Title of the invention: A non-temporary computer-readable medium containing a person state detection device, a person state detection method, and a program.
Technical field
[0001]
The present invention relates to a non-temporary computer-readable medium in which a person state detection device, a person state detection method, and a program are stored.
Background technology
[0002]
In recent years, in surveillance systems and the like, a technique for detecting a state such as a person's posture or behavior from an image of a surveillance camera has been used. For example, Patent Documents 1 and 2 are known as techniques related to detecting the posture of a person. Patent Document 1 discloses a technique for recognizing a person's posture from changes over time in an image area of ​​the person. Patent Document 2 discloses a technique for determining a person's posture from the height, width, and depth of a three-dimensional person area in a distance image. In addition, Non-Patent Document 1 is known as a technique related to estimating the skeleton of a person.
Prior art literature
Patent documents
[0003]
Patent Document 1: Japanese Unexamined Patent Publication No. 2010-237873
Patent Document 2: Japanese Unexamined Patent Publication No. 2008-146583
Non-patent literature
[0004]
Non-Patent Document 1: Zhe Cao, Tomas Simon, Shih-En Wei, Yaser Sheikh, "Realtime Multi-Person 2D Pose Optimization using Part Affinity Fields", The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, P. 7291-7299
Outline of the invention
Problems to be solved by the invention
[0005]
As described above, in Patent Document 1, the posture of a person is detected based on the change in the image area of ​​the person, but since an upright image is indispensable, it may be detected accurately depending on the posture of the person. Can not. Further, Patent Document 2 is premised on the acquisition of three-dimensional information of a distance image. For this reason, there is a problem that it is difficult to accurately detect the state of a person from a two-dimensional image of a person in a related technique.
[0006]
In view of such problems, the present disclosure provides a non-temporary computer-readable medium in which a person state detection device, a person state detection method, and a program capable of improving the detection accuracy of a person's state can be improved. With the goal.
Means to solve the problem
[0007]
The person state detection device according to the present disclosure includes an acquisition means for acquiring a two-dimensional image of a person, a skeletal structure detection means for detecting the two-dimensional skeletal structure of the person based on the acquired two-dimensional image, and a skeletal structure detection means. An estimation means for estimating the upright height of the person on the two-dimensional image space based on the detected two-dimensional skeleton structure, the estimated upright height and the region of the person in the two-dimensional image. It is provided with a state detecting means for detecting the state of the person based on the height of the person.
[0008]
In the person state detection method according to the present disclosure, a two-dimensional image of a person is acquired, the two-dimensional skeletal structure of the person is detected based on the acquired two-dimensional image, and the detected two-dimensional skeletal structure is detected. The height of the person in the upright position on the two-dimensional image space is estimated based on the above, and the state of the person is determined based on the estimated height of the person in the upright position and the height of the area of ​​the person in the two-dimensional image. It is to detect.
[0009]
The person state detection program according to the present disclosure acquires a two-dimensional image of a person, detects the two-dimensional skeletal structure of the person based on the acquired two-dimensional image, and the detected two-dimensional skeletal structure. The height of the person in the upright position on the two-dimensional image space is estimated based on the above, and the state of the person is determined based on the estimated height of the person in the upright position and the height of the area of ​​the person in the two-dimensional image. It is a person state detection program for causing a computer to execute a process to be detected.
Effect of the invention
[0010]
According to the present disclosure, it is possible to provide a non-temporary computer-readable medium in which a person state detection device, a person state detection method, and a program capable of improving the detection accuracy of a person state can be provided.
A brief description of the drawing
[0011]
[Fig. 1] Fig. 1 is a flowchart showing a related monitoring method.
FIG. 2 is a configuration diagram showing an outline of a person state detection device according to an embodiment.
FIG. 3 is a configuration diagram showing a configuration of a person state detection device according to the first embodiment.
FIG. 4 is a flowchart showing a person state detection method according to the first embodiment.
FIG. 5 is a flowchart showing a method of calculating the number of height pixels according to the first embodiment.
FIG. 6 is a diagram showing a human body model according to the first embodiment.
FIG. 7 is a diagram showing a detection example of a skeletal structure according to the first embodiment.
FIG. 8 is a diagram showing a detection example of a skeletal structure according to the first embodiment.
FIG. 9 is a diagram showing a detection example of a skeletal structure according to the first embodiment.
FIG. 10 is a diagram showing a human body model according to the second embodiment.
FIG. 11 is a flowchart showing a method of calculating the number of height pixels according to the second embodiment.
FIG. 12 is a diagram showing a detection example of a skeletal structure according to a second embodiment.
FIG. 13 is a histogram for explaining a method for calculating the number of height pixels according to the second embodiment.
FIG. 14 is a flowchart showing a method of calculating the number of height pixels according to the third embodiment.
FIG. 15 is a diagram showing a detection example of a skeletal structure according to a third embodiment.
FIG. 16 is a diagram showing a three-dimensional human body model according to the third embodiment.
FIG. 17 is a diagram for explaining a method for calculating the number of height pixels according to the third embodiment.
FIG. 18 is a diagram for explaining a method for calculating the number of height pixels according to the third embodiment.
FIG. 19 is a diagram for explaining a method for calculating the number of height pixels according to the third embodiment.
FIG. 20 is a configuration diagram showing an outline of computer hardware according to an embodiment.
Embodiment for carrying out the invention
[0012]
Hereinafter, embodiments will be described with reference to the drawings. In each drawing, the same elements are designated by the same reference numerals, and duplicate explanations are omitted as necessary.
[0013]
(Examination leading to the embodiment)
In recent years, image recognition technology utilizing machine learning has been applied to various systems. As an example, consider a surveillance system that monitors images from a surveillance camera.
[0014]
FIG. 1 shows a monitoring method in a related monitoring system. As shown in FIG. 1, the surveillance system acquires an image from the surveillance camera (S101), detects a person from the acquired image (S102), and performs state recognition and attribute recognition (S103) of the person. For example, it recognizes the posture and behavior (behavior, movement line) of the person as the state of the person, and recognizes the age, gender, height, etc. of the person as the attributes of the person. Further, in the monitoring system, data is analyzed from the state and attributes of the recognized person (S104), and actions such as coping are performed based on the analysis result (S105). For example, an alert is displayed based on the recognized posture or action, or a person with an attribute such as the recognized height is monitored.
[0015]
As in the state recognition in this example, there is an increasing demand for detecting the posture and behavior of a person, especially in a surveillance system, from the surveillance camera image. For example, postures and behaviors include crouching, falling asleep, and falling.
[0016]
The inventors examined a method for detecting a state such as a person's posture or behavior from an image, and found that it is difficult to detect it easily with related technology, and it is not always possible to detect it accurately. I found a problem. With the development of deep learning in recent years, it is possible to detect the above-mentioned posture, etc. by collecting and learning a large amount of images of the posture, etc. of the detection target. However, it is difficult to collect this training data and the cost is high. Further, for example, if a part of the body of a person is hidden, the state of the person may not be detected.
[0017]
Therefore, the inventors examined a method of using skeletal estimation technology using machine learning to detect the state of a person. For example, in a related skeleton estimation technique such as OpenPose disclosed in Non-Patent Document 1, the skeleton of a person is estimated by learning the image data in which various patterns are correctly answered. In the following embodiments, by utilizing such a skeleton estimation technique, it is possible to easily detect the state of a person and improve the detection accuracy.
The skeletal structure estimated by a skeletal estimation technique such as OpenPose is composed of "key points" which are characteristic points of joints and the like and "bones (bone links)" which indicate links between key points. .. Therefore, in the following embodiments, the skeletal structure will be described using the terms "key point" and "bone", but unless otherwise specified, the "key point" corresponds to the "joint" of a person and ". "Bone" corresponds to the "bone" of a person.
[0018]
(Outline of embodiment)
FIG. 2 shows an outline of the person state detection device 10 according to the embodiment. As shown in FIG. 2, the person state detection device 10 includes an acquisition unit 11, a skeletal structure detection unit 12, an estimation unit 13, and a state detection unit 14.
[0019]
The acquisition unit 11 acquires a two-dimensional image of a person. The skeletal structure detection unit 12 detects the two-dimensional skeletal structure of a person based on the two-dimensional image acquired by the acquisition unit 11. The estimation unit 13 estimates the height of a person in an upright position in a two-dimensional image space based on the two-dimensional skeletal structure detected by the skeletal structure detection unit 12. The state detection unit 14 detects a state such as a person's posture and behavior based on the height in an upright position estimated by the estimation unit 13 and the height of a person's area in a two-dimensional image.
[0020]
As described above, in the embodiment, the two-dimensional skeletal structure of a person is detected from the two-dimensional image, and the height of the person on the two-dimensional image space estimated based on the two-dimensional skeletal structure when the person is upright and the person in the image. By detecting the state of the person from the height of the person, it can be easily detected, and it can be detected accurately regardless of the posture of the person.
[0021]
(Embodiment 1)
Hereinafter, the first embodiment will be described with reference to the drawings. FIG. 3 shows the configuration of the person state detection device 100 according to the present embodiment. The person state detection device 100 constitutes the person state detection system 1 together with the camera 200. For example, the person state detection device 100 and the person state detection system 1 are applied to the monitoring method in the monitoring system as shown in FIG. 1, detect states such as the posture and behavior of a person, and display an alarm according to the detection. Is done. The camera 200 may be provided inside the person state detection device 100.
[0022]
As shown in FIG. 3, the person state detection device 100 includes an image acquisition unit 101, a skeleton structure detection unit 102, a height calculation unit 103, a person area calculation unit 104, a state detection unit 105, and a storage unit 106. The configuration of each part (block) is an example, and may be composed of other parts as long as the method (operation) described later is possible. Further, the person state detection device 100 is realized by, for example, a computer device such as a personal computer or a server that executes a program, but may be realized by one device or by a plurality of devices on a network. May be good.
[0023]
The storage unit 106 stores information (data) necessary for the operation (processing) of the person state detection device 100. For example, the storage unit 106 is a non-volatile memory such as a flash memory, a hard disk device, or the like. The storage unit 106 stores an image acquired by the image acquisition unit 101, an image processed by the skeletal structure detection unit 102, data for machine learning, and the like. The storage unit 106 may be an external storage device or an external storage device on the network. That is, the person state detection device 100 may acquire necessary images, machine learning data, and the like from an external storage device.
[0024]
The image acquisition unit 101 acquires a two-dimensional image captured by the camera 200 from the camera 200 connected communicably. The camera 200 is an imaging unit such as a surveillance camera that captures a person, and the image acquisition unit 101 acquires an image of a person from the camera 200.
[0025]The skeletal structure detection unit 102 detects the two-dimensional skeletal structure of a person in the image based on the acquired two-dimensional image. The skeletal structure detection unit 102 detects the skeletal structure of a person based on the characteristics of the recognized person's joints and the like by using a skeletal estimation technique using machine learning. The skeletal structure detection unit 102 uses, for example, a skeletal estimation technique such as OpenPose of Non-Patent Document 1.
[0026]
The height calculation unit (height estimation unit) 103 calculates (estimates) the height (referred to as the number of height pixels) of a person in a two-dimensional image when standing upright, based on the detected two-dimensional skeletal structure. It can also be said that the number of height pixels is the height of a person in a two-dimensional image (the length of the whole body of the person in the two-dimensional image space). The height calculation unit 103 obtains the number of height pixels (number of pixels) from the length (length on the two-dimensional image space) of each bone of the detected skeleton structure. In the present embodiment, the number of height pixels is obtained by totaling the lengths of the bones from the head to the foot among the bones of the skeletal structure. If the skeletal structure detection unit 102 (skeletal estimation technique) does not output the crown and feet, it can be corrected by multiplying by a constant if necessary.
[0027]
The person area calculation unit 104 calculates the height of the person in the image (referred to as the person area height) based on the acquired two-dimensional image. The person area calculation unit 104 extracts a person area in the image and calculates the height (number of pixels) of the person area in the vertical direction. For example, when a person is crouching, the number of pixels from the top of the head to the toes in that state is calculated.
[0028]
The state detection unit 105 detects the state of a person based on the calculated number of height pixels and the height of the person area. In this example, postures such as standing upright, crouching, and falling asleep are detected as the state of the person. The state detection unit 105 may detect an action (fall, etc.) from a temporal change in posture as a state of a person. The state detection unit 105 obtains the ratio between the number of height pixels and the height of the person area, and detects the posture of the person from the ratio.
[0029]
4 and 5 show the operation of the person state detection device 100 according to the present embodiment. FIG. 4 shows the flow from image acquisition to state detection in the person state detection device 100, and FIG. 5 shows the flow of the height pixel number calculation process (S203) of FIG.
[0030]
As shown in FIG. 4, the person state detection device 100 acquires an image from the camera 200 (S201). The image acquisition unit 101 acquires an image of a person for detection of a skeletal structure or a person area.
[0031]
Subsequently, the person state detection device 100 detects the skeleton structure of the person based on the acquired image of the person (S202). FIG. 6 shows the skeletal structure of the human body model 300 detected at this time, and FIGS. 7 to 9 show an example of detecting the skeletal structure. The skeletal structure detection unit 102 detects the skeletal structure of the human body model (two-dimensional skeletal model) 300 as shown in FIG. 6 from a two-dimensional image by using a skeletal estimation technique such as OpenPose. The human body model 300 is a two-dimensional model composed of key points such as human joints and bones connecting the key points.
[0032]
The skeletal structure detection unit 102, for example, extracts feature points that can be key points from an image, and detects each key point of a person by referring to information obtained by machine-learning the image of the key points. In the example of FIG. 6, as key points of a person, head A1, neck A2, right shoulder A31, left shoulder A32, right elbow A41, left elbow A42, right hand A51, left hand A52, right waist A61, left waist A62, right knee A71. , Left knee A72, right foot A81, left foot A82 are detected. Further, as the bones of the person connecting these key points, the bone B1 connecting the head A1 and the neck A2, the bones B21 and B22 connecting the neck A2 and the right shoulder A31 and the left shoulder A32, respectively, the right shoulder A31 and the left shoulder A32 and the right Bone B31 and B32 connecting elbow A41 and left elbow A42, respectively, bone B41 and bone B42 connecting right elbow A41 and left elbow A42 with right hand A51 and left hand A52, respectively, connecting neck A2 and right waist A61 and left waist A62, respectively. Bones B51 and B52, right waist A61 and left waist A62, right knee A71 and left knee A72, respectively, bone B61 and bone B62, right knee A71 and left knee A72, right foot A81 and left foot A82, respectively. B72 is detected.
[0033]
FIG. 7 is an example of detecting an upright person. In FIG. 7, an upright person is imaged from the front, and bones B1, bone B51 and bone B52, bones B61 and bone B62, bones B71 and bone B72 viewed from the front are detected without overlapping, and the right foot is detected. Bone B61 and Bone B71 are slightly bent more than Bone B62 and Bone B72 of the left foot. FIG. 8 is an example of detecting a person in a crouched state. In FIG. 8, a crouching person is imaged from the right side, and bone B1, bone B51 and bone B52, bone B61 and bone B62, bone B71 and bone B72, respectively, viewed from the right side are detected, and bone B61 of the right foot is detected. And bone B71 and bone B62 and bone B72 of the left foot are greatly bent and overlapped. FIG. 9 is an example of detecting a person who is sleeping. In FIG. 9, a sleeping person is imaged from diagonally left front, and bone B1, bone B51 and bone B52, bone B61 and bone B62, bone B71 and bone B72 viewed from diagonally left front are detected, respectively, and the right foot. Bone B61 and B71 and bone B62 and bone B72 of the left foot are bent and overlapped.
[0034]
Subsequently, the person state detection device 100 performs a height pixel number (H1) calculation process based on the detected skeleton structure (S203). In the height pixel number calculation process, as shown in FIG. 5, the height calculation unit 103 acquires the length of each bone (S211) and totals the acquired lengths of each bone (S212). The height calculation unit 103 acquires the length of a bone on a two-dimensional image of a foot from the head of a person, and obtains the number of height pixels. That is, from the image in which the skeletal structure is detected, among the bones of FIG. 6, bone B1 (length L1), bone B51 (length L21), bone B61 (length L31) and bone B71 (length L41), or , Bone B1 (length L1), bone B52 (length L22), bone B62 (length L32) and bone B72 (length L42) are acquired. The length of each bone can be obtained from the coordinates of each key point in the two-dimensional image. The sum of these is calculated as the number of height pixels (H1) by multiplying L1 + L21 + L31 + L41 or L1 + L22 + L32 + L42 by a correction constant. When both values ​​can be calculated, for example, the longer value is taken as the number of height pixels. That is, each bone has the longest length in the image when it is imaged from the front, and it is displayed short when it is tilted in the depth direction with respect to the camera. Therefore, it is more likely that the longer bone is imaged from the front, which is considered to be closer to the true value. Therefore, it is preferable to select the longer value.
[0035]
In the example of FIG. 7, bone B1, bone B51 and bone B52, bone B61 and bone B62, bone B71 and bone B72 are detected without overlapping. The total of these bones, L1 + L21 + L31 + L41 and L1 + L22 + L32 + L42, is obtained, and for example, the value obtained by multiplying L1 + L22 + L32 + L42 on the left foot side where the detected bone length is long by a correction constant is taken as the number of height pixels.
[0036]
In the example of FIG. 8, bone B1, bone B51 and bone B52, bone B61 and bone B62, bone B71 and bone B72 are detected, respectively, and bone B61 and bone B71 of the right foot and bone B62 and bone B72 of the left foot overlap each other. .. The total of these bones, L1 + L21 + L31 + L41 and L1 + L22 + L32 + L42, is obtained, and for example, the value obtained by multiplying L1 + L21 + L31 + L41 on the right foot side where the detected bone length is long by a correction constant is taken as the number of height pixels.
[0037]
In the example of FIG. 9, bone B1, bone B51 and bone B52, bone B61 and bone B62, bone B71 and bone B72 are detected, respectively, and bone B61 and bone B71 of the right foot and bone B62 and bone B72 of the left foot overlap each other. .. The total of these bones, L1 + L21 + L31 + L41 and L1 + L22 + L32 + L42, is obtained, and for example, the value obtained by multiplying L1 + L22 + L32 + L42 on the left foot side where the detected bone length is long by a correction constant is taken as the number of height pixels.
[0038]
On the other hand, as shown in FIG. 4, the person state detection device 100 calculates the height of the person area (H2) in the image based on the image acquired from the camera 200 (S204). The person area calculation unit 104 detects the upper end and the lower end of the person area in the image in the height direction, and calculates the height (number of pixels) thereof. The height of the person area can be obtained from the coordinates of the end of the person area (upper end and lower end in the height direction) in the image.
[0039]
In FIG. 7, the entire person is shown from the front while the person is upright. In this example, the length from the upper end of the head (top of the head) to the lower end of the foot (the tip of the left foot) of the person is defined as the height of the person area (H2). In FIG. 8, the entire person is shown from the right side while the person is crouching down. In this example, the length from the upper end of the head (top of the head) to the lower end of the foot (the base of the right toe) of the person as the end of the person area is defined as the person area height (H2). In the figure, the whole person is shown from the front left with the person lying down. In this example, the length from the upper end of the head (top of the head) to the lower end of the foot (right knee) of the person as the end of the person area is defined as the person area height (H2).
[0040]
Subsequently, the person state detection device 100 detects the state of a person based on the number of height pixels (H1) and the height of the person area (H2) (S205 to S208). The state detection unit 105 obtains the ratio of the number of height pixels (H1) to the height of the person region (H2), that is, H2 / H1 (S205). In this example, the upright state, the crouching state, and the sleeping state are detected by comparing H2 / H1 with the threshold value (0.5 = 1/2 and 0.2 = 1/5). The threshold value is an example and is not limited to this. In addition, other thresholds are used to detect a state in which both hands are raised (for example, when H2 / H1 = 1 or more) or a state in which the person is sitting on a chair (for example, when H2 / H1 = 0.5 to 0.7). You may.
[0041]
The state detection unit 105 detects that a person is upright when H2 / H1 is larger than 0.5 (S206). For example, in FIG. 7, if H1 = L1 + L22 + L32 + L42 = 100 pixels and H2 = 110 pixels from the upper end of the head to the lower end of the foot, H2 / H1 = 110/100 = 1.1, and the person is upright. Judge.
[0042]
Further, when H2 / H1 is 0.5 or less and larger than 0.2, the state detection unit 105 detects that the person is crouching (S207). For example, in FIG. 8, if H1 = L1 + L21 + L31 + L41 = 110 pixels and H2 = 50 pixels from the upper end of the head to the lower end of the foot, H2 / H1 = 50/110 = 0.45, and the person is crouching. to decide.
[0043]
Further, the state detection unit 105 detects that the person is asleep, for example, when H2 / H1 is 0.2 or less (S208). For example, in FIG. 9, if H1 = L1 + L22 + L32 + L42 = 110 pixels and H2 = 20 pixels from the upper end of the head to the lower end of the foot, H2 / H1 = 20/110 = 0.18, and it is determined that the person is sleeping. do.
[0044]
As described above, in the present embodiment, the skeletal structure of a person is detected from a two-dimensional image, and the number of height pixels (height when standing upright on a two-dimensional image space) obtained from the detected skeletal structure and the image. Detects the state of a person based on the ratio of the height of the person area. As a result, the state of the person can be easily detected because only the height ratio is required without using complicated calculation or machine learning. For example, by detecting the skeletal structure using the skeletal estimation technique, it is possible to detect the state of a person without collecting training data. Also, the bones of a person Since the information of the case structure is used, the state of the person can be detected regardless of the posture of the person.
[0045]
Further, in the present embodiment, the number of height pixels is obtained by totaling the lengths of the detected bones of the skeletal structure as the estimated height. Since the height can be obtained by summing the lengths of the bones from the head to the legs, the height can be estimated by a simple method and the state of the person can be detected. In addition, since it is only necessary to detect the skeleton from the head to the foot by skeletal estimation technology using machine learning, the height can be estimated accurately even when the entire person is not always shown in the image, such as when crouching down. , Can detect the state of a person.
[0046]
(Embodiment 2)
Next, the second embodiment will be described. In the present embodiment, in the height pixel number calculation process of the first embodiment, the number of height pixels is used by using a human body model showing the relationship between the length of each bone and the length of the whole body (height on the two-dimensional image space). Is calculated. The process is the same as that of the first embodiment except for the height pixel number calculation process.
[0047]
FIG. 10 is a human body model (two-dimensional skeleton model) 301 showing the relationship between the length of each bone in the two-dimensional image space and the length of the whole body in the two-dimensional image space used in the present embodiment. As shown in FIG. 10, the relationship between the length of each bone of an average person and the length of the whole body (the ratio of the length of each bone to the length of the whole body) is associated with each bone of the human body model 301. For example, the length of the head bone B1 is the length of the whole body x 0.2 (20%), the length of the bone B41 of the right hand is the length of the whole body x 0.15 (15%), and the length of the right foot. The length of bone B71 is the length of the whole body × 0.25 (25%). By storing the information of the human body model 301 in the storage unit 106, the average length of the whole body can be obtained from the length of each bone. In addition to the human body model of an average person, a human body model may be prepared for each attribute of the person such as age, gender, and nationality. As a result, the length (height) of the whole body can be appropriately obtained according to the attributes of the person.
[0048]
FIG. 11 is a height pixel number calculation process according to the present embodiment, and shows the flow of the height pixel number calculation process (S203) of FIG. 4 in the first embodiment. In the height pixel number calculation process of the present embodiment, as shown in FIG. 11, the height calculation unit 103 acquires the length of each bone (S301). The height calculation unit 103 acquires the lengths (lengths in the two-dimensional image space) of all the bones in the skeletal structure detected as in the first embodiment. FIG. 12 is an example in which a person in a crouched state is imaged from diagonally right behind and the skeletal structure is detected. In this example, since the face and left side of the person are not shown, the bones of the head and the bones of the left arm and the left hand cannot be detected. Therefore, the lengths of the detected bones B21, B22, B31, B41, B51, B52, B61, B62, B71, and B72 are acquired.
[0049]
Subsequently, the height calculation unit 103 calculates the number of height pixels from the length of each bone based on the human body model (S302). The height calculation unit 103 refers to the human body model 301 showing the relationship between each bone and the length of the whole body as shown in FIG. 10, and obtains the number of height pixels from the length of each bone. For example, since the length of the bone B41 on the right hand is the length of the whole body × 0.15, the number of height pixels based on the bone B41 is obtained by the length of the bone B41 / 0.15. Further, since the length of the bone B71 of the right foot is the length of the whole body × 0.25, the number of height pixels based on the bone B71 is obtained from the length of the bone B71 / 0.25.
[0050]
The human body model referred to at this time is, for example, a human body model of an average person, but a human body model may be selected according to the attributes of the person such as age, gender, and nationality. For example, when a person's face is shown in the captured image, the attribute of the person is identified based on the face, and the human body model corresponding to the identified attribute is referred to. It is possible to recognize the attributes of a person from the features of the face in the image by referring to the information obtained by machine-learning the face for each attribute. Further, when the attribute of the person cannot be identified from the image, the human body model of the average person may be used.
Further, the number of height pixels calculated from the length of the bone may be corrected by the camera parameter. The camera parameter is an image imaging parameter, and for example, the camera parameter is the posture, position, imaging angle, focal distance, and the like of the camera 200. The camera 200 can capture an object whose length is known in advance, and obtain camera parameters from the image. For example, when the camera is taken at a high position and looking down at a person, the horizontal length of the shoulder-width bones, etc. is not affected by the depression angle of the camera in the two-dimensional skeletal structure, but the vertical length of the neck-waist bones, etc. The length decreases as the depression angle of the camera increases. Then, the number of height pixels calculated from the horizontal length of the shoulder-width bone or the like tends to be larger than the actual number. Therefore, by utilizing the camera parameters, it is possible to know at what angle the person is looking down at the camera, and it is possible to correct the two-dimensional skeleton structure as if it was taken from the front using this depression angle information. This makes it possible to calculate the number of height pixels more accurately.
[0051]
Subsequently, the height calculation unit 103 calculates the optimum value of the number of height pixels (S303). The height calculation unit 103 calculates the optimum value of the number of height pixels from the number of height pixels obtained for each bone. For example, as shown in FIG. 13, a histogram of the number of height pixels obtained for each bone is generated, and a large number of height pixels is selected from the histogram. That is, the number of height pixels longer than the others is selected from the plurality of height pixels obtained based on the plurality of bones. For example, the top 30% is set as a valid value, and in FIG. 13, the number of height pixels by bones B71, B61, and B51 is selected. The average number of selected height pixels may be obtained as the optimum value, or the largest number of height pixels may be used as the optimum value. Since the height is calculated from the length of the bone in the two-dimensional image, the length of the bone is taken from the front when the bone is not made from the front, that is, when the bone is tilted in the depth direction when viewed from the camera. It will be shorter than the case. Then, a value having a large number of height pixels is more likely to be imaged from the front than a value having a small number of height pixels, and is a more plausible value. Therefore, a larger value is set as the optimum value.
[0052]
As described above, in the present embodiment, the number of height pixels is based on the detected bones of the skeletal structure using a human body model showing the relationship between the bones on the two-dimensional image space and the length of the whole body. By obtaining, the height of the person is estimated, and the state of the person is detected as in the first embodiment. As a result, even if not all the skeletons from the head to the legs can be obtained, the height can be estimated from some bones and the state of the person can be detected. In particular, by adopting a larger value among the heights (number of height pixels) obtained from a plurality of bones, the height can be estimated accurately and the state of the person can be detected.
[0053]
(Embodiment 3)
Next, the third embodiment will be described. In the present embodiment, in the height pixel number calculation process of the first embodiment, the height pixel number is calculated by fitting (fitting) the three-dimensional human body model to the two-dimensional skeleton structure. The process is the same as that of the first embodiment except for the height pixel number calculation process.
[0054]
FIG. 14 is a height pixel number calculation process according to the present embodiment, and shows the flow of the height pixel number calculation process (S203) of FIG. 4 in the first embodiment. In the height pixel number calculation process of the present embodiment, as shown in FIG. 14, the height calculation unit 103 first calculates the camera parameters based on the image captured by the camera 200 (S401). The camera parameter is an image imaging parameter as in the second embodiment. The height calculation unit 103 extracts an object whose length is known in advance from a plurality of images captured by the camera 200, and obtains a camera parameter from the size (number of pixels) of the extracted object. The camera parameters may be obtained in advance, and the obtained camera parameters may be acquired as needed.
Subsequently, the height calculation unit 103 adjusts the arrangement and height of the three-dimensional human body model (S402). The height calculation unit 103 prepares a three-dimensional human body model for calculating the number of height pixels for the two-dimensional skeleton structure detected as in the first embodiment, and arranges them in the same two-dimensional image based on the camera parameters. do. Specifically, the "relative positional relationship between the camera and the person in the real world" is specified from the camera parameters and the two-dimensional skeleton structure. For example, assuming that the position of the camera is the coordinate (0,0,0), the coordinate (x, y, z) of the position where the person is standing (or sitting) is specified. Then, by assuming an image in which a three-dimensional human body model is placed at the same position (x, y, z) as the specified person and captured, the two-dimensional skeletal structure and the three-dimensional human body model are superimposed.
[0055]
FIG. 15 is an example in which a crouching person is imaged diagonally from the front left and the two-dimensional skeletal structure 401 is detected. The two-dimensional skeleton structure 401 has two-dimensional coordinate information. It is preferable that all bones are detected, but some bones may not be detected. For this two-dimensional skeleton structure 401, a three-dimensional human body model 402 as shown in FIG. 16 is prepared. The three-dimensional human body model (three-dimensional skeleton model) 402 has three-dimensional coordinate information and is a model of a skeleton having the same shape as the two-dimensional skeleton structure 401. Then, as shown in FIG. 17, the prepared three-dimensional human body model 402 is arranged and superimposed on the detected two-dimensional skeleton structure 401. In addition, the height of the three-dimensional human body model 402 is adjusted so as to match the two-dimensional skeleton structure 401.
[0056]
The three-dimensional human body model 402 prepared at this time may be a model in a state close to the posture of the two-dimensional skeleton structure 401 as shown in FIG. 17, or may be a model in an upright state. For example, a technique of estimating a posture in a three-dimensional space from a two-dimensional image using machine learning may be used to generate a three-dimensional human body model 402 of the estimated posture. By learning the information of the joints in the two-dimensional image and the joints in the three-dimensional space, the three-dimensional posture can be estimated from the two-dimensional image.
[0057]
Subsequently, the height calculation unit 103 fits the three-dimensional human body model into the two-dimensional skeletal structure (S403). As shown in FIG. 18, the height calculation unit 103 superimposes the three-dimensional human body model 402 on the two-dimensional skeletal structure 401 so that the postures of the three-dimensional human body model 402 and the two-dimensional skeletal structure 401 match. Transform the dimensional human body model 402. That is, the height, body orientation, and joint angle of the three-dimensional human body model 402 are adjusted and optimized so that there is no difference from the two-dimensional skeletal structure 401. For example, the joints of the three-dimensional human body model 402 are rotated within the range of movement of the person, the entire three-dimensional human body model 402 is rotated, and the overall size is adjusted. The fitting of the three-dimensional human body model and the two-dimensional skeleton structure is performed in the two-dimensional space (two-dimensional coordinates). That is, a three-dimensional human body model is mapped in a two-dimensional space, and the three-dimensional human body model is converted into a two-dimensional skeleton structure in consideration of how the deformed three-dimensional human body model changes in the two-dimensional space (image). Optimize.
[0058]
Subsequently, the height calculation unit 103 calculates the number of height pixels of the fitted three-dimensional human body model (S404). As shown in FIG. 19, the height calculation unit 103 obtains the number of height pixels of the three-dimensional human body model 402 in that state when the difference between the three-dimensional human body model 402 and the two-dimensional skeleton structure 401 disappears and the postures match. With the optimized three-dimensional human body model 402 upright, the length of the whole body in the two-dimensional space is obtained based on the camera parameters. For example, the number of height pixels is calculated from the length of bones (number of pixels) from the head to the foot when the three-dimensional human body model 402 is upright. Similar to the first embodiment, the lengths of the bones from the head to the foot of the three-dimensional human body model 402 may be totaled.
[0059]
As described above, in the present embodiment, the three-dimensional human body model is fitted to the two-dimensional skeleton structure based on the camera parameters, and the height pixels are based on the three-dimensional human body model.By obtaining the number, the height of the person is estimated, and the state of the person is detected as in the first embodiment. As a result, even if all the bones are not shown in the front, that is, even if all the bones are shown diagonally and the error is large, the height can be estimated accurately and the state of the person can be detected. can. If the methods of the first to third embodiments are applicable, the height may be estimated using all the methods (or a method in which any of them is combined). In that case, the one closer to the average height of the person may be set as the optimum value.
[0060]
Note that each configuration in the above-described embodiment is configured by hardware and / or software, and may be composed of one hardware or software, or may be composed of a plurality of hardware or software. The functions (processing) of the person state detection devices 10 and 100 may be realized by a computer 20 having a processor 21 such as a CPU (Central Processing Unit) and a memory 22 which is a storage device, as shown in FIG. For example, a program (personal state detection program) for performing the method in the embodiment may be stored in the memory 22, and each function may be realized by executing the program stored in the memory 22 on the processor 21.
[0061]
These programs are stored using various types of non-transitory computer readable medium and can be supplied to the computer. Non-temporary computer-readable media include various types of tangible storage media. Examples of non-temporary computer-readable media include magnetic recording media (eg, flexible discs, magnetic tapes, hard disk drives), optomagnetic recording media (eg, optomagnetic discs), CD-ROMs (Read Only Memory), CD-Rs, etc. Includes CD-R / W, semiconductor memory (eg, mask ROM, PROM (Programmable ROM), EPROM (Erasable PROM), flash ROM, RAM (random access memory)). The program may also be supplied to the computer by various types of transient computer readable medium. Examples of temporary computer readable media include electrical signals, optical signals, and electromagnetic waves. The temporary computer-readable medium can supply the program to the computer via a wired communication path such as an electric wire and an optical fiber, or a wireless communication path.
[0062]
Further, the present disclosure is not limited to the above-described embodiment, and can be appropriately changed without departing from the spirit. For example, although the state of a person is detected in the above, the state of an animal other than a person having a skeletal structure (mammalia, reptile, bird, amphibian, fish, etc.) may be detected.
[0063]
Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above embodiments. The structure and details of the present disclosure may be modified in various ways within the scope of the present disclosure as may be understood by those skilled in the art.
[0064]
A part or all of the above embodiment may be described as in the following appendix, but is not limited to the following.
(Appendix 1)
Acquisition means for acquiring a two-dimensional image of a person,
A skeletal structure detecting means for detecting the two-dimensional skeletal structure of the person based on the acquired two-dimensional image,
An estimation means for estimating the height of the person in the upright position on the two-dimensional image space based on the detected two-dimensional skeleton structure,
A state detecting means for detecting the state of the person based on the estimated height when standing upright and the height of the area of ​​the person in the two-dimensional image.
A person status detection device equipped with.
(Appendix 2)
The state detecting means detects the state of the person based on the ratio of the height of the person when standing upright to the height of the area of ​​the person.
The person state detection device described in Appendix 1.
(Appendix 3)
The state detecting means detects that the person is in an upright state based on the result of comparison between the ratio and a predetermined threshold value.
Person status detection device described in Appendix 2.
(Appendix 4)
The state detecting means detects that the person is crouching based on the result of comparison between the ratio and a predetermined threshold value.
Person status detection device described in Appendix 2.
(Appendix 5)
The state detecting means detects that the person is in a sleeping state based on the result of comparison between the ratio and a predetermined threshold value.
Person status detection device described in Appendix 2.
(Appendix 6)
The estimation means estimates the height of the person in an upright position based on the length of the bone in the two-dimensional image space included in the two-dimensional skeletal structure.
The person state detection device according to any one of Supplementary note 1 to 5.
(Appendix 7)
The estimation means estimates the height of the person in an upright position based on the total length of the bones from the foot to the head included in the two-dimensional skeletal structure.
The person state detection device described in Appendix 6.
(Appendix 8)
The estimation means estimates the height of the person in an upright position based on a two-dimensional skeletal model showing the relationship between the length of the bone and the length of the whole body of the person in the two-dimensional image space.
The person state detection device described in Appendix 6.
(Appendix 9)
The estimation means estimates the height of the person in an upright position based on a three-dimensional skeleton model fitted to the two-dimensional skeleton structure based on the imaging parameters of the two-dimensional image.
The person state detection device according to any one of Supplementary note 1 to 5.
(Appendix 10)
Acquire a two-dimensional image of a person,
Detecting the 2D skeletal structure of the person based on the acquired 2D image,
Based on the detected two-dimensional skeletal structure, the height of the person in the two-dimensional image space when standing upright is estimated.
The state of the person is detected based on the estimated height when standing upright and the height of the area of ​​the person in the two-dimensional image.
Person status detection method.
(Appendix 11)
In the detection of the state, the state of the person is detected based on the ratio of the height of the person when standing upright and the height of the area of ​​the person.
The person state detection method described in Appendix 10.
(Appendix 12)
Acquire a two-dimensional image of a person,
Detecting the 2D skeletal structure of the person based on the acquired 2D image,
Based on the detected two-dimensional skeletal structure, the height of the person in the two-dimensional image space when standing upright is estimated.
The state of the person is detected based on the estimated height when standing upright and the height of the area of ​​the person in the two-dimensional image.
A person state detection program for causing a computer to execute processing.
(Appendix 13)
In the detection of the state, the state of the person is detected based on the ratio of the height of the person when standing upright and the height of the area of ​​the person.
The person status detection program described in Appendix 12.
(Appendix 14)
Equipped with a camera and a person status detection device
The person state detection device is
Acquisition means for acquiring a two-dimensional image of a person taken from the camera,
A skeletal structure detecting means for detecting the two-dimensional skeletal structure of the person based on the acquired two-dimensional image,
An estimation means for estimating the height of the person in the upright position on the two-dimensional image space based on the detected two-dimensional skeleton structure,
A state detecting means for detecting the state of the person based on the estimated height when standing upright and the height of the area of ​​the person in the two-dimensional image.
A person status detection system equipped with.
(Appendix 15)
The state detecting means detects the state of the person based on the ratio of the height of the person when standing upright to the height of the area of ​​the person.
The person status detection system described in Appendix 14.
Description of the sign
[0065]
1 Person status detection system
10 Person status detection device
11 Acquisition department
12 Skeletal structure detector
13 Estimator
14 Status detector
20 Computer
21 processor
22 Memory
100 Person status detector
101 Image acquisition department
102 Skeletal structure detector
103 Height calculation department
104 Person area calculation unit
105 Status detector
106 Storage unit
200 camera
300, 301 human body model
401 Two-dimensional skeleton structure
402 3D human body model
The scope of the claims
[Claim 1]
Acquisition means for acquiring a two-dimensional image of a person,
A skeletal structure detecting means for detecting the two-dimensional skeletal structure of the person based on the acquired two-dimensional image,
An estimation means for estimating the height of the person in the upright position on the two-dimensional image space based on the detected two-dimensional skeleton structure,
A state detecting means for detecting the state of the person based on the estimated height when standing upright and the height of the area of ​​the person in the two-dimensional image.
A person status detection device equipped with.
[Claim 2]
The state detecting means detects the state of the person based on the ratio of the height of the person when standing upright to the height of the area of ​​the person.
The person state detection device according to claim 1.
[Claim 3]
The state detecting means detects that the person is crouching based on the result of comparison between the ratio and a predetermined threshold value.
The person state detection device according to claim 2.
[Claim 4]
The state detecting means detects that the person is in a sleeping state based on the result of comparison between the ratio and a predetermined threshold value.
The person state detection device according to claim 2.
[Claim 5]
The estimation means estimates the height of the person in an upright position based on the length of the bone in the two-dimensional image space included in the two-dimensional skeletal structure.
The person state detection device according to any one of claims 1 to 4.
[Claim 6]
The estimation means estimates the height of the person in an upright position based on the total length of the bones from the foot to the head included in the two-dimensional skeletal structure.
The person state detection device according to claim 5.
[Claim 7]
The estimation means estimates the height of the person in an upright position based on a two-dimensional skeletal model showing the relationship between the length of the bone and the length of the whole body of the person in the two-dimensional image space.
The person state detection device according to claim 5.
[Claim 8]
The estimation means estimates the height of the person in an upright position based on a three-dimensional skeleton model fitted to the two-dimensional skeleton structure based on the imaging parameters of the two-dimensional image.
The person state detection device according to any one of claims 1 to 4.
[Claim 9]
Acquire a two-dimensional image of a person
Detecting the 2D skeletal structure of the person based on the acquired 2D image,
Based on the detected two-dimensional skeletal structure, the height of the person in the two-dimensional image space when standing upright is estimated.
The state of the person is detected based on the estimated height when standing upright and the height of the area of ​​the person in the two-dimensional image.
Person status detection method.
[Claim 10]
Acquire a two-dimensional image of a person
Detecting the 2D skeletal structure of the person based on the acquired 2D image,
Based on the detected two-dimensional skeletal structure, the height of the person in the two-dimensional image space when standing upright is estimated.
The state of the person is detected based on the estimated height when standing upright and the height of the area of ​​the person in the two-dimensional image.
A non-temporary computer-readable medium containing a program for causing a computer to execute processing.