearing a helmet. The image of the head when wearing a helmet has unique image features. For example, in the case of helmets such as full-face type, jet type, and cap type, in the image from the direction in which the face is reflected, there are unique features when wearing those helmets. On the other hand, the image of the head when not wearing a helmet also has unique image features. Therefore, such unique image features of the image of the head when wearing a helmet and the image features of the image of the head when not wearing a helmet are pre-learned and stored in a database, and this database is used to identify the image features of the rider's head area and determine whether the rider is wearing a helmet.

[0027] Next, specific embodiments of the present invention will be described.

[0028] FIG. 3 is a block diagram of a helmet wearing determination system according to an embodiment of the present invention.

[0029] The helmet wearing determination system according to an embodiment of the present invention includes a photographing device 1, a helmet wearing determination device 2, and a notification device 3.

[0030] The photographing device 1 photographs a two-wheeler traveling on a road and is installed at a position where an image that can easily detect the above-described rider's head area can be photographed.

[0031] Specifically, as shown in FIG. 4, it is installed at a position where the two-wheeler can take a picture within the range from directly in front to directly sideways (mainly diagonally forward) with respect to the traveling direction of the two-wheeler. There are several types of helmets, such as full-face type, jet type, and cap type. Therefore, when taking a picture from the direction in which the face of the rider of the two-wheeler is reflected, the difference between the case of wearing a helmet and the case of not wearing a helmet is likely to be prominent. Furthermore, the photographing device 1 is preferably installed in front of a point where the two-wheeler decelerates, such as an intersection, a stop line, a curve, a speed bump, etc. By taking a picture in front of a point where the speed of the two-wheeler slows down, it becomes easier to obtain a photographed image with less motion blur. As a result, it becomes easier to extract features from the photographed image, and the accuracy of helmet wearing determination can be improved.

[0032] The helmet wearing determination device 2 includes a moving body detection unit 21, a category classification unit 22, a vehicle body position detection unit 23, a rider's head region estimation unit 24, a helmet wearing determination unit 25, a category determination dictionary 26, a vehicle body detection dictionary 27, a rider's head region estimation dictionary 28, and a helmet wearing determination dictionary 29.

[0033] The moving body detection unit 21 detects a moving body in the image from the photographing device 1. Various methods for detecting a moving body have been proposed conventionally, and an appropriate method may be selected.

[0034] The category classification unit 22 uses the category determination dictionary 26 to identify a two-wheeler (motorcycle) from the moving body and outputs the entire image of the moving two-wheeler (motorcycle) to the vehicle body position detection unit 23. The category determination dictionary 26 has information for identifying a two-wheeler (motorcycle) stored in a database.

[0035] Specifically, the category classification unit 22 receives the detection of the moving object from the moving object detection unit 21 and classifies the type (category) of the moving object. For the classification of the type (category) of the moving object, the size of the search range for identifying the type (category) of the moving object is determined in advance, the probability that the moving object is other than a two-wheeled vehicle is calculated based on the feature amounts within the search range, and based on this result, the moving object other than a two-wheeled vehicle is excluded. For example, within a predetermined search range, a feature such that two circles are arranged substantially linearly can be used as a feature for calculating the probability that the moving object is other than a two-wheeled vehicle. Also, the features of the position and number of the headlight can be used as features for calculating the probability that it is not a two-wheeled vehicle (motorcycle). Such characteristic parts of the image are learned in advance, and the data is stored in a database as a category determination dictionary 25. Then, the category classification unit 22 uses the category determination dictionary 25 to determine the probability that the detected moving object is not a two-wheeled vehicle. When this probability exceeds a predetermined threshold value, the entire image of the moving object is not output to the vehicle body position detection unit 23, and in other cases, the entire image of the detected moving object is output to the vehicle body position detection unit 23. For example, the category classification unit 22 does not output to the vehicle body position detection unit 23 a moving object that is determined to be not a two-wheeled vehicle with a probability of 90% (=predetermined threshold value) using the category determination dictionary 25. On the other hand, a moving object that is determined to be not a two-wheeled vehicle with a probability of 90% or less, for example, 85%, may be a two-wheeled vehicle, so it is output to the vehicle body position detection unit 23 for detailed verification.

[0036] The vehicle body position detection unit 23 specifies a range corresponding to the vehicle body part of the motorcycle using the vehicle body detection dictionary 26 from the image of the moving object output from the category classification unit 22. For example, in the example of the overall image of the motorcycle (bike) shown in FIG. 5 (an image taken from directly side-on), the range is specified as the vehicle body part range with the length connecting the front tip of the front wheel and the rear tip of the rear wheel of the motorcycle as the horizontal width and the vertical length of the wheel as the vertical width. Also, in the example of the overall image of the motorcycle (bike) shown in FIG. 6 (an image taken from obliquely in front), the range is specified as the vehicle body part range with the length connecting the front tip of the front wheel and the rear tip of the rear wheel of the motorcycle as the horizontal width and a range slightly larger than the width between the front tip and the upper tip on the installation side of the front and rear wheels of the motorcycle. Note that the wheel shape of the motorcycle, the distance between the wheels, etc. required for the specification are learned in advance and the data is stored in a database as the vehicle body detection dictionary 27.

[0037] Note that since not necessarily only the image of the motorcycle is output from the category classification unit 22, there may be a case where the vehicle body position detection unit 23 cannot specify the vehicle body part of the motorcycle using the vehicle body detection dictionary 26. In this case, the moving object is excluded from the processing target assuming it is not a motorcycle.

[0038] The passenger head region estimation unit 24 detects a passenger head region estimated to be the head of a person riding in a region above the vehicle body specified by the vehicle body position detection unit 23. As a detection method, a contour shape corresponding to a human head is detected from the contour shape of an image in the upper region of the vehicle body specified by the vehicle body position detection unit 23, and a predetermined size region surrounding the contour shape is detected as the passenger head region. Since the human head is a sphere, even if a helmet is worn, its contour shape has the characteristic of being on an arc to some extent. The cob shape having such characteristics of the human head is learned in advance and registered in the passenger head region estimation dictionary 28. Further, when two or more people are riding on a two-wheeled vehicle, etc., a unique valley formed by the head of the person riding in front of the two-wheeled vehicle, the arm of the person riding behind the two-wheeled vehicle from the back, and the line from the chest to the head is formed between the heads. Therefore, if the shape of this unique valley is learned in advance and registered in the passenger head region estimation dictionary 28, the passenger head region including the head can be specified from the shape of the contour of the upper position. Furthermore, by combining the above two methods, the accuracy of specifying the passenger head region including the head can be further improved.

[0039] The helmet wearing determination unit 25 compares the image features of the passenger head region with the image features corresponding to the head when wearing a helmet or the image features corresponding to the head when not wearing a helmet, and determines whether the passenger is wearing a helmet.

[0040] For example, as shown in FIG. 7, the image of the head of a rider wearing a helmet such as a full-face type, a jet type, or a cap type has unique features. On the other hand, as shown in FIG. 7, the image of the head when not wearing a helmet also has unique features. Such unique features include, for example, the feature that when wearing a helmet, a straight line occurs across the face slightly above the center of the head. Also, as another feature, when wearing a helmet, the head shape is more clearly rounded than when not wearing a helmet. Furthermore, as another feature, when wearing a helmet, the head shape is fixed over time, while when not wearing a helmet, the hair, etc. flutters and is not fixed. The image features of these features are pre-learned using HOG features, luminance gradient features (directional features), CCS features, Haar like features, and as statistical discrimination methods, Support Vector Machine, Generalized Learning Vector Quantization, AdaBoost, Real AdaBoost, pseudo-Bayes discrimination, etc., and registered in the helmet wearing determination dictionary 29.

[0041] Then, the helmet wearing determination unit 25 determines whether the rider on the captured image is wearing a helmet by determining which of the image features of the helmet-wearing time or the image features of the non-helmet-wearing time the image features of the rider's head region resemble using the helmet wearing determination dictionary 29.

[0042] For example, when using the helmet wearing determination dictionary 29 in which the above-described features are stored, if it can be detected that a straight line occurs across slightly above the center of the head from the image features of the head in the rider's head region, it is determined that the rider is wearing a helmet. On the other hand, when such a straight line cannot be detected, it is determined that the rider is not wearing a helmet.

[0043] Also, when it is possible to clearly detect that the head shape has a distinct roundness based on the image features of the head in the passenger's head region using the helmet-wearing determination dictionary 29 in which the above-described features are stored, it is determined that the helmet is being worn. On the other hand, when the distinct roundness cannot be detected, it is determined that the helmet is not being worn.

[0044] Furthermore, when it is possible to detect that the temporal change in the image features of the head in the passenger's head region is fixed using the helmet-wearing determination dictionary 29 in which the above-described features are stored, it is determined that the helmet is being worn. On the other hand, when it is not fixed, it is determined that the helmet is not being worn.

[0045] The notification device 3 has a display unit 30 that displays the captured image of the imaging device 1, and is a device that notifies the monitor of the determination result of the helmet-wearing determination unit 25.

[0046] The notification device 3 receives the determination result of the helmet-wearing determination unit 25, the coordinates of the region surrounding the passenger and the two-wheeled vehicle body from the vehicle body position detection unit 23, and the coordinates of the passenger's head region from the passenger head region estimation unit 24. Then, as shown in FIG. 8, the captured image of the imaging device 1 is displayed on the display unit 30, and when the determination result of the helmet-wearing determination unit 25 is that the passenger is not wearing a helmet, a figure surrounding the region including at least the two-wheeled vehicle body and the passenger's head is displayed on the captured image, and a figure surrounding the passenger's head region is displayed.

[0047] Furthermore, the features of the passenger (for example, the driver) may be displayed clearly (for example, largely). Examples of the features of the passenger (for example, the driver) include the face, clothing, motorcycle (color, vehicle type, modification, etc.). Using face recognition and object recognition technologies, these features are recognized from the image, the recognized part is cut out from the image, and the cut-out image is displayed largely separately from the image as shown in FIG. 8 (for example, the driver's face is displayed largely). By displaying in this way, the features of the passenger (for example, the driver) can be grasped.

[0048] Furthermore, in combination with a person verification system or an object verification system, the recognized features described above (e.g., face, clothing, color of the motorcycle, motorcycle model, modifications, etc.) can be verified against the information of the person verification system or the object verification system, and the verification result can be displayed. Examples of the verification result include criminal record information, clothing brand, motorcycle model and modification details, etc. By configuring in this way, it can also be used for crime prevention and detection.

[0049] According to the present embodiment configured as described above, it is not necessary to recognize the face of the motorcycle rider, and it is possible to determine whether the motorcycle rider is wearing a helmet or not.

[0050] Furthermore, when the motorcycle rider is not wearing a helmet, the motorcycle and the helmetless head are displayed in an easily distinguishable manner, so that the attention of the monitor can be attracted.

[0051] Also, in the helmet wearing determination device 2, the helmet wearing determination unit 25 may include an infrared detection unit (not shown). The infrared detection unit may be, for example, an infrared sensor. The helmet wearing determination unit 25 may determine whether a helmet is being worn based on, for example, the detection result of infrared rays by the infrared detection unit. Alternatively, the helmet wearing determination unit 25 may weight the result of the comparison of the above-described image features with the detection result of infrared rays, etc., to determine whether a helmet is being worn. By including the infrared detection unit, the helmet wearing determination unit 25 can determine whether a helmet is being worn with higher accuracy.

[0052] Note that, in the above-described embodiment, the helmet wearing determination device 2 is configured by hardware, but it is also possible to be realized by a computer program. In this case, a processor operating according to a program stored in a program memory realizes the same functions and operations as those of the above-described embodiment. Also, it is possible to realize only some of the functions of the above-described embodiment by a computer program.

[0053] In the above-described embodiments, the object to be determined was described with reference to a two-wheeled vehicle. However, if the head of a person can be detected, the present invention is not limited to two-wheeled vehicles, and can naturally be applied to unicycles, three-wheeled vehicles, and other objects as well.

[0054] Also, some or all of the above-described embodiments can be described as in the following supplementary notes, but are not limited thereto.

[0055] (Supplementary Note 1) An imaging means installed at a predetermined position for imaging a two-wheeled vehicle traveling on a road, processing the captured image captured by the imaging means, estimating a passenger head region corresponding to the head of a person riding on the two-wheeled vehicle traveling on the road, comparing the image features of the passenger head region with the image features corresponding to the head when wearing a helmet or / and when not wearing a helmet, and a helmet wearing determination means for determining whether the passenger is wearing a helmet or not A helmet wearing determination system having the same.

[0056] (Supplementary Note 2) The helmet wearing determination means compares the image features corresponding to the head when wearing a helmet or / and when not wearing a helmet, which are pre-learned and stored, with the image features of the passenger head region, and determines whether the passenger is wearing a helmet or not The helmet wearing determination system according to Supplementary Note 1.

[0057] (Supplementary Note 3) The passenger head region estimation means specifies the position of the two-wheeled vehicle from the captured image, and estimates the passenger head region from the image region at the upper position of the specified two-wheeled vehicle. The helmet wearing determination system according to Supplementary Note 1 or Supplementary Note 2.

[0058] (Supplementary Note 4) A notification means for notifying the determination result of the helmet wearing determination means having The helmet wearing determination system according to any one of Supplementary Notes 1 to 3.

[0059] (Appendix 5) The notification means notifies when the determination result of the helmet wearing determination means indicates that the helmet is not being worn. The helmet wearing determination system according to Appendix 4.

[0060] (Appendix 6) The notification means has a display means for displaying the captured image, and when the determination result of the helmet wearing determination means indicates that the helmet is not being worn, a figure surrounding at least the area including the two-wheeler body and the rider's head identified by the rider's head region estimation means is displayed on the captured image. The helmet wearing determination system according to Appendix 5.

[0061] (Appendix 7) The photographing means is installed at a position where the two-wheeler can be photographed in a range from obliquely forward to directly sideward with respect to the traveling direction of the two-wheeler. The helmet wearing determination system according to any one of Appendices 1 to 6.

[0062] (Appendix 8) The photographing means is installed in front of the point where the two-wheeler decelerates. The helmet wearing determination system according to any one of Appendices 1 to 7.

[0063] (Appendix 9) Process the captured image of a photographing device that is photographing a road, estimate the rider's head region corresponding to the head of a person riding on the two-wheeler traveling on the road, compare the image features extracted from the rider's head region with the image features corresponding to the head when wearing a helmet and / or when not wearing a helmet, and determine whether the rider is wearing a helmet or not. Helmet wearing determination method.

[0064] (Appendix 10) Learn in advance the image features corresponding to the head when wearing a helmet and / or when not wearing a helmet. The helmet wearing determination method according to Appendix 9.

[0065] (Appendix 11) Identify the position of the two-wheeler from the captured image, Estimate the rider's head region from the image region at the upper position of the specified two-wheeler. The helmet wearing determination method according to Appendix 9 or Appendix 10.

[0066] (Appendix 12) Notify the result of the determination. The helmet wearing determination method according to any one of Appendices 9 to 11.

[0067] (Appendix 13) Notify when the result of the determination is non-wearing of a helmet. The helmet wearing determination method according to Appendix 12.

[0068] (Appendix 14) When the result of the determination is non-wearing of a helmet, display and notify a figure surrounding a region including at least the two-wheeler body and the rider's head on the captured image. The helmet wearing determination method according to Appendix 13.

[0069] (Appendix 15) Photograph the two-wheeler in a range from diagonally in front to directly across with respect to the traveling direction of the two-wheeler. The helmet wearing determination method according to any one of Appendices 9 to 14.

[0070] (Appendix 16) Photograph the two-wheeler from before the point where the two-wheeler decelerates. The helmet wearing determination method according to Appendix 15.

[0071] (Appendix 17) Process the captured image of the imaging device photographing the road, estimate the rider's head region corresponding to the head of the person riding on the two-wheeler traveling on the road, Compare the image features of the rider's head region with the image features corresponding to the head when wearing a helmet and / or when not wearing a helmet, and determine whether the rider is wearing a helmet or not, a helmet wearing determination means A helmet wearing determination device having the same.

[0072] (Appended Note 18) A process of processing a captured image of a photographing device that is photographing a road and estimating a passenger head region corresponding to the head of a person riding on a two-wheeled vehicle traveling on the road, a process of comparing the image features of the passenger head region with the image features corresponding to the head when wearing a helmet and / or when not wearing a helmet, and determining whether the passenger is wearing a helmet or not A program for causing a computer to execute.

[0073] Although the present invention has been described by giving the above preferred embodiments, the present invention is not necessarily limited to the above embodiments, and various modifications can be made and implemented within the scope of its technical idea.

[0074] This application claims the priority based on Japanese Patent Application No. 2013-239598 filed on November 20, 2013, and incorporates all of its disclosures herein.

Explanation of Reference Numerals

[0075] 1 Photographing device 2 Helmet wearing determination device 3 Notification device 21 Moving body detection unit 22 Category classification unit 23 Vehicle body position detection unit 24 Passenger head region estimation unit 25 Helmet wearing determination unit 26 Category determination dictionary 27 Vehicle body detection dictionary 28 Passenger head region estimation dictionary 29 Helmet wearing determination dictionary 30 Display unit

Claims

1. A helmet wearing determination process for determining whether a helmet is worn or not on a head shown in an image based on the similarity between the image features of an image captured by an imaging device and the image features registered in a dictionary in which the image features of a first image including a head wearing a helmet and the image features of a second image including a head not wearing a helmet are registered. A program for causing a computer to execute the process.

2. The image features of the image captured by the imaging device are the image features extracted from the image. The program according to Claim 1.

3. An extraction process for extracting the image features of the image from the image captured by the imaging device. The program according to Claim 1 or 2, further causing a computer to execute the extraction process.

4. The head is the head of a rider of a two-wheeled vehicle. The program according to any one of Claims 1 to 3.

5. The helmet wearing determination process determines whether a helmet is worn or not on a head shown in the image based on the similarity between the image features of the first image and the second image and the image features of the image captured by the imaging device. The program according to any one of Claims 1 to 4.

6. A helmet wearing determination means for determining whether a helmet is worn or not on a head shown in an image based on the similarity between the image features of an image captured by an imaging device and the image features registered in a dictionary in which the image features of a first image including a head wearing a helmet and the image features of a second image including a head not wearing a helmet are registered. A helmet wearing determination system including the helmet wearing determination means.

7. The image features of the image captured by the imaging device are the image features extracted from the image. The helmet wearing determination system according to Claim 6.

8. An extraction means for extracting the image features of the image from the image captured by the imaging device. The helmet wearing determination system according to Claim 6 or 7, further including the extraction means.

9. The head is the head of a rider of a two-wheeled vehicle. The helmet wearing determination system according to any one of Claims 6 to 8.

10. The helmet wearing determination means determines whether a helmet is worn or not on a head shown in the image based on the similarity between the image features of the first image and the second image and the image features of the image captured by the imaging device. The helmet wearing determination system according to any one of claims 6 to 9.

11. A helmet wearing determination process for determining whether a helmet is worn or not worn on the head shown in the image based on the similarity between the image features of the image captured by the imaging device and the image features registered in the dictionary, using the dictionary in which the image features of the first image including the head wearing a helmet and the image features of the second image including the head not wearing a helmet are registered. A helmet wearing determination method including the above.

12. The image features of the image captured by the imaging device are the image features extracted from the image. The helmet wearing determination method according to claim 11.

13. An extraction process for extracting the image features of the image from the image captured by the imaging device. The helmet wearing determination method according to claim 11 or 12, further including the above.

14. The head is the head of a two-wheeler rider. The helmet wearing determination method according to any one of claims 11 to 13.

15. The helmet wearing determination process determines whether a helmet is worn or not worn on the head shown in the image based on the similarity between the image features of the first image and the second image and the image features of the image captured by the imaging device. The helmet wearing determination method according to any one of claims 11 to 14.

Images