WO2023058221A1 - 画像処理装置及び画像処理方法 - Google Patents
画像処理装置及び画像処理方法 Download PDFInfo
- Publication number
- WO2023058221A1 WO2023058221A1 PCT/JP2021/037332 JP2021037332W WO2023058221A1 WO 2023058221 A1 WO2023058221 A1 WO 2023058221A1 JP 2021037332 W JP2021037332 W JP 2021037332W WO 2023058221 A1 WO2023058221 A1 WO 2023058221A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- operator
- image
- posture
- imaging device
- image processing
- Prior art date
Links
- 238000012545 processing Methods 0.000 title claims abstract description 77
- 238000003672 processing method Methods 0.000 title claims abstract description 9
- 238000003384 imaging method Methods 0.000 claims abstract description 99
- 238000006073 displacement reaction Methods 0.000 claims abstract description 19
- 238000001514 detection method Methods 0.000 claims description 39
- 238000000034 method Methods 0.000 claims description 30
- 238000004364 calculation method Methods 0.000 claims description 27
- 238000012937 correction Methods 0.000 claims description 13
- 230000036544 posture Effects 0.000 description 96
- 230000009471 action Effects 0.000 description 42
- 210000000245 forearm Anatomy 0.000 description 34
- 230000006870 function Effects 0.000 description 33
- 210000001503 joint Anatomy 0.000 description 28
- 230000006399 behavior Effects 0.000 description 24
- 210000002414 leg Anatomy 0.000 description 19
- 238000010586 diagram Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 11
- 210000000707 wrist Anatomy 0.000 description 11
- 238000010801 machine learning Methods 0.000 description 10
- 210000003423 ankle Anatomy 0.000 description 9
- 210000003811 finger Anatomy 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 6
- 210000002310 elbow joint Anatomy 0.000 description 6
- 230000010365 information processing Effects 0.000 description 5
- 230000001681 protective effect Effects 0.000 description 5
- 210000000629 knee joint Anatomy 0.000 description 4
- 210000000323 shoulder joint Anatomy 0.000 description 4
- 239000000284 extract Substances 0.000 description 3
- 239000011521 glass Substances 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 210000003813 thumb Anatomy 0.000 description 3
- 239000000853 adhesive Substances 0.000 description 2
- 230000001070 adhesive effect Effects 0.000 description 2
- 239000002390 adhesive tape Substances 0.000 description 2
- 230000037237 body shape Effects 0.000 description 2
- 244000309466 calf Species 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 210000001145 finger joint Anatomy 0.000 description 2
- 210000004932 little finger Anatomy 0.000 description 2
- 210000003857 wrist joint Anatomy 0.000 description 2
- 210000001015 abdomen Anatomy 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000005357 flat glass Substances 0.000 description 1
- 210000004247 hand Anatomy 0.000 description 1
- 210000001624 hip Anatomy 0.000 description 1
- 210000004394 hip joint Anatomy 0.000 description 1
- 210000003127 knee Anatomy 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 210000001699 lower leg Anatomy 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 210000000689 upper leg Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
Definitions
- the present invention relates to an image processing device and an image processing method for recognizing human behavior.
- a pair of three-dimensional image data and two-dimensional image data captured by a camera is acquired from the photographer side, the position and orientation of the camera are estimated based on the three-dimensional image data, and the estimated position and orientation of the camera are combined with the previously obtained position and orientation.
- extracting a skeletal posture parameter representing the skeletal posture of the photographer based on the body shape parameter representing the body shape of the photographer and the three-dimensional image data extracting the posture feature of the photographer based on the skeletal posture parameter; Extracting image feature values based on two-dimensional image data, and recognizing the behavior of the photographer based on the posture feature value, the image feature value, and pre-learned classifier parameters for identifying the behavior of the photographer.
- Patent Document 1 A method for doing so is known (Patent Document 1).
- the problem to be solved by the present invention is to provide an image processing device and an image processing method that can correctly recognize the posture of a person even when the mounting position of the camera is shifted.
- the present invention acquires an image including the person from an imaging device attached to at least one of a person and an article worn by the person, detects the posture of the person from the image, and detects the position of the person from the image.
- the above problem is solved by calculating the positional deviation of the imaging device using a reference position as a reference for calculating the deviation, and correcting the posture using the positional deviation.
- the human posture can be correctly recognized.
- FIG. 1 is a block diagram showing one embodiment of an image processing system including an image processing apparatus according to the present invention
- FIG. FIG. 2 is a plan view showing the wearable terminal including the imaging device of FIG. 1 and the operator's left forearm and left hand
- FIG. 2 is a perspective view showing the head of a human (operator) and a helmet to which a camera, which is the imaging device of FIG. 1, is attached
- FIG. 2 is a plan view showing the camera, which is the imaging device of FIG. 1, and the operator's right arm
- 2B is a diagram showing an example of an image of an operator captured by the wearable terminal of FIG. 2A
- FIG. 2C is a diagram showing an example of an image of an operator captured by the camera of FIG. 2B
- FIG. 2C is a diagram showing an example of an image of an operator captured by the camera of FIG. 2C;
- 3B is a diagram showing an example of a method for detecting the posture of the operator shown in FIG. 3A;
- FIG. 3C is a diagram showing an example of a method for detecting the posture of the operator shown in FIG. 3B;
- FIG. 3D is a diagram showing an example of a method for detecting the posture of the operator shown in FIG. 3C;
- FIG. 2B is a diagram showing another example of an image of an operator captured by the wearable terminal of FIG. 2A;
- FIG. FIG. 2D is a diagram showing another example of an image of an operator captured by the camera of FIG. 2C;
- FIG. 2C is a diagram showing another example of an image of an operator captured by the camera of FIG. 2B (No. 1);
- FIG. 2C is a diagram showing another example of the image of the operator captured by the camera of FIG. 2B (No. 2);
- 2 is a flowchart showing an example of an information processing procedure in the image processing system of FIG. 1;
- FIG. 1 is a block diagram showing an image processing system 1 according to the invention.
- the image processing system 1 is a device that detects actions of an operator (corresponding to a human according to the present invention) who performs a predetermined operation. Predetermined operations are not particularly limited. , also referred to as a "dealer.”
- An operator whose behavior is detected by the image processing system 1 (hereinafter also simply referred to as an "operator”) is not particularly limited, and examples thereof include vehicle crew, factory workers, dealer mechanics, and the like.
- the actions detected by the image processing system 1 are, for example, the action of a vehicle occupant reaching for a switch on an on-board device such as an air conditioner, and the action of an assembly factory worker pulling a torque wrench out of a tool box. and the action of touching the switch that raises and lowers the lift that lifts the vehicle by the dealer's mechanic.
- the image processing system 1 of the present embodiment detects the above-described behavior, thereby determining whether or not the function associated with the operation of the on-vehicle device has been properly manifested. It is possible to confirm whether or not the assembly work was performed according to a predetermined procedure, and whether or not the mechanics engaged in vehicle maintenance at the vehicle dealer performed maintenance according to the manual.
- the image processing system 1 includes an imaging device 11, a display device 12, and an image processing device 13.
- Devices that make up the image processing system 1 are connected to each other by known means such as a wired or wireless LAN so that they can exchange data with each other.
- the numbers of the imaging device 11 and the number of the display devices 12 are not particularly limited as long as they are at least one or more.
- the imaging device 11 and the display device 12 do not have to be installed together with the image processing device 13 and may be installed at a location away from the image processing device 13 .
- the imaging device 11 and the display device 12 are installed near the assembly line of the assembly plant, and the image processing device 13 is installed in a central control room away from the assembly line or in a remote server away from the assembly plant. may
- the imaging device 11 is a device for acquiring image data of objects existing around the operator, and is, for example, a camera including an imaging device such as a CCD, an ultrasonic camera, an infrared camera, or the like.
- Objects include objects existing around the operator in addition to the operator.
- objects include switches and touch panels around the vehicle occupants, parts assembled by workers and tools used, and vehicles maintained by dealer mechanics.
- the imaging device 11 can detect the operator's body posture such as the dashboard, roof and seat of the vehicle, the assembly line of the assembly plant, the vicinity of the work table and the tools used by the workers, and the lift of the dealer. installed in a position where
- the imaging device 11 of the present embodiment is attached to at least one of the operator who performs the above-described predetermined operation and the article worn by the operator. At this time, the imaging device 11 is attached so that at least a part of the operator's body can be photographed.
- the imaging device 11 is attached to the operator, for example, the imaging device 11 is attached to the operator's head, upper arm, forearm, abdomen, thigh and lower leg.
- a mounting method for example, a hook-and-loop fastener or a buckle is used to wrap and fix the operator's body, an adhesive tape is used to affix the operator's body, and an adhesive is used for operation. and a method of adhering to a person's body.
- FIG. 2A is a plan view showing the operator's left forearm 73b and left hand 73c, with the left forearm 73b covered by clothing.
- the wearable terminal 3 is worn by wrapping it around a portion of the operator's left forearm 73b near the wrist with a band 31 having a hook-and-loop fastener.
- the wearable terminal 3 includes an imaging device 11, and the imaging device 11 captures an image within a field of view 4 indicated by broken lines in FIG. 2A. Accordingly, an image including the body of the operator wearing the wearable terminal 3 can be captured.
- the items worn by the operator are the clothes and protective gear that the operator wears when performing a predetermined operation.
- Clothing and protective gear includes jackets, coats, wristwatches, rings, earrings and piercings worn by vehicle occupants, and work clothes, caps, safety shoes and helmets worn by assembly plant workers and dealer mechanics. , masks and protective glasses.
- a method of attaching the imaging device 11 to the clothes and protective gear worn by the operator for example, a method of fixing it to a coat or work clothes using a hook-and-loop fastener, and a member such as a clamp to a hat, a helmet, a mask, protective glasses, or the like. and attaching the imaging device 11 by gripping the member, and a method of adhering the device to a wrist watch or safety shoes using an adhesive tape or adhesive.
- FIG. 2B is a perspective view showing the operator's head 71 and the helmet 5 attached to the head 71.
- the camera 6 is attached downward to the brim 51 of the helmet 5 worn by the operator so that the operator's body can be photographed.
- the camera 6 is attached by attaching a bracket 52 for holding the camera 6 to the helmet 5 and holding the camera 6 by the bracket 52 .
- the helmet 5 may be provided with a portion for sandwiching the camera 6, and the portion may be integrally molded as the helmet 5 to sandwich the camera 6 therebetween.
- the camera 6 captures an image within the range of the field of view 4a indicated by the dashed line in FIG. 2B. Accordingly, an image including the body of the operator wearing the helmet 5 can be captured.
- FIG. 2C is a plan view showing the operator's right arm, which consists of the right upper arm 74a, the right forearm 74b, and the right hand 74c.
- the camera 6 is attached by wrapping it around the operator's right forearm 74b near the elbow with a band 61 having a hook-and-loop fastener.
- the camera 6 captures an image within the field of view 4b indicated by the dashed line in FIG. 2C. Accordingly, an image including the right hand 74c of the operator wearing the camera 6 can be captured.
- FIG. 2C is a plan view showing the operator's right arm, which consists of the right upper arm 74a, the right forearm 74b, and the right hand 74c.
- the camera 6 is attached by wrapping it around the operator's right forearm 74b near the elbow with a band 61 having a hook-and-loop fastener.
- the camera 6 captures an image within the field of view 4b indicated by the das
- the camera 6 is mounted so as to photograph the back side of the operator's right hand 74c, but the camera 6 may be mounted so as to photograph the palm side of the operator's right hand 74c. Which side is photographed is appropriately selected according to the behavior of the operator to be detected.
- the wearable terminal 3 is worn on the left arm in the example of FIG. 2A and the camera 6 is worn on the right arm in the example of FIG. 2C, the wearable terminal 3 and the camera 6 may be worn on either the left or right arm. .
- these mounting positions can be set to appropriate positions within a range in which the behavior of the operator can be appropriately detected.
- the display device 12 is a device for notifying the operator of actions detected by the image processing device 13 .
- the display device 12 is, for example, a liquid crystal display, a projector, or the like, and may include a speaker.
- the display device 12 is installed near the operator, such as the dashboard of the vehicle, the work place of the worker in the assembly plant, and the position where the operator can be notified of necessary information. If there is a supervisor to monitor the operation, it will be installed near the supervisor. In this case, if the supervisor is located away from the operator, the display device 12 will be installed at a location away from the operator. Also, the display device 12 may be attached to the operator as a wearable terminal such as glasses. Furthermore, instead of the display device 12, only a speaker that emits an alarm sound according to the detected action may be used.
- the image processing device 13 is a device, such as a computer, for estimating the behavior of the operator from the posture of the operator performing a predetermined operation.
- the image processing device 13 can consider the positional deviation of the imaging device 11 when detecting the posture of the operator.
- the image processing device 13 cooperates with the imaging device 11 and the display device 12 to obtain image data from the imaging device 11 at predetermined time intervals, and processes the obtained image data to detect the posture of the operator. Then, correction of the positional deviation of the imaging device 11 and estimation of the behavior of the operator are performed.
- the image processing device 13 includes a CPU (Central Processing Unit) 131 which is a processor, and a program stored therein. It has a ROM (Read Only Memory) 132 and a RAM (Random Access Memory) 133 that functions as an accessible storage device.
- the CPU 131 is an operation circuit for functioning as the image processing device 13 by executing programs stored in the ROM 132 . Note that the image processing device 13 does not have to be installed together with the imaging device 11 and the display device 12, and may be installed in a remote server away from these devices.
- a program used in the image processing device 13 includes a processing section 2 which is a functional block for realizing the functions of the image processing device 13 described above.
- the processing unit 2 controls and cooperates with the imaging device 11 and the display device 12 to obtain image data including the operator from the imaging device 11 (data acquisition function) and acquire the acquired image data.
- a function for detecting the posture of the operator after processing and correcting the positional deviation of the imaging device 11, and a function for estimating the action of the operator from the posture of the operator (action estimation function). function
- the processing unit 2 includes, as shown in FIG. and In FIG. 1, each part is extracted and shown for convenience.
- the image processing apparatus 13 shown in FIG. 1 includes all of the above functional blocks, but it is not necessary for a single image processing apparatus 13 to include all of the functional blocks. It may be provided in another device included in the system 1 or in another information processing device (not shown).
- the detection unit 22 may be provided in the imaging device 11 . In this case, the functions of the detection unit 22 are executed using the CPU, ROM, and RAM of the imaging device 11 .
- each functional block it is not necessary for a single device to execute all the processing of each functional block, and the function of each functional block may be realized across multiple devices connected in a state where data can be exchanged.
- the CPU, ROM, and RAM of the imaging device 11 are used to perform part of the processing for realizing the function of the calculator 23 .
- the acquisition unit 21 has a function of acquiring an image including the operator from the imaging device 11 .
- the imaging device 11 of the present embodiment is installed at a position and in an orientation in which an image including the operator's body can be captured. By acquiring image data at predetermined time intervals, an image including the operator can be acquired. Further, the image data acquired by the acquisition unit 21 is not limited to still images, and may be time-series images such as moving images. Examples of images acquired by the acquisition unit 21 are shown in FIGS. 3A to 3C.
- FIG. 3A is an image captured by the wearable terminal 3 shown in FIG. 2A, and is an image including the operator 7 who is a passenger of the vehicle.
- the operator 7 who is a vehicle occupant wears the wearable terminal 3 on the forearm 73b of the left arm.
- a forearm 73b is shown on the right side, and a right upper arm 74a and a right forearm 74b are shown on the left side.
- the entire head 71 and neck 72 of the operator 7 are photographed, and a part of the body 75 is also photographed.
- FIG. 3B is an image captured by the camera 6 shown in FIG. 2B, and is an image including the operator 7 who is a worker at the vehicle assembly plant.
- the operator 7 who is a worker has the camera 6 attached to the brim 51 of the helmet 5 worn on the head 71 facing downward.
- a portion above 51 that is, a portion covered by the helmet 5) cannot be photographed. Therefore, as shown in FIG. 3B, an image of the upper portion of the head 71 cannot be acquired.
- a neck 72 and a torso 75 are shown below the head 71. On the left side of the torso 75 are shown a right upper arm 74a, a right forearm 74b, and a right hand 74c.
- An upper arm 73a, a left forearm 73b, and a left hand 73c are shown. Since the camera 6 photographs the operator 7 from the front direction of the operator 7, the left and right are reversed as in FIG. 3A. A portion of the left leg 76 and a portion of the right leg 77 are also photographed below the body 75 .
- FIG. 3C is an image captured by the camera 6 shown in FIG. 2C, and is an image including the right hand 74c of the operator who is a worker at the vehicle assembly plant. Since the operator who is a worker wears the camera 6 near the elbow of the forearm 74b of the right arm, when the right hand 74c is photographed from the camera 6, as shown in FIG. , and the right hand 74c.
- the detection unit 22 has a function of detecting the posture of the operator 7 from the image acquired by the acquisition unit 21.
- the posture of the operator 7 is the stance or posture of the body of the operator 7, and more specifically, the position of each joint of the body of the operator 7 and the relationship of connection between the joints.
- the positions of the joints of the body of the operator 7 and the parts of the body that connect the joints are known, the positions of the joints in a three-dimensional space defined by an orthogonal coordinate system.
- the operator's body stance can be represented geometrically. .
- the detection unit 22 performs the processing shown in FIG. 4A to detect the posture of the operator 7.
- the joints of the operator 7 are detected from among the operator 7 included in the image by pattern matching or the like.
- a characteristic portion such as a portion where the orientation of the body of the operator 7 is changed is extracted, and the extracted characteristic portion is compared with the class dictionary stored in the database 14. By doing so, it is determined to which class the feature part belongs. If a class for each joint is set in the class dictionary, it can be determined whether or not the characteristic portion corresponds to the joint.
- a point P1 is plotted at a portion corresponding to the joint of the neck 72 of the operator 7
- a point P2 is plotted at a portion corresponding to the right shoulder joint
- a right elbow joint is plotted.
- a point P3 is plotted at the portion corresponding to
- a point P4 is plotted at the portion corresponding to the left shoulder joint
- a point P5 is plotted at the portion corresponding to the left elbow joint.
- a straight line corresponding to the right shoulder is drawn between points P1 and P2
- a straight line corresponding to the right upper arm 74a is drawn between points P2 and P3
- a straight line is drawn between points P1 and P4.
- a straight line corresponding to the left shoulder is drawn between them, and a straight line corresponding to the upper arm 73a of the left arm is drawn between points P4 and P5.
- the image shown in FIG. 3A does not include the right and left hands of the operator 7, but since there is no joint between the elbow joint and the wrist, the forearm 74b of the right arm included in FIG. It is presumed that the right hand exists on the line and the left hand exists on the extension line of the left forearm 73b. Therefore, as shown in FIG. 4A, a straight line is drawn from point P3 toward the estimated position of the right hand (not shown), and a straight line is drawn from point P5 toward the estimated position of the left hand (not shown).
- the detection unit 22 performs the processing shown in FIG. 4B to detect the posture of the operator 7.
- joints are first detected by a process similar to the pattern matching described in FIG. 4A. Then, as in FIG. 4A, dots are plotted at the portions determined to be joints. Specifically, as shown in FIG. 4B, a point P6 is plotted at the portion corresponding to the joint of the neck 72 of the operator 7, a point P7 is plotted at the portion corresponding to the right shoulder joint, and a right elbow joint is plotted.
- Point P8 is plotted on the part corresponding to the right wrist joint
- Point P9 is plotted on the part corresponding to the right wrist joint
- Point P10 is plotted on the part corresponding to the left shoulder joint
- Point P10 is plotted on the part corresponding to the left elbow joint P11 is plotted
- point P12 is plotted at the portion corresponding to the joint of the left wrist.
- a point P13 is plotted at the portion corresponding to the hip joint
- a point P14 is plotted at the portion corresponding to the joint at the base of the right leg
- a point P15 is plotted at the portion corresponding to the right knee joint
- a point P15 is plotted at the portion corresponding to the right knee joint.
- a point P16 is plotted at the portion corresponding to the joint at the base of the knee
- a point P17 is plotted at the portion corresponding to the left knee joint.
- a straight line corresponding to the trunk 75 is drawn between the points P6 and P13, and a straight line corresponding to the base of the leg is drawn between the points P13 and P14 and between the points P13 and P16.
- a straight line corresponding to the right leg 77 is drawn between the points P14 and P15, and a straight line corresponding to the left leg 76 is drawn between the points P16 and P17.
- the image shown in FIG. 3B does not include the right ankle and left ankle of the operator 7, but since there is no joint between the knee joint and the ankle, the right leg 77 included in FIG. It is presumed that the right ankle exists on the extension line and the left ankle exists on the extension line of the left leg 76 . Therefore, as shown in FIG. 4B, straight lines are drawn from points P15 and P17 toward the estimated positions of both ankles (not shown). These straight lines pass through the centers of right leg 77 and left leg 76, for example, in a direction perpendicular to the lengthwise direction of right leg 77 and left leg 76. As shown in FIG. In this way, the detection unit 22 obtains the positions of the joints of the body of the operator 7 included in FIG. The posture of the operator 7 is detected by representing it with a straight line.
- the detection unit 22 performs the processing shown in FIG. 4C to detect the posture of the operator's right hand 74c.
- joints are first detected by a process similar to the pattern matching described in FIG. 4A. Then, as in FIG. 4A, dots are plotted at the portions determined to be joints. Specifically, as shown in FIG. 4C, a point P18 is plotted at the joint of the right wrist of the operator, points P19 to P20 are plotted at the joint of the thumb, and points P19 to P20 are plotted at the joint of the index finger.
- Points P21 to P23 are plotted in the corresponding portion
- points P24 to P26 are plotted in the portion corresponding to the joint of the middle finger
- points P27 to P29 are plotted in the portion corresponding to the joint of the ring finger
- the joint of the little finger is plotted.
- Plot points P30-P32 on the part.
- a straight line corresponding to the thumb is drawn between points P19 and P20
- a straight line corresponding to the index finger is drawn between points P21 and P22 and between points P22 and P23
- points P24 and P24 are drawn.
- a straight line corresponding to the middle finger is drawn between points P25 and between points P25 and P26
- a straight line corresponding to the ring finger is drawn between points P27 and P28 and between points P28 and P29
- a straight line corresponding to the little finger is drawn between points P30 and P31 and between points P31 and P32.
- FIG. 3C The image shown in FIG. 3C does not include the operator's right elbow, but since there is no joint between the elbow and the wrist, the right elbow is on the extension line of the right forearm 74b included in FIG. 4C. presumed to exist. Therefore, as shown in FIG. 4C, a straight line is drawn from point P18 toward the estimated position of the right elbow (not shown). This straight line passes through the center of the right forearm 74b, for example, in a direction perpendicular to the length of the right forearm 74b. In this way, the detection unit 22 determines the position of each joint of the operator's right hand 74c included in FIG. By representing with a straight line, the posture of the operator's right hand 74c is detected.
- the detection unit 22 can detect the posture of the operator 7 based on the relative positional relationship between the parts of the operator's 7 body and objects around the operator 7 . For example, the positions of the head 71 and arms of the operator 7 are calculated with respect to the ground on which the operator 7 is standing, and the posture of the operator 7 is detected based on the calculation result. Alternatively, or in addition to this, the detection unit 22 can detect the posture from the orientation of the parts of the body of the operator 7 . For example, the posture of the operator 7 is detected from the eye position (line of sight), hand orientation, leg orientation, and body orientation 75 of the operator 7 .
- joints for which points are set by the detection unit 22 are not limited to the joints shown in FIGS. 4A to 4C.
- finer joints may be detected and plotted with more points than those shown in Figures 4A and 4B.
- dots may be plotted at portions corresponding to the finger joints of the left hand 73c and right hand 74c to detect how the fingers of the left hand 73c and right hand 74c are shaped. . This makes it possible to detect detailed postures, such as whether or not a vehicle occupant is touching onboard equipment of the vehicle, or whether or not an assembly line worker is holding a tool.
- the detection unit 22 uses a posture estimation model trained in advance to estimate the posture of the operator 7 from an image including the operator 7. may be used.
- the posture estimation model is stored, for example, in database 14 shown in FIG.
- an image including the operator 7 is associated with the pose of the operator 7 detected in the image.
- the posture of the operator 7 is output.
- the posture estimation model is machine-learned by the first machine-learning unit 221 shown in FIG. Specifically, using the past images and the posture detection result of the operator 7 stored in the database 14, the posture estimation model detects an appropriate posture for the input image data, and the result Let it learn to output
- the posture estimation model may be a trained model that has undergone machine learning.
- a trained model is a model that has been learned in advance by machine learning so that appropriate output data can be obtained for certain input data. At least, a program that performs operations from input data to output data, and a weighting coefficient (parameter) used for the calculation.
- the learned model is configured by a computer (especially , a processor (CPU 131). By using such a learned model, it is possible to detect the posture of the operator 7 who performs an operation other than the learned operation.
- a neural network comprises an input layer, an intermediate layer, and an output layer, each layer containing at least one neuron.
- Input data including image data acquired by the acquisition unit 21 is input to the input layer, and the input data is output to the intermediate layer.
- the intermediate layer extracts the data of the operator 7 from the data input from the input layer.
- the posture is detected from the extracted data of the operator 7 .
- the output layer outputs the data input from the intermediate layer as output data including posture data.
- parameters for the intermediate layer the positions of the joints of the operator 7 and the connection relationships between the joints are considered.
- the calculation unit 23 has a function of identifying a reference position as a reference for calculating positional displacement from the image obtained by the obtaining unit 21 and calculating the positional displacement of the imaging device 11 using the identified reference position.
- the imaging device 11 of the present embodiment is attached to at least one of the operator 7 and an article worn by the operator 7, but the attached imaging device 11 is not necessarily completely fixed. While the operator 7 is performing a predetermined operation, the mounting position of the imaging device 11 may move from the position where it was initially mounted. For example, in FIG. 2A, when the band 31 rotates with respect to the wrist and the wearable terminal 3 rotates and moves toward the thumb side of the left hand 73c, the acquisition unit 21 detects the operator's position as shown in FIG. 5A.
- An image is acquired with 7 rotated counterclockwise (in the direction of arrow 8). 2C, when the band 61 is loosened and the camera 6 rotates clockwise, the acquisition unit 21 obtains an image in which the right hand 74c rotates clockwise (in the direction of the arrow 9) as shown in FIG. 5B. is obtained.
- the posture of the operator 7 may not be detected correctly by the detection unit 22. Therefore, the function of the calculation unit 23 is used to calculate the positional deviation of the mounting position of the imaging device 11, and the correction unit 24 described later. corrects the posture detected by the detection unit 22 by . Thereby, the posture of the operator 7 is correctly detected.
- the positional deviation of the imaging device 11 is calculated with respect to a reference position in a predetermined state.
- the predetermined state includes, for example, the initial state when the imaging device 11 is worn (hereinafter also referred to as "initial state"), the state when the operator 7 starts a predetermined operation, and a predetermined time from a certain time point (for example, the present). A previous state or state at a certain time. How much the reference position has moved (displaced) with respect to the reference position in these states is calculated, and the calculated result is used as the positional displacement of the imaging device 11 .
- the calculation unit 23 identifies a part of the body of the operator 7 included in the image as a reference position, and calculates the positional displacement using the identified part of the body.
- the parts of the body mentioned above are parts of the body that are identified in order to calculate the positional deviation as simply as possible, and the parts themselves do not have movable parts such as joints. Specifically, when the imaging device 11 is worn by the operator 7 , there is no joint between the imaging device 11 and the part where the imaging device 11 is worn. When attached to an article worn by the operator 7, it is a part where no joint exists between the article that the operator 7 is wearing and the part that is in contact with the article. The reason why there is no portion movable by a joint in the part itself is that when calculating the positional deviation using the part of the body as the reference position, the part at the reference position may have moved due to movement of the joint, or the mounting position of the imaging device 11 may be changed.
- the portion of the body of the operator 7 to which the imaging device 11 is mounted cannot be calculated.
- the portion that is in contact with the article worn by the operator 7 is excluded from the candidates for the reference position.
- the part on which the imaging device 11 is mounted and the part in contact with the article worn by the operator 7 are parts of the body of the operator 7 .
- the part identified as the reference position and the part to which the imaging device 11 is attached may be the same part.
- the part identified as the reference position part and the part in contact with the article worn by the operator 7 may be the same part.
- a preset part of the body of the operator 7 included in the image data is detected by pattern matching. Then, when a preset site is identified, the site is set as a reference position for calculating the positional deviation.
- the site is set as a reference position for calculating the positional deviation.
- FIG. 2A when the wearable terminal 3 is worn on the wrist of the left hand 73c, the left forearm 73b existing between the wrist and the left elbow joint is used as a reference position. Set in advance. The forearm 73b itself of the left arm does not have a movable part by a joint, and there is no joint between the worn wearable terminal 3 and the forearm 73b of the left arm.
- the left hand 73c since the left hand 73c has finger joints, it does not serve as a reference position for calculating the positional deviation.
- FIG. 2B when the helmet 5 is worn on the head 71, the nose not covered by the helmet 5 is set in advance as a reference position. The nose itself has no articulated parts and there are no joints between the part covered by the helmet 5 and the nose.
- the jaw and mouth do not serve as the reference position for calculating the positional deviation because the mandibular joint moves the lower jaw.
- FIG. 2B when the helmet 5 is worn on the head 71, the nose not covered by the helmet 5 is set in advance as a reference position. The nose itself has no articulated parts and there are no joints between the part covered by the helmet 5 and the nose.
- the jaw and mouth do not serve as the reference position for calculating the positional deviation because the mandibular joint moves the lower jaw.
- the forearm 74b of the right arm when the camera 6 is attached to a portion of the forearm 74b of the right arm near the elbow, the forearm 74b of the right arm is set in advance as a reference position. .
- the forearm 74b of the right arm itself does not have a movable part by a joint, and there is no joint between the mounted camera 6 and the forearm 74b of the right arm (especially the wrist side of the forearm 74b of the right arm).
- the right hand 74c since the right hand 74c is moved by the joint of the right wrist, it does not become the reference position for calculating the positional deviation.
- FIGS. 6A and 6B show a method of calculating positional deviation when the nose is identified as a reference position for calculating positional deviation and the identified nose is set as the reference position.
- FIG. 6A is a plan view showing an example of an image obtained when the head 71 is photographed by the camera 6 shown in FIG. 2B.
- FIG. 6A includes a head 71 and a nose 711, and the nose 711 is set as a reference position for calculating positional deviation.
- the positional deviation in the case of FIG. 6A is calculated with respect to the position when the imaging device 11 is first worn, that is, the position in the initial state shown in FIG. 6A. In this case, assuming that the helmet 5 rotates counterclockwise, as shown in FIG.
- the head 71 is rotated clockwise by an angle ⁇ with respect to the center 712 of the head 71 of .
- the position of the nose 711 in the image does not change even if the operator 7 rotates. Actually, the operator 7 himself did not rotate, but only the helmet 5 rotated.
- the calculation unit 23 calculates the position of the nose in the initial state. The distance and angle (angle ⁇ in the case of FIG.
- the nose 6B by which the nose has moved with respect to the position of 711 (that is, the position of the center 712) are detected.
- the positional deviation caused by the rotation of the helmet 5 (camera 6) is calculated.
- the calculated positional deviation information is output to the correction unit 24 .
- the mounting position of the imaging device 11 in the initial state or the position of the nose 711 is also referred to as a starting position for calculating the positional deviation.
- the part to be the reference position may be identified from the time-series images acquired by the acquisition unit 21 without presetting the part to be the reference position. For example, in time-series images captured at a predetermined cycle, a part that has moved only a distance within a predetermined range (for example, 5 mm or less) in which displacement of the mounting position of the imaging device 11 can be detected is detected. , identifies the site as the reference position. For example, when the imaging device 11 is attached to the ankle of the right leg facing upward, a portion whose movement amount in the image is 5 mm or less is detected in time-series images taken at a cycle of 1 minute. In this case, since the operator 7 is photographed upward from the ankle of the right leg, the calf of the right leg hardly moves with respect to the ankle. Therefore, the calf of the right leg is identified as the reference position.
- a predetermined range for example, 5 mm or less
- the calculation unit 23 calculates the displacement from the amount of movement of the part in the time-series images.
- the movement of the nose 711 shown in FIGS. 6A and 6B is acquired as time-series images from when the helmet 5 is worn until the helmet 5 rotates as shown in FIG. It is possible to calculate in which direction and how much the hollow nose 711 has moved, and to use the calculation result as a positional deviation.
- the starting position for calculating the positional deviation may be set to the first position of the part in the time-series images.
- the calculating unit 23 uses positional displacement estimation that has been learned in advance so as to estimate the positional displacement from the position of the part set as the reference position in the image when calculating the positional displacement from the image acquired by the acquiring unit 21.
- a model may be used.
- the positional deviation estimation model is stored, for example, in the database 14 shown in FIG.
- an image including the operator 7 is associated with the misregistration calculated in the image. output.
- the positional deviation estimation model is machine-learned by the second machine-learning unit 231 shown in FIG. Specifically, using past images and misregistration calculation results stored in the database 14, a misregistration estimation model estimates appropriate misregistration for the input image data, and outputs the results. Let it learn to output.
- the misalignment estimation model may be a learned model that has undergone machine learning.
- the trained model is a computer (particularly, a processor) so that output data including positional deviation is output based on the input data.
- a certain CPU 131) is activated.
- a neural network comprises an input layer, an intermediate layer, and an output layer, each layer containing at least one neuron.
- Input data including image data acquired by the acquisition unit 21 is input to the input layer, and the input data is output to the intermediate layer.
- the intermediate layer extracts the data of the operator 7 from the data input from the input layer.
- the positional deviation is estimated from the extracted data of the operator 7 .
- the output layer outputs the data input from the intermediate layer as output data including misalignment data.
- the positions of the joints of the operator 7 and the connection relationships between the joints are considered.
- the calculation unit 23 identifies the pattern as a reference position, and identifies the imaging device specified from the pattern. 11 positions may be used to calculate the misalignment.
- the pattern is not particularly limited as long as it serves as a reference when calculating the positional deviation in the image, and examples thereof include a plurality of intersecting lines, star-shaped marks, and the like.
- a mask worn by a vehicle occupant is patterned with grid lines, and the grid lines are used to calculate the displacement of the mounting position of the imaging device 11 .
- the calculation unit 23 may be calculated based on the position of the imaging device 11 and the posture of the operator 7 detected at 22 .
- a camera installed in front of an assembly plant worker is used to detect the position of the imaging device 11 and the posture of the operator 7, and the behavior of the operator 7 is estimated using the detection results.
- the correction unit 24 has a function of correcting the orientation detected by the detection unit 22 using the positional deviation of the imaging device 11 calculated by the calculation unit 23 .
- the correction unit 24 geometrically transforms the geometrically expressed posture of the operator 7 acquired by the detection unit 22. I do. For example, as shown in FIG. 6B, when the helmet 5 is rotated counterclockwise by an angle ⁇ , the detected orientation is converted to rotate clockwise by an angle ⁇ . As a result, the posture of the operator 7 before the helmet 5 rotates and the wearing position shifts, that is, the correct posture of the operator 7 can be detected.
- the helmet 5 when the helmet 5 is displaced in the front-rear direction and in the left-right direction with respect to the operator 7, at least one linear transformation out of translation, rotation, enlargement, and reduction is performed on the detected posture, and the helmet 5 is shifted.
- a process for correcting movement that is, positional deviation
- the wearable terminal 3 shown in FIG. 2A and the camera 6 shown in FIG. 2C can also be corrected for positional deviation by performing the same geometric conversion.
- the estimation unit 25 has a function of estimating the behavior of the operator 7 from the posture of the operator 7 corrected by the correction unit 24 .
- the action of the operator 7 refers to any action necessary to complete a predetermined operation, such as the act of reaching out to the switch of the on-vehicle equipment such as an air conditioner by the occupant of the vehicle to operate the on-vehicle equipment, assembly, etc.
- a factory worker taking out a torque wrench from a tool box and the action of a dealer mechanic touching a switch for raising and lowering a lift that lifts a vehicle
- a vehicle occupant presses a switch for raising and lowering the window glass of the vehicle.
- the action of a vehicle occupant touching a touch panel to change the display of a map on a navigation device The assembly factory worker's action of connecting a coupler connected to a sensor and a coupler connected to an electronic control unit (ECU). assembly plant workers tightening bolts using tools to attach the exhaust manifold to the engine block, dealer mechanics fitting spark plugs into the engine, dealer mechanics , the action of tightening a bolt using a torque wrench.
- ECU electronice control unit
- the estimation unit 25 determines that the vehicle occupant is a vehicle occupant based on the positions of points P1 to P3 shown in FIG. 4A and the shape of a straight line drawn from the points P1 to P3. It is presumed that the operator 7 is extending his right hand (not shown) to the operation unit of the on-vehicle device in order to operate the on-board device. Further, when the posture shown in FIG. 4B is detected, from the positions of points P7 to P9 and P10 to P12 shown in FIG.
- characteristic portions such as the positions of the joints of the operator 7, the relationship between the joints, and the shapes of the body parts that connect the joints are extracted and extracted. It is determined to which class the characteristic portion belongs by comparing the obtained characteristic portion with the class dictionary stored in the database 14 . If an action corresponding to a posture is set as a class in the class dictionary, the action can be estimated from the posture.
- a model can be used.
- the action estimation model is stored, for example, in the database 14 shown in FIG.
- the detected posture of the operator 7 and the action of the operator 7 are associated so that they correspond to each other. 7 actions are output.
- the behavior estimation model is machine-learned by the third machine-learning unit 251 shown in FIG. Specifically, using the past detection results of the posture of the operator 7 and the estimated behavior of the operator 7 stored in the database 14, the behavior estimation model generates learn to estimate appropriate actions and output the results.
- the action estimation model may be a learned model that has undergone machine learning.
- the trained model is configured by a computer (especially , a processor (CPU 131). By using such a learned model, it is possible to appropriately detect the actions of the operator 7 who performs an operation other than the learned operation. It should be noted that the input posture data may or may not have been corrected by the correction unit 24 .
- a neural network comprises an input layer, an intermediate layer, and an output layer, each layer containing at least one neuron.
- Input data including posture data detected by the detection unit 22 is input to the input layer, and the input data is output to the intermediate layer.
- the intermediate layer extracts posture data of the operator 7 from the data input from the input layer.
- the action is estimated from the extracted posture data of the operator 7 .
- the output layer outputs the data input from the intermediate layer as output data including action data.
- the parameters in the intermediate layer for example, the positions of points indicating the joints of the operator 7 and the shape of straight lines indicating the connection relationship between the joints are considered.
- the estimating unit 25 does not specify the attributes of the operator 7.
- an action corresponding to the posture can be specified.
- the attributes of the operator 7 include, for example, whether the operator 7 is a vehicle crew member, a worker working on an assembly line at an assembly plant, or a mechanic at a vehicle dealer maintenance plant. It is a feature about the role of the operator 7 and the place where the operator 7 exists.
- the output unit 26 has a function of outputting the behavior estimated by the estimation unit 25 to the display device 12 .
- the actions of the operator 7 output from the output unit 26 are received by the display device 12 . Then, it is displayed on the display device 12 and presented to the operator 7, the supervisor of the operator 7, and the like. By confirming the action presented by the display device 12, the operator 7 and the supervisor can determine whether the action of the operator 7 is necessary for the predetermined operation and whether the operator 7 performs the predetermined operation in a predetermined procedure. It is possible to confirm whether or not the operator 7 has skipped necessary procedures.
- the processing unit 2 determines whether the operator 7's action is necessary for the predetermined operation, whether the operator 7 is performing the predetermined operation in a predetermined procedure, and whether the operator 7 performs the predetermined operation. It may be determined whether or not 7 is skipping a necessary step.
- the predetermined operation, the procedure of the predetermined operation, the action required for the predetermined operation, and the like are stored in the database 14 and acquired by the processing unit 2 as necessary.
- FIG. 7 is an example of a flowchart showing information processing in the image processing system 1 of this embodiment. The processing described below is executed at predetermined time intervals by the CPU 131, which is the processor of the image processing device 13, while the operator 7 is performing a predetermined operation.
- step S1 the function of the acquisition unit 21 acquires image data captured by the imaging device 11 .
- step S2 it is determined whether or not image data including the operator 7 has been acquired. If the operator 7 is not included in the obtained image data, the process returns to step S1. On the other hand, when the operator 7 is included in the obtained image data, the process proceeds to step S3.
- step S3 the posture of the operator 7 included in the image is detected by the function of the detection unit 22.
- step S4 it is determined whether or not the posture of the operator 7 has been properly detected. If the posture of the operator 7 could not be properly detected, the process returns to step S1. On the other hand, if the posture of the operator 7 can be properly detected, the process proceeds to step S5.
- step S5 the function of the calculation unit 23 calculates how much the mounting position of the imaging device 11 is deviated from, for example, the reference position in the initial state.
- step S6 it is determined whether or not the mounting position of the imaging device 11 is shifted. Specifically, it is determined whether or not the calculated amount of positional deviation is within a range in which the posture of the operator 7 can be detected correctly. If the calculated amount of positional deviation is within the range in which the posture of the operator 7 can be detected correctly, the process proceeds to step S8. On the other hand, if the calculated amount of positional deviation exceeds the range in which the posture of the operator 7 can be detected correctly, the process proceeds to step S7.
- step S7 the posture of the operator 7 is corrected by the function of the correction unit 24 based on the amount of deviation calculated by the calculation unit 23.
- step S8 the action of the operator 7 is estimated from the posture of the operator 7 by the function of the estimation unit 25.
- step S9 it is determined whether or not the behavior of the operator 7 has been appropriately estimated. If the behavior of the operator 7 could not be properly estimated, the execution of the routine is stopped and the information processing is terminated. On the other hand, if the action of the operator 7 can be appropriately estimated, the process proceeds to step S10.
- step S10 the function of the output unit 26 outputs data including the estimated behavior from the image processing device 13 to the display device 12.
- the function of the processing unit 2 determines whether or not the operator 7 is appropriately performing the predetermined operation based on the behavior estimated by the estimation unit 25. FIG. At this time, the determination result may be output to the display device 12 .
- step S11 the execution of the routine is stopped and the information processing is terminated. Note that step S2, step S4, step S6, and steps S9 to S11 are not essential steps, and can be provided as necessary.
- an image including the operator 7 is captured from the imaging device 11 attached to at least one of the operator 7 performing the predetermined operation and the article worn by the operator 7.
- An acquisition unit 21 for acquiring, a detection unit 22 for detecting the posture of the operator 7 from the image, and a positional deviation of the imaging device 11 is calculated from the image using a reference position as a standard for calculating positional deviation. and a correction unit 24 that corrects the posture using the positional deviation.
- an action estimation model that has been learned in advance so as to estimate the action of the operator 7 from the posture is used to estimate the action of the operator 7 from the corrected posture.
- An estimation unit 25 for estimating behavior is provided. Thereby, the behavior of the operator 7 can be estimated by associating the posture and the behavior of the operator 7 .
- the detection unit 22 uses a posture estimation model that has been learned in advance so as to estimate the posture of the operator 7 from the image, and performs the operation from the image.
- the posture of the person 7 is detected. As a result, it is possible to perform highly accurate posture detection using the detection results obtained so far.
- the calculation unit 23 identifies a part of the body of the operator 7 included in the image as the reference position, and calculates the reference position relative to the reference position in a predetermined state. Calculate the misalignment. As a result, it is possible to easily and accurately calculate the positional deviation using the parts of the body.
- the reference position is the operator's position other than the part where the imaging device 11 is attached. 7 of the body, the part itself does not have a joint that can be moved by a joint and has no joint with the part to which the imaging device 11 is attached.
- the operator 7 is a part where there is no joint between the article and the part that is in contact with the wearer. This makes it possible to more accurately identify body parts.
- the acquisition unit 21 acquires time-series images including the reference position, and the calculation unit 23 moves the reference position in the time-series images. Quantity is used to calculate the misalignment. Thereby, the positional deviation of the imaging device 11 can be calculated more accurately.
- the calculation unit 23 uses a positional deviation estimation model learned in advance so as to estimate the positional deviation from the position of the reference position in the image. Calculate the deviation. This makes it possible to calculate the amount of positional deviation simply by inputting an image.
- the acquisition unit 21 acquires an image including a pattern for specifying the position of the imaging device 11, and the calculation unit 23 calculates the pattern as the reference.
- the positional deviation is calculated using the position of the imaging device 11 identified as the position and specified from the pattern. Thereby, the positional deviation of the imaging device 11 can be calculated more accurately.
- the detection unit 22 detects the position of the imaging device 11 and the posture of the operator 7 using detection devices installed around the operator 7 .
- the calculation unit 23 calculates the positional deviation based on the position of the imaging device 11 and the posture of the operator 7 detected by the detection unit 22 . Thereby, the positional deviation of the imaging device 11 can be calculated more accurately.
- the processor performs an image pickup attached to at least one of the operator 7 performing the predetermined operation and the article worn by the operator 7.
- An image including the operator 7 is acquired from the device 11, the posture of the operator 7 is detected from the image, and the image of the imaging device 11 is detected using a reference position as a reference for calculating positional deviation from the image.
- An image processing method is provided that calculates misalignment and corrects the pose using the misalignment. As a result, the posture of the operator 7 can be correctly recognized even when the mounting position of the imaging device 11 (for example, a camera) is displaced.
- Output unit 3 Wearable terminal 31 Bands 4, 4a, 4b Field of view 5 Helmet 51 Collar 52 Bracket 6 Camera 61 Band 7 Operator (human) 71 Head 711 Nose 712 Center 72 Neck 73a Left upper arm 73b Left forearm 73c Left hand 74a Right upper arm 74b Right forearm 74c Right hand 75 Torso 76 Left leg 77 Right leg 8, 9... Arrows P1 to P32... Point ⁇ ... Angle
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
Description
図1は、本発明に係る画像処理システム1を示すブロック図である。画像処理システム1は、所定操作を行う操作者(本発明に係る人間に相当する。)の行動を検出する装置である。所定操作は特に限定されず、たとえば、車両の乗員(ドライバーを含む。以下同じ。)による車載機器の操作と、工場において組立て作業に従事する作業員の工具の操作と、車両の販売店(以下、「ディーラー」とも言う。)において、車両の整備に従事する整備士の整備用具の操作とが挙げられる。画像処理システム1により行動を検出される操作者(以下、単に「操作者」とも言う。)は特に限定されず、車両の乗員、工場の作業員及びディーラーの整備士などが挙げられる。画像処理システム1により検出される行動は、たとえば、車両の乗員がエアコンなどの車載機器を操作するために、車載機器のスイッチに手を延ばす行動と、組立て工場の作業員が工具箱からトルクレンチを取り出す行動と、ディーラーの整備士が車両を持ち上げるリフトを上下させるスイッチに触れる行動とが挙げられる。本実施形態の画像処理システム1により上述したような行動を検出することで、車載機器の操作に付随した機能が適切に発現したか否か、車両の組立て工場において、組立て作業に従事する作業員の組立て作業が、予め決められた手順に従って行われたか否か、車両の販売店において、車両の整備に従事する整備士の整備がマニュアルに従って行われたか否かなどを確認することができる。
画像処理装置13で用いるプログラムは、上述した画像処理装置13の機能を実現するための機能ブロックである処理部2を含む。処理部2は、撮像装置11と、表示装置12とを制御して協働させることで、撮像装置11から操作者を含む画像データを取得する機能(データ取得機能)と、取得した画像データを処理し、撮像装置11の位置ずれを補正したうえで操作者の姿勢を検出する機能(姿勢検出機能及び位置ずれ補正機能)と、操作者の姿勢から操作者の行動を推定する機能(行動推定機能)とを実現する。これらの機能に対応する機能ブロックとして、処理部2は、図1に示すように、取得部21と、検出部22と、算出部23と、補正部24と、推定部25と、出力部26とを備える。図1には、各部を便宜的に抽出して示す。
図7を参照して、画像処理装置13が情報を処理する際の手順を説明する。図7は、本実施形態の画像処理システム1における情報の処理を示すフローチャートの一例である。以下に説明する処理は、操作者7が所定操作を行っている間、画像処理装置13のプロセッサであるCPU131により所定の時間間隔で実行される。
以上のとおり、本実施形態によれば、所定操作を行う操作者7、及び前記操作者7が身に付ける物品のうち少なくとも一方に装着された撮像装置11から、前記操作者7を含む画像を取得する取得部21と、前記画像から前記操作者7の姿勢を検出する検出部22と、前記画像から、位置ずれを算出する基準としての基準位置を用いて前記撮像装置11の位置ずれを算出する算出部23と、前記位置ずれを用いて前記姿勢を補正する補正部24と、を備える、画像処理装置13が提供される。これにより、撮像装置11(たとえばカメラ)の装着位置がずれた場合でも、操作者7の姿勢を正しく認識することができる。
11…撮像装置
12…表示装置
13…画像処理装置
131…CPU
132…ROM
133…RAM
14…データベース
2…処理部
21…取得部
22…検出部
221…第1機械学習部
23…算出部
231…第2機械学習部
24…補正部
25…推定部
251…第3機械学習部
26…出力部
3…ウェアラブル端末
31…バンド
4、4a、4b…視野
5…ヘルメット
51…つば
52…ブラケット
6…カメラ
61…バンド
7…操作者(人間)
71…頭部
711…鼻
712…中心
72…首
73a…左腕の上腕
73b…左腕の前腕
73c…左手
74a…右腕の上腕
74b…右腕の前腕
74c…右手
75…胴体
76…左脚
77…右脚
8、9…矢印
P1~P32…点
α…角度
Claims (10)
- 人間及び前記人間が身に付ける物品のうち少なくとも一方に装着された撮像装置から、前記人間を含む画像を取得する取得部と、
前記画像から前記人間の姿勢を検出する検出部と、
前記画像から、位置ずれを算出する基準としての基準位置を用いて前記撮像装置の位置ずれを算出する算出部と、
前記位置ずれを用いて前記姿勢を補正する補正部と、を備える、画像処理装置。 - 前記姿勢から前記人間の行動を推定するように予め学習された行動推定モデルを用いて、補正された前記姿勢から前記人間の行動を推定する推定部を備える、請求項1に記載の画像処理装置。
- 前記検出部は、前記画像から前記人間の姿勢を推定するように予め学習された姿勢推定モデルを用いて、前記画像から前記人間の姿勢を検出する、請求項1又は2に記載の画像処理装置。
- 前記算出部は、前記画像に含まれる前記人間の身体の一部を前記基準位置として識別し、所定状態の前記基準位置に対する前記位置ずれを算出する、請求項1~3のいずれか一項に記載の画像処理装置。
- 前記基準位置は、
前記撮像装置が前記人間に装着されている場合には、前記撮像装置が装着された部分以外の前記人間の身体の部位のうち、部位自体に関節により可動する部分が存在せず、且つ、前記撮像装置が装着された部分との間に関節が存在しない部位であり、
前記撮像装置が前記物品に装着されている場合には、前記物品と接触している部分以外の前記人間の身体の部位のうち、部位自体に関節により可動する部分が存在せず、且つ、前記人間が身に付けた前記物品と接触している部分との間に関節が存在しない部位である、請求項4に記載の画像処理装置。 - 前記取得部は、前記基準位置を含む時系列の画像を取得し、
前記算出部は、前記時系列の画像における前記基準位置の移動量を用いて前記位置ずれを算出する、請求項4又は5に記載の画像処理装置。 - 前記算出部は、前記画像における前記基準位置の位置から前記位置ずれを推定するように予め学習された位置ずれ推定モデルを用いて前記位置ずれを算出する、請求項4又は5に記載の画像処理装置。
- 前記取得部は、前記撮像装置の位置を特定するための模様を含む画像を取得し、
前記算出部は、前記模様を前記基準位置として識別し、前記模様から特定された前記撮像装置の位置を用いて前記位置ずれを算出する、請求項1~7のいずれか一項に記載の画像処理装置。 - 前記検出部は、前記人間の周囲に設置された検出装置を用いて、前記撮像装置の位置と、前記人間の姿勢とを検出し、
前記算出部は、前記検出部にて検出された、前記撮像装置の位置と前記人間の姿勢とに基づいて前記位置ずれを算出する、請求項1~8のいずれか一項に記載の画像処理装置。 - プロセッサにより実行される画像処理方法において、
前記プロセッサは、
人間及び前記人間が身に付ける物品のうち少なくとも一方に装着された撮像装置から、前記人間を含む画像を取得し、
前記画像から前記人間の姿勢を検出し、
前記画像から、位置ずれを算出する基準としての基準位置を用いて前記撮像装置の位置ずれを算出し、
前記位置ずれを用いて前記姿勢を補正する、画像処理方法。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2023552653A JPWO2023058221A1 (ja) | 2021-10-08 | 2021-10-08 | |
PCT/JP2021/037332 WO2023058221A1 (ja) | 2021-10-08 | 2021-10-08 | 画像処理装置及び画像処理方法 |
CN202180103139.7A CN118076972A (zh) | 2021-10-08 | 2021-10-08 | 图像处理装置以及图像处理方法 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2021/037332 WO2023058221A1 (ja) | 2021-10-08 | 2021-10-08 | 画像処理装置及び画像処理方法 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2023058221A1 true WO2023058221A1 (ja) | 2023-04-13 |
Family
ID=85804060
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2021/037332 WO2023058221A1 (ja) | 2021-10-08 | 2021-10-08 | 画像処理装置及び画像処理方法 |
Country Status (3)
Country | Link |
---|---|
JP (1) | JPWO2023058221A1 (ja) |
CN (1) | CN118076972A (ja) |
WO (1) | WO2023058221A1 (ja) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2016099982A (ja) | 2014-11-26 | 2016-05-30 | 日本電信電話株式会社 | 行動認識装置、行動学習装置、方法、及びプログラム |
-
2021
- 2021-10-08 JP JP2023552653A patent/JPWO2023058221A1/ja active Pending
- 2021-10-08 CN CN202180103139.7A patent/CN118076972A/zh active Pending
- 2021-10-08 WO PCT/JP2021/037332 patent/WO2023058221A1/ja active Application Filing
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2016099982A (ja) | 2014-11-26 | 2016-05-30 | 日本電信電話株式会社 | 行動認識装置、行動学習装置、方法、及びプログラム |
Non-Patent Citations (2)
Title |
---|
ANONYMOUS: "Developed motion capture technology using a compact chest- mounted camera", TOKYO INSTITUTE OF TECHNOLOGY, 21 October 2020 (2020-10-21), XP093056932, Retrieved from the Internet <URL:https://www.titech.ac.jp/news/2020/048131> [retrieved on 20230622] * |
HATANAKA TETSUO, WATANABE Y., KOMURO T., ISHIKAWA M.: "A Method of Human Gait Estimation Using a Wearable Camera System outline of human gait estimation", PROCEEDINGS OF THE 2009 JSME CONFERENCE ON ROBOTICS AND MECHATRONICS, 24 May 2009 (2009-05-24), XP093056924, Retrieved from the Internet <URL:http://ishikawa-vision.org/members/watanabe/hatanaka_ROBOMEC2009.pdf> [retrieved on 20230622] * |
Also Published As
Publication number | Publication date |
---|---|
CN118076972A (zh) | 2024-05-24 |
JPWO2023058221A1 (ja) | 2023-04-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108268129B (zh) | 对动作捕捉手套上的多个传感器进行校准的方法和装置及动作捕捉手套 | |
US9278453B2 (en) | Biosleeve human-machine interface | |
Liu et al. | A glove-based system for studying hand-object manipulation via joint pose and force sensing | |
TWI549655B (zh) | 關節活動度量測裝置及其量測方法 | |
US20150054850A1 (en) | Rehabilitation device and assistive device for phantom limb pain treatment | |
KR20120025582A (ko) | 로봇의 교시 데이터를 작성하는 방법 및 로봇 교시 시스템 | |
CN107856014B (zh) | 基于手势识别的机械臂位姿控制方法 | |
CN110978064B (zh) | 人机协作中人体安全评估方法及*** | |
EP2418562B1 (en) | Modelling of hand and arm position and orientation | |
US11667031B2 (en) | Teaching method | |
Xu et al. | Development of a hybrid motion capture method using MYO armband with application to teleoperation | |
JPWO2010027015A1 (ja) | モーションキャプチャ装置 | |
JP4765075B2 (ja) | ステレオ画像を利用した物体の位置および姿勢認識システムならびに物体の位置および姿勢認識方法を実行するプログラム | |
WO2023058221A1 (ja) | 画像処理装置及び画像処理方法 | |
JP6491121B2 (ja) | 身体歪み検知システム | |
CN115919250A (zh) | 一种人体动态关节角测量*** | |
JP2006130580A (ja) | ロボットによる任意形状物体の把持方法 | |
JP2020054433A (ja) | 身体姿勢検知システム | |
JP4882930B2 (ja) | 行動推定装置 | |
JP3884249B2 (ja) | 人間型ハンドロボット用教示システム | |
JP2016206081A (ja) | 動作推測装置および動作推測方法 | |
Ogata et al. | Estimating movements of human body for the shirt-type wearable device mounted on the strain sensors based on convolutional neural networks | |
CN112057083B (zh) | 可穿戴人体上肢位姿采集设备及采集方法 | |
Kim et al. | Thumb joint angle estimation for soft wearable hand robotic devices | |
JP4878462B2 (ja) | コミュニケーションロボット |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21959963 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2023552653 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2021959963 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2021959963 Country of ref document: EP Effective date: 20240508 |