CN111046701A - Tumble detection method and tumble detection equipment - Google Patents
Tumble detection method and tumble detection equipment Download PDFInfo
- Publication number
- CN111046701A CN111046701A CN201811191665.XA CN201811191665A CN111046701A CN 111046701 A CN111046701 A CN 111046701A CN 201811191665 A CN201811191665 A CN 201811191665A CN 111046701 A CN111046701 A CN 111046701A
- Authority
- CN
- China
- Prior art keywords
- information
- video
- skeleton information
- skeleton
- character
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
Abstract
The embodiment of the invention discloses a fall detection method and fall detection equipment, which are used for detecting falls of a monitored object under the condition that the monitored object does not need to wear the detection equipment, so that the convenience of fall detection is improved. The embodiment of the invention obtains the video of the collected object; determining whether the acquisition object is a moving person according to the video; if so, extracting the character skeleton information of the character from the video; matching the character skeleton information with sample skeleton information in an action library to obtain a matching result; and finally, determining whether the acquired object has a tumbling event or not according to the matching result.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a tumble detection method and tumble detection equipment.
Background
The detection of human body falls is an important part of human body movement behavior recognition, and especially for old people who are at home alone, if the old people fall and do not know, the detection may bring unforeseen consequences. Therefore, as the aging process of society is accelerated, detection of accidental falls for "empty nesters" is being urgently required.
At present, in order to detect the accidental fall of the empty nest old people, the action parameters in the user movement process can be detected by means of some wearable detection equipment for detecting the user movement actions, and then whether the user falls or not is detected. However, the wearable detection device needs to be carried about by the monitored object, which is troublesome and the experience of the monitored object is poor.
Disclosure of Invention
The embodiment of the invention provides a fall detection method and fall detection equipment, which are used for detecting falls of a monitored object under the condition that the monitored object does not need to wear the detection equipment, so that the convenience of fall detection is improved.
In a first aspect, an embodiment of the present invention provides a fall detection method, including:
acquiring a video of a collection object;
determining whether the acquisition object is a moving person according to the video;
if so, extracting the character skeleton information of the character from the video;
matching the character skeleton information with sample skeleton information in an action library to obtain a matching result;
and determining whether the collecting object has a falling event or not according to the matching result.
In some embodiments, said acquiring a video of a subject comprises:
determining whether the infrared information of the acquisition object is detected or not through a pyroelectric infrared sensor;
and if the infrared information is detected, acquiring the video of the acquisition object according to the transmitting position of the infrared information.
In a second aspect, an embodiment of the present invention further provides a fall detection device, including:
the acquisition unit is used for acquiring a video of a collection object;
a first determination unit, configured to determine whether the captured object is a moving person according to the video;
an extraction unit configured to extract person skeleton information of a person from the video when the capture object is the moving person;
the matching unit is used for matching the character skeleton information with sample skeleton information in an action library to obtain a matching result;
and the second determining unit is used for determining whether the collecting object has a tumbling event according to the matching result.
In some embodiments, the second determination unit comprises:
the first determining subunit is used for determining that the collecting object has a tumbling event when the matching is successful;
and the second determining subunit is used for determining whether the acquired object has a tumbling event or not according to the size information and the movement trend information of the character skeleton information if the matching fails.
In some embodiments, the second determining subunit comprises:
the first judgment module is used for judging whether the size information of the figure skeleton information meets the falling size condition or not to obtain a size judgment result;
the second judgment module is used for judging whether the movement trend information of the figure skeleton information conforms to the falling trend characteristic or not to obtain a trend judgment result;
and the determining module is used for determining whether the collecting object has a tumbling event according to the size judging result and the trend judging result.
In some embodiments, the fall detection device further comprises:
and the storage unit is used for storing the character skeleton information into the action library when the falling event of the collection object is determined.
In some embodiments, the saving unit is specifically configured to:
storing skeleton characteristic point information in the character skeleton information into the action library;
the matching unit is specifically configured to:
matching skeleton characteristic point information in the character skeleton information with sample skeleton characteristic point information in sample skeleton information in the action library.
In some embodiments, the fall detection device further comprises:
the sending unit is used for sending fall reminding information to the monitoring terminal, and the fall reminding information comprises a video corresponding to the fall event.
In some embodiments, the fall detection device further comprises:
the receiving unit is used for receiving feedback information sent by the monitoring terminal, and the feedback information comprises the falling event detection result;
and the correction unit is used for correcting the action library according to the feedback information.
In some embodiments, the first determining unit is specifically configured to:
detecting whether a collection object in the video moves;
and if so, determining whether the acquired object meets the characteristics of the person.
In some embodiments, the extracted unit is specifically configured to:
extracting the target outline of the person from the video by using a background difference method;
and extracting the human skeleton information from the target contour by using a morphological algorithm.
In some embodiments, the obtaining unit is specifically configured to:
determining whether the infrared information of the acquisition object is detected or not through a pyroelectric infrared sensor;
and if the infrared information is detected, acquiring the video of the acquisition object according to the transmitting position of the infrared information.
Yet another aspect of the present application provides a computer-readable storage medium having stored therein instructions, which when executed on a computer, cause the computer to perform the method of the above-described aspects.
Yet another aspect of the present application provides a computer program product containing instructions which, when run on a computer, cause the computer to perform the method of the above-described aspects.
The tumble detection device in the embodiment of the invention acquires a video of a collected object; then determining whether the acquisition object is a moving person according to the video; if so, extracting the character skeleton information of the character from the video; matching the character skeleton information with sample skeleton information in an action library to obtain a matching result; and finally, determining whether the acquired object has a tumbling event or not according to the matching result. Whether this scheme is direct judges through the video that fall detection equipment obtained and gathers object (control object) and take place the fall incident, does not need control object to dress detection equipment, has improved the convenience that falls and detect.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a schematic view of a fall detection system according to an embodiment of the present invention;
fig. 2 is a first flowchart of a fall detection method according to an embodiment of the present invention;
fig. 3 is a schematic flow chart of a fall detection method according to an embodiment of the present invention;
fig. 4a is a schematic structural diagram of a fall detection apparatus according to an embodiment of the present invention;
fig. 4b is a schematic structural diagram of a fall detection apparatus according to an embodiment of the present invention;
fig. 4c is a schematic structural diagram of a fall detection apparatus according to an embodiment of the present invention;
fig. 4d is a schematic structural diagram of a fall detection apparatus according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a hardware structure of the fall detection apparatus according to the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, an embodiment of the present invention provides a fall detection system, including: the system comprises a fall detection device 10 and a monitoring terminal 20, wherein the fall detection device 10 in the scheme can be an internet protocol Camera (IPC, IP Camera), a lens of the IPC has a 360-degree (omnibearing) rotation function, the monitoring terminal 20 comprises a mobile phone, a tablet computer, a notebook computer and other devices, the fall detection device 10 is connected with the monitoring terminal 20 through a network 30, and the network 30 comprises a router, a gateway and other network entities.
The above example of fig. 1 is only an example of a system architecture for implementing the embodiment of the present invention, and the embodiment of the present invention is not limited to the system architecture shown in fig. 1, and various embodiments of the present invention are proposed based on the system architecture.
In an embodiment, a fall detection method is provided, which may be executed by a processor in a fall detection device, as shown in fig. 2, the fall detection method comprising:
201. and acquiring a video of the acquisition object.
Wherein, imbed pyroelectric infrared sensor (PIR in the fall detection equipment in this application, when falling detection equipment detected the infrared information of collection object through PIR, then the explanation has the living beings to get into monitoring range this moment, at this moment, acquires the video of collection object according to the transmission position of this infrared information, specifically, turns to the direction of this infrared information transmission position with the camera to acquire the video of this collection object.
202. And determining whether the acquisition object is a moving person according to the video.
After the captured video of the captured object is obtained, whether the captured object is a moving person is determined according to the video, specifically:
(1) detecting whether a collection object in a video moves;
(2) and if so, determining whether the acquired object meets the characteristics of the person.
When the acquisition object in the video is detected to be moving, whether the acquisition object meets the characteristics of a person or not is determined through image algorithm processing, namely whether the moving object is a person or not is determined, wherein if the acquisition object is determined to be a static object, the process is stopped, and if the acquisition object is determined to be a moving object but the acquisition object does not meet the characteristics of the person, the process is also stopped.
203. If the person is a moving person, the person skeleton information of the person is extracted from the video.
When the acquired acquisition object is determined to be a moving person, at this time, the person skeleton information of the person needs to be extracted from the video, specifically:
extracting a target outline of a person from the video by using a background difference method;
and extracting the human skeleton information from the target contour by using a morphological algorithm.
More specifically, the extraction of skeletal information from a video goes through nine steps: graying an image, extracting a target contour by a background difference method, enhancing contrast by using a CLAHE algorithm, Gaussian filtering, performing edge detection by using a Sobel (Solel) operator, performing wavelet denoising, binarizing by a maximum inter-class error method, performing morphological operation and median filtering, and finally extracting character skeleton information from the video.
Further, the character skeleton information includes skeleton feature point information, and feature points in the skeleton feature point information may be set at the corresponding parietal bone, hand bone, elbow bone, crotch bone, knee bone, shoulder bone, and the like in the image of the character skeleton in the character skeleton information, and specific positions are not limited herein.
204. And matching the character skeleton information with the sample skeleton information in the action library to obtain a matching result.
In this embodiment, skeleton information actions of various previously detected fall events are stored in the action library, and after the fall detection device extracts the person skeleton information of a person in a video, the person skeleton information is matched with sample skeleton information in the action library, that is, whether an action matched with the detected skeleton action (skeleton information of multiple frames of images in the video) exists in the action library is detected, and then a matching result is obtained, where the matching result includes: and the matching success result indicates that skeleton information matched with the detected skeleton information exists in the action library, and the matching failure result indicates that the skeleton information matched with the detected skeleton information does not exist in the action library.
The action library may be located in the cloud or in the local storage, and is not limited herein.
Specifically, feature points in the person skeleton information may be matched with feature points of sample skeleton information in the action library.
205. And determining whether the acquired object has a tumbling event or not according to the matching result.
When the character skeleton information extracted from the video is matched with the skeleton information in the action library to obtain a matching result, because the matching result has two types, the scheme executes different steps according to different matching results, specifically:
(1) and if the matching is successful, determining that the falling event of the collection object occurs.
At this time, it is indicated that sample skeleton information matched with the human body skeleton information extracted from the video exists in the action library, and since the sample skeleton information stored in the action library corresponds to the falling action, if the matching is successful at this time, it is indicated that the falling event occurs to the collection object at this time.
After judging that the collecting object has a fall event, sending reminding information to the monitoring terminal, wherein the reminding information comprises a video corresponding to the fall object, the length of the video can be 10 seconds, and specifically, the video can be: triggering the fall detection device to automatically record a 10s video from the beginning of the recording when the moving character is detected after the fall event is determined to have occurred. Of course, the video length may also be other lengths, which may be specifically set by a user, and the specific length is not limited herein.
(2) And if the matching fails, determining whether the acquired object has a tumbling event or not according to the size information and the movement trend information of the character skeleton information.
When the matching fails, it is indicated that skeleton information matched with the extracted character skeleton information does not exist in the action library at this moment, and whether the acquired object has a tumbling event or not needs to be determined according to the size information and the movement trend information of the character skeleton information, which is specifically as follows:
a. and judging whether the size information of the figure skeleton information meets the falling size condition or not to obtain a size judgment result.
Since the human body has a certain ratio, for example, the arm is on the upper half of the human body and the foot is on the lower half of the human body, if a person walks upright, the width and height of the human body are in a certain ratio (i.e., size), that is, the width and height of the skeleton information are also in a certain ratio. In one example, if the skeleton information has a width of 180 cm and a height of 30 cm, the width-to-height ratio in the size information of the skeleton information at this time is 6, and if the fall size condition at this time is: and if the width-height ratio is greater than 1, determining that the size information of the figure skeleton information meets the falling size condition, and determining that the size judgment result meets the size condition.
And if the size information of the figure skeleton information is judged not to be in accordance with the falling size condition, stopping the process.
b. And judging whether the movement trend information of the character skeleton information conforms to the falling trend characteristics or not to obtain a trend judgment result.
If the falling event of the collection object is judged only according to the size information of the figure skeleton information, misjudgment is likely to happen, so that the scheme judges whether the falling event of the collection object is caused according to the movement trend information of the figure skeleton information besides the size information.
Because the speed of the person is high when the person really falls down and the gravity center of the person tends to descend, a plurality of false falls can be eliminated according to the characteristics of the human body.
In one example, one frame of video is about 0.03 second, two images separated by 10 frames are taken as a judgment standard of motion trend information, that is, the motion trend information of the character skeleton information is judged once every 0.3 second, if the falling trend characteristic is: if the speed of lowering of the center of gravity in the skeleton information is greater than a predetermined value, the person is specified at this timeThe movement trend information of the skeleton information conforms to the falling trend characteristic, wherein the certain value can be 1.5 m/s2That is, if the velocity of lowering of the center of gravity corresponding to the human skeleton information is calculated at this time to be 2 m/s2At this time, it is indicated that the movement trend information of the character skeleton information conforms to the falling trend characteristic.
According to the scheme, the character skeleton information in the video is extracted firstly, whether the collected object falls down or not is roughly judged according to the proportion of the character skeleton information (the size information of the skeleton information), and the motion trend information is used for screening false falls under the condition that the collected object is judged to fall down, so that the true fall state is left.
In addition, in some embodiments, step b in this step may be performed after step a, before step a, or simultaneously with step a, which is not limited herein.
c. And determining whether the collecting object has a tumbling event or not according to the size judgment result and the trend judgment result.
In this embodiment, after the size determination result and the trend determination result are obtained, whether a tumble event occurs in the acquisition object in the video is determined according to the size determination result and the trend determination result.
When the size judgment result and the trend judgment result both determine that the falling event occurs to the collection object (namely, the size information of the figure skeleton information conforms to the falling size condition, and the movement trend information of the figure skeleton information conforms to the falling trend characteristic), it can be determined that the falling event occurs to the collection object at the moment.
When the falling event of the collection object is uncertain in any one of the size judgment result and the trend judgment result (namely the size information of the figure skeleton information does not accord with the falling size condition and/or the movement trend information of the figure skeleton information does not accord with the falling trend characteristic), the collection object can be judged not to have the falling event.
It should be noted that, after it is determined that the object is captured and a fall event occurs, a reminding message is sent to the monitoring terminal, where the reminding message includes a video corresponding to the object, the length of the video may be 10 seconds, and specifically, the video may be: triggering the fall detection device to automatically record a 10s video from the beginning of the recording when the moving character is detected after the fall event is determined to have occurred. Of course, the video length may also be other lengths, which may be specifically set by a user, and the specific length is not limited herein.
In addition, if the falling event of the collection object is determined according to the size judgment result and the trend judgment result, at the moment, the cloud starts self-learning, the character skeleton information is added into the action library, and the sample skeleton information in the action library is enriched.
After receiving the reminding information, a user in the monitoring terminal can feed back whether the falling event in the reminding information is accurate or not to the falling detection equipment through the monitoring terminal, if not, the falling detection equipment starts deep self-learning after receiving the feedback information, the figure skeleton information corresponding to the feedback information in the action library is reprocessed, and the figure skeleton information can be removed in the action library.
Finally, inaccurate falling actions (character skeleton information) in the action library can be eliminated through a large amount of user feedback information and continuous polling, and the false alarm rate is reduced.
The tumble detection device in the embodiment of the invention acquires a video of a collected object; then determining whether the collected object is a moving person according to the video; if so, extracting the character skeleton information of the character from the video; matching the character skeleton information with sample skeleton information in an action library to obtain a matching result; and finally, determining whether the acquired object has a tumbling event or not according to the matching result. Whether this scheme is direct judges through the video that fall detection equipment obtained and gathers object (control object) and take place the fall incident, does not need control object to dress detection equipment, has improved the convenience that falls and detect.
In an embodiment, a fall detection method is provided, which may be executed by a processor in a fall detection device, as shown in fig. 3, the fall detection method comprising:
301. and determining whether the infrared information of the acquisition object is detected or not through a pyroelectric infrared sensor.
Embedding PIR in the fall detection equipment in this application, when fall detection equipment passes through PIR and detects the infrared information of gathering the object, then explain this moment that there is the living beings to get into monitoring range.
302. And if the infrared information is detected, acquiring the video of the acquisition object according to the transmitting position of the infrared information.
If the infrared information is detected, at the moment, the camera turns to the direction of the infrared information transmitting position to acquire the video of the acquisition object.
303. Whether a collection object in the video moves is detected.
After the video of the acquisition object is acquired, it is further required to detect whether the acquisition object in the video moves, specifically, it may be determined whether the acquisition formation moves by using the position of the acquisition object in the multi-frame picture in the video, if the position of the acquisition object in the multi-frame picture is different, it indicates that the acquisition object is moving at this time, otherwise, the acquisition object is not moving, wherein if it is determined that the acquisition object is a stationary object (not moving), the process needs to be stopped.
304. And if so, determining whether the acquired object meets the characteristics of the person.
When it is determined that the object to be captured is a moving object to be captured, it is necessary to further determine whether the moving object to be captured is a person, specifically, it may be further determined whether the object to be captured conforms to a character feature, specifically, it is determined whether the object to be captured conforms to the character feature by image algorithm processing, where when the object to be captured does not conform to the character feature, the process needs to be stopped.
305. And if the image accords with the character characteristics, extracting the target outline of the character from the video by using a background difference method.
Firstly, preprocessing an image: the method mainly comprises the steps of graying and filtering the image.
And then extracting the target contour: and subtracting the N frame image from the N +1 frame image, subtracting the N +1 frame image from the N frame image, and adding the two subtracted images to increase the contrast of target contour extraction.
306. And extracting the human skeleton information from the target contour by using a morphological algorithm.
When the target contour of a person is extracted from a video, the person skeleton information needs to be extracted from the target contour by using a morphological algorithm, and specifically, the extraction of the skeleton information from the video goes through nine steps: the method comprises the steps of graying an image, extracting a target contour by a background difference method, enhancing contrast by a CLAHE algorithm, carrying out Gaussian filtering, carrying out edge detection by a Sobel (Solel) operator, carrying out wavelet denoising, carrying out binarization by a maximum inter-class error method, carrying out morphological operation and median filtering, and finally extracting skeleton information of a person from a video.
Further, the character skeleton information includes skeleton feature point information, and feature points in the skeleton feature point information may be set at the corresponding parietal bone, hand bone, elbow bone, crotch bone, knee bone, shoulder bone, and the like in the image of the character skeleton in the character skeleton information, and specific positions are not limited herein.
307. And determining whether the feature point information in the character skeleton information is matched with the feature point information of the sample skeleton information in the action library.
In the embodiment, skeleton information actions of various previously detected fall events are stored in the action library, and after the fall detection device extracts the person skeleton information of a person in a video, the person skeleton information is matched with sample skeleton information in the action library, that is, whether an action matched with the detected skeleton information action (skeleton information of multiple frames of images in the video) exists in the action library is detected, and then a matching result is obtained.
Wherein the matching result comprises: and the matching success result indicates that skeleton information matched with the detected skeleton information exists in the action library, and the matching failure result indicates that the skeleton information matched with the detected skeleton information does not exist in the action library.
The action library may be located in the cloud or in the local storage, and is not limited herein.
Specifically, the feature point information in the person skeleton information may be matched with the feature point information of the sample skeleton information in the action library.
At this time, the fact that the character skeleton information extracted from the video exists in the action library is indicated, and since the character skeleton information stored in the action library corresponds to the falling actions, if the matching is successful at this time, the fact that the falling event occurs to the collection object at this time is indicated.
After judging that the collecting object has a fall event, sending reminding information to the monitoring terminal, wherein the reminding information comprises a video corresponding to the fall object, the length of the video can be 10 seconds, and specifically, the video can be: triggering the fall detection device to automatically record a 10s video from the beginning of the recording when the moving character is detected after the fall event is determined to have occurred. Of course, the video length may also be other lengths, which may be specifically set by a user, and the specific length is not limited herein.
308. If not, judging whether the size information of the figure skeleton information meets the falling size condition or not to obtain a size judgment result.
If the feature points of the character skeleton information are not matched with the feature points of the skeleton information in the action library, then a fall detection and identification system is required to be entered (namely, size judgment and movement trend information judgment of the character skeleton information are carried out), firstly, whether the size information of the character skeleton information meets a fall size condition is required to be judged, namely, whether the width-height ratio of the character skeleton information meets the fall size condition is judged (at the moment, the fall size condition is that the width-height ratio is larger than 1).
Since the human body has a certain ratio, for example, if a person walks upright, the width and height of the person are in a certain ratio (i.e., size), i.e., the width and height of the skeleton information are also in a certain ratio. In one example, if the skeleton information has a width of 180 cm and a height of 30 cm, the width-to-height ratio in the size information of the skeleton information at this time is 6, and if the fall size condition at this time is: and if the width-height ratio is greater than 1, determining that the size information of the figure skeleton information meets the falling size condition, and determining that the size judgment result meets the size condition.
And if the size information of the figure skeleton information is judged not to be in accordance with the falling size condition, stopping the process.
309. And judging whether the movement trend information of the character skeleton information conforms to the falling trend characteristics or not to obtain a trend judgment result.
If the falling event of the collection object is judged only according to the size information of the figure skeleton information, misjudgment is likely to happen, so that the scheme judges whether the falling event of the collection object is caused according to the movement trend information of the figure skeleton information besides the size information.
Because the speed of the person is high when the person really falls down and the gravity center of the person tends to descend, a plurality of false falls can be eliminated according to the characteristics of the human body.
In one example, a frame time of a video is about 0.03 second, two images separated by 10 frames are taken as a judgment standard of motion trend information, that is, the motion trend information of the human skeleton information is judged once every 0.3 second, and the falling trend characteristic is as follows: if the speed of gravity center lowering in the skeleton information is greater than a certain value, it indicates that the movement trend information of the character skeleton information conforms to the falling trend characteristic, wherein the certain value may be 1.5 m/s2That is, if the velocity of lowering of the center of gravity corresponding to the human skeleton information is calculated at this time to be 2 m/s2At this time, it is indicated that the movement trend information of the character skeleton information conforms to the falling trend characteristic.
Namely, whether the movement trend information of the figure skeleton information accords with the falling trend characteristic or not is judged, and the obtained trend judgment result comprises the following steps:
judging whether the gravity center descending speed in the figure skeleton information is larger than a certain value, if so, judging that the figure skeleton information movement trend information conforms to the falling trend, if not, judging that the figure skeleton information movement trend information does not conform to the falling trend, wherein the certain value can be 1.5 m/s2The center of gravity lowering speed may be a center of gravity lowering speed of the person skeleton information calculated every 0.3 seconds from the person whose movement is detected.
In one embodiment, the method firstly extracts the character skeleton information in the video, firstly roughly judges whether the collection object falls or not according to the proportion of the character skeleton information (size information of the skeleton information), and then screens false falls by using the movement trend information under the condition that the collection object is judged to be in the falling state, so that the real falling state is left.
In addition, in some embodiments, step 309 may be performed after step 308, before step 308, or simultaneously with step 308, which is not limited herein.
310. And determining whether the collecting object has a tumbling event or not according to the size judgment result and the trend judgment result.
In this embodiment, after the size determination result and the trend determination result are obtained, whether a tumble event occurs in the acquisition object in the video is determined according to the size determination result and the trend determination result.
When the size judgment result and the trend judgment result both determine that the falling event occurs to the collection object (namely, the size information of the figure skeleton information conforms to the falling size condition, and the movement trend information of the figure skeleton information conforms to the falling trend characteristic), it can be determined that the falling event occurs to the collection object at the moment.
When the falling event of the collection object is uncertain in any one of the size judgment result and the trend judgment result (namely the size information of the figure skeleton information does not accord with the falling size condition and/or the movement trend information of the figure skeleton information does not accord with the falling trend characteristic), the collection object can be judged not to have the falling event.
311. And if the falling event occurs, sending falling reminding information to the monitoring terminal.
After judging that the collecting object has a fall event, sending reminding information to the monitoring terminal, wherein the reminding information comprises a video corresponding to the fall object, the length of the video can be 10 seconds, and specifically, the video can be: triggering the fall detection device to automatically record a 10s video from the beginning of the recording when the moving character is detected after the fall event is determined to have occurred. Of course, the video length may also be other lengths, which may be specifically set by a user, and the specific length is not limited herein.
312. And storing the characteristic point information in the character skeleton information into an action library.
If the falling event of the collection object is determined according to the size judgment result and the trend judgment result, the cloud starts self-learning at the moment, the characteristic points are marked on the skeleton image in the character skeleton information, and the characteristic point information is stored in the action library to enrich the action library.
313. And receiving feedback information sent by the monitoring terminal.
After receiving the reminding information, the user in the monitoring terminal can feed back whether the falling event in the reminding information is accurate or not to the falling detection equipment through the monitoring terminal.
314. And correcting the action library according to the feedback information.
If the information is inaccurate, the falling detection equipment starts deep self-learning after receiving the feedback information, the figure skeleton information corresponding to the feedback information in the action library is reprocessed, and the figure skeleton information can be removed from the action library.
Finally, inaccurate falling actions (character skeleton information) in the action library can be eliminated through a large amount of user feedback information and continuous polling, and the false alarm rate is reduced.
The tumble detection device in the embodiment of the invention acquires a video of a collected object; then determining whether the collected object is a moving person according to the video; if so, extracting the character skeleton information of the character from the video; matching the character skeleton information with sample skeleton information in an action library to obtain a matching result; and finally, determining whether the acquired object has a tumbling event or not according to the matching result. Whether this scheme is direct judges the collection object (control object) through the video that fall detection equipment obtained and whether takes place the fall incident, does not need control object to dress detection equipment, has improved the convenience that falls to detect, realizes the with low costs of this scheme moreover, only needs imbed software algorithm in fall detection equipment and handles etc..
In an embodiment, there is also provided a fall detection device, as shown in fig. 4a, which may comprise: an acquisition unit 401, a first determination unit 402, an extraction unit 403, a matching unit 404, and a second determination unit 405;
an obtaining unit 401, configured to obtain a video of a collection object;
a first determination unit 402, configured to determine whether the capture object is a moving person according to the video;
an extraction unit 403 for extracting person skeleton information of a person from the video when the capture object is a moving person;
a matching unit 404, configured to match the person skeleton information with sample skeleton information in the action library to obtain a matching result;
and a second determining unit 405, configured to determine whether the collecting object has a fall event according to the matching result.
In one embodiment, referring to fig. 4b, the second determining unit 405 includes:
the first determining subunit 4051 is configured to determine that the collecting object has a fall event when the matching is successful;
and a second determining sub-unit 4052, configured to determine whether the collecting object has a tumbling event according to the size information and the movement trend information of the character skeleton information if the matching fails.
Wherein the second determining subunit 4052 includes:
the first judging module 40521 is configured to judge whether the size information of the figure skeleton information meets a fall size condition, so as to obtain a size judgment result;
a second judging module 40522, configured to judge whether the movement trend information of the character skeleton information conforms to a tumbling trend characteristic, to obtain a trend judgment result;
the determining module 40523 is configured to determine whether the collecting object has a tumbling event according to the size determination result and the trend determination result.
In an embodiment, referring to fig. 4c, the fall detection apparatus further comprises:
and the storage unit 406 is configured to store the character skeleton information into the action library when it is determined that the collection object has a fall event.
In some embodiments, the saving unit 406 is specifically configured to:
storing the characteristic point information in the character skeleton information into an action library;
the matching unit 404 is specifically configured to:
matching skeleton characteristic point information in the character skeleton information with sample skeleton characteristic point information in sample skeleton information in the action library.
In some embodiments, the fall detection device further comprises:
and the sending unit is used for sending fall reminding information to the monitoring terminal, and the fall reminding information comprises a video corresponding to the fall event.
In an embodiment, referring to fig. 4d, the fall detection apparatus further comprises:
a receiving unit 407, configured to receive feedback information sent by the monitoring terminal, where the feedback information includes a tumble event detection result;
and a correction unit 408 for performing correction processing on the action library according to the feedback information.
In some embodiments, the first determining unit 402 is specifically configured to:
detecting whether a collection object in a video moves;
and if so, determining whether the acquired object meets the characteristics of the person.
In some embodiments, the extracted unit 403 is specifically configured to:
extracting a target outline of a person from the video by using a background difference method;
and extracting the human skeleton information from the target contour by using a morphological algorithm.
In some embodiments, the obtaining unit 401 is specifically configured to:
determining whether infrared information of an acquisition object is detected or not through a pyroelectric infrared sensor;
and if the infrared information is detected, acquiring the video of the acquisition object according to the transmitting position of the infrared information.
In the embodiment of the present invention, an obtaining unit 401 obtains a video of a collection object; the first determination unit 402 determines whether the captured object is a moving person from the video; if yes, the extraction unit 403 extracts the character skeleton information of the character from the video; the matching unit 404 matches the character skeleton information with the sample skeleton information in the action library to obtain a matching result; finally, the second determination unit 405 determines whether the object has a fall event according to the matching result. Whether this scheme is direct judges the collection object (control object) through the video that fall detection equipment obtained and whether takes place the fall incident, does not need control object to dress detection equipment, has improved the convenience that falls to detect, realizes the with low costs of this scheme moreover, only needs imbed software algorithm in fall detection equipment and handles etc..
Referring to fig. 5, an embodiment of the present invention provides a fall detection apparatus 500, which may include one or more processing cores of a processor 501, one or more computer-readable storage media of a memory 502, a Radio Frequency (RF) circuit 503, a power supply 504, an input unit 505, and a display unit 506. It will be appreciated by those skilled in the art that the fall detection device structure shown in fig. 5 does not constitute a limitation of the fall detection device, and may comprise more or fewer components than those shown, or some components in combination, or a different arrangement of components. Wherein:
the processor 501 is a control center of the fall detection apparatus, connects various parts of the entire fall detection apparatus by using various interfaces and lines, and executes various functions and processes data of the fall detection apparatus by operating or executing software programs and/or modules stored in the memory 502 and calling data stored in the memory 502, thereby integrally monitoring the fall detection apparatus. Optionally, processor 501 may include one or more processing cores; preferably, the processor 501 may integrate an application processor, which mainly handles operating systems, user interfaces, application programs, etc., and a modem processor, which mainly handles wireless communications. It will be appreciated that the modem processor described above may not be integrated into the processor 501.
The memory 502 may be used to store software programs and modules, and the processor 501 executes various functional applications and data processing by operating the software programs and modules stored in the memory 502.
The RF circuit 503 may be used for receiving and transmitting signals during the process of transmitting and receiving information.
The fall detection device also includes a power source 504 (e.g., a battery) for powering the various components, which may preferably be logically connected to the processor 501 via a power management system to manage charging, discharging, and power consumption management functions via the power management system.
The fall detection apparatus may further include an input unit 505, and the input unit 505 may be used to receive input numeric or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.
The fall detection device may further comprise a display unit 506, which display unit 506 may be used to display information input by or provided to the user as well as various graphical user interfaces of the fall detection device, which may be constituted by graphics, text, icons, video and any combination thereof. Specifically, in this embodiment, the processor 501 in the fall detection apparatus loads an executable file corresponding to a process of one or more application programs into the memory 502 according to the following instructions, and the processor 501 runs the application programs stored in the memory 502, so as to implement various functions as follows:
acquiring a video of a collection object;
determining whether the collected object is a moving person according to the video;
if so, extracting the character skeleton information of the character from the video;
matching the character skeleton information with sample skeleton information in an action library to obtain a matching result;
and determining whether the acquired object has a tumbling event or not according to the matching result.
It will be understood by those skilled in the art that all or part of the steps of the methods of the above embodiments may be performed by instructions or by associated hardware controlled by the instructions, which may be stored in a computer readable storage medium and loaded and executed by a processor.
To this end, embodiments of the present invention provide a storage medium having stored therein a plurality of instructions, which can be loaded by a processor to perform the steps of any of the fall detection methods provided by the embodiments of the present invention. For example, the instructions may perform the steps of:
acquiring a video of a collection object;
determining whether the collected object is a moving person according to the video;
if so, extracting the character skeleton information of the character from the video;
matching the character skeleton information with sample skeleton information in an action library to obtain a matching result;
and determining whether the acquired object has a tumbling event or not according to the matching result.
The above operations can be implemented in the foregoing embodiments, and are not described in detail herein.
Wherein the storage medium may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like.
Since the instructions stored in the storage medium can execute the steps in any of the fall detection methods provided in the embodiments of the present invention, the beneficial effects that can be achieved by any of the fall detection methods provided in the embodiments of the present invention can be achieved, which are detailed in the foregoing embodiments and will not be described again here.
The embodiments of the invention have been described in detail, and specific examples are used herein to explain the principles and implementations of the invention, and the above descriptions are only used to help understand the method and the core ideas of the invention; meanwhile, for those skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A fall detection method, comprising:
acquiring a video of a collection object;
determining whether the acquisition object is a moving person according to the video;
if so, extracting the character skeleton information of the character from the video;
matching the character skeleton information with sample skeleton information in an action library to obtain a matching result;
and determining whether the collecting object has a falling event or not according to the matching result.
2. The method of claim 1, wherein said determining whether a fall event has occurred in said subject from said matching results comprises:
if the matching is successful, determining that the acquired object has a tumbling event;
and if the matching fails, determining whether the acquired object has a tumbling event or not according to the size information and the movement trend information of the character skeleton information.
3. The method of claim 2, wherein determining whether the subject has a fall event according to the size information and the movement tendency information of the human skeleton information comprises:
judging whether the size information of the figure skeleton information meets the falling size condition or not to obtain a size judgment result;
judging whether the movement trend information of the figure skeleton information conforms to the falling trend characteristics or not to obtain a trend judgment result;
and determining whether the collecting object has a falling event or not according to the size judgment result and the trend judgment result.
4. The method of claim 3, wherein after determining whether the subject has suffered a fall event based on the size determination and the trend determination, the method further comprises:
and if the fact that the collecting object falls down is determined, the character skeleton information is stored in the action library.
5. The method of claim 4, wherein saving the character armature information to the action library comprises:
storing skeleton characteristic point information in the character skeleton information into the action library;
the matching of the character skeleton information with the skeleton information in the action library comprises:
matching skeleton characteristic point information in the character skeleton information with sample skeleton characteristic point information in sample skeleton information in the action library.
6. The method of claim 4, wherein after determining that the subject has suffered a fall event, the method further comprises:
sending fall reminding information to the monitoring terminal, wherein the fall reminding information comprises a video corresponding to the fall event.
7. The method of any one of claims 6, wherein after sending the fall alert message to the monitoring terminal, the method further comprises:
receiving feedback information sent by the monitoring terminal, wherein the feedback information comprises a falling event detection result;
after the saving of the person skeleton information into the action library, the method further comprises:
and correcting the action library according to the feedback information.
8. The method of any of claims 1-7, wherein said determining from the video whether the capture object is a moving person comprises:
detecting whether a collection object in the video moves;
and if so, determining whether the acquired object meets the characteristics of the person.
9. The method of any of claims 1-7, wherein the extracting the person skeleton information for the person from the video comprises:
extracting the target outline of the person from the video by using a background difference method;
and extracting the human skeleton information from the target contour by using a morphological algorithm.
10. A fall detection device, comprising:
the acquisition unit is used for acquiring a video of a collection object;
a first determination unit, configured to determine whether the captured object is a moving person according to the video;
an extraction unit configured to extract person skeleton information of a person from the video when the capture object is the moving person;
the matching unit is used for matching the character skeleton information with sample skeleton information in an action library to obtain a matching result;
and the second determining unit is used for determining whether the collecting object has a tumbling event according to the matching result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811191665.XA CN111046701A (en) | 2018-10-12 | 2018-10-12 | Tumble detection method and tumble detection equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811191665.XA CN111046701A (en) | 2018-10-12 | 2018-10-12 | Tumble detection method and tumble detection equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111046701A true CN111046701A (en) | 2020-04-21 |
Family
ID=70229836
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811191665.XA Pending CN111046701A (en) | 2018-10-12 | 2018-10-12 | Tumble detection method and tumble detection equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111046701A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030058341A1 (en) * | 2001-09-27 | 2003-03-27 | Koninklijke Philips Electronics N.V. | Video based detection of fall-down and other events |
| CN103186902A (en) * | 2011-12-29 | 2013-07-03 | 爱思开电讯投资（中国）有限公司 | Trip detecting method and device based on video |
| CN104317199A (en) * | 2014-09-16 | 2015-01-28 | 江苏大学 | Mobile smart housekeeper |
| CN104581082A (en) * | 2015-01-29 | 2015-04-29 | 深圳市中兴移动通信有限公司 | Home monitoring system and home monitoring method |
| CN104794463A (en) * | 2015-05-11 | 2015-07-22 | 华东理工大学 | System and method for achieving indoor human body falling detection based on Kinect |
| CN105448041A (en) * | 2016-01-22 | 2016-03-30 | 苏州望湖房地产开发有限公司 | A human body falling intelligent control system and method |
| CN106358021A (en) * | 2016-11-01 | 2017-01-25 | 成都宏软科技实业有限公司 | Smart community monitoring system |
| CN107085922A (en) * | 2017-05-25 | 2017-08-22 | 湖北酷焰智能科技有限公司 | Tumble alarm device and the tumble alarm method applied to the tumble alarm device |
| CN107220604A (en) * | 2017-05-18 | 2017-09-29 | 清华大学深圳研究生院 | A kind of fall detection method based on video |
-
2018
- 2018-10-12 CN CN201811191665.XA patent/CN111046701A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030058341A1 (en) * | 2001-09-27 | 2003-03-27 | Koninklijke Philips Electronics N.V. | Video based detection of fall-down and other events |
| CN103186902A (en) * | 2011-12-29 | 2013-07-03 | 爱思开电讯投资（中国）有限公司 | Trip detecting method and device based on video |
| CN104317199A (en) * | 2014-09-16 | 2015-01-28 | 江苏大学 | Mobile smart housekeeper |
| CN104581082A (en) * | 2015-01-29 | 2015-04-29 | 深圳市中兴移动通信有限公司 | Home monitoring system and home monitoring method |
| CN104794463A (en) * | 2015-05-11 | 2015-07-22 | 华东理工大学 | System and method for achieving indoor human body falling detection based on Kinect |
| CN105448041A (en) * | 2016-01-22 | 2016-03-30 | 苏州望湖房地产开发有限公司 | A human body falling intelligent control system and method |
| CN106358021A (en) * | 2016-11-01 | 2017-01-25 | 成都宏软科技实业有限公司 | Smart community monitoring system |
| CN107220604A (en) * | 2017-05-18 | 2017-09-29 | 清华大学深圳研究生院 | A kind of fall detection method based on video |
| CN107085922A (en) * | 2017-05-25 | 2017-08-22 | 湖北酷焰智能科技有限公司 | Tumble alarm device and the tumble alarm method applied to the tumble alarm device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110287923B (en) | Human body posture acquisition method, device, computer equipment and storage medium | |
| Tabar et al. | Smart home care network using sensor fusion and distributed vision-based reasoning | |
| CN108009466B (en) | Pedestrian detection method and device | |
| CN111104910B (en) | Garbage delivery behavior supervision method and related products | |
| CN102750709A (en) | Method and device for detecting fight by using video | |
| CN110685119B (en) | Washing machine and system and method for guaranteeing safety of children | |
| CN114469076B (en) | Identity-feature-fused fall identification method and system for solitary old people | |
| CN111814510A (en) | Detection method and device for remnant body | |
| CN112949417A (en) | Tumble behavior identification method, equipment and system | |
| CN113657150A (en) | Fall detection method and device and computer readable storage medium | |
| CN115393563A (en) | Package detection method and system and electronic equipment | |
| CN110276310A (en) | Smog of smoking determines method and device, storage medium, electronic device | |
| CN113615166B (en) | Accident detection device and accident detection method | |
| CN113361469A (en) | Method, device and equipment for identifying coverage state and storage medium | |
| CN102801964A (en) | Intelligent home monitoring method based on motion detection and system thereof | |
| CN111046701A (en) | Tumble detection method and tumble detection equipment | |
| CN110505438A (en) | A kind of acquisition methods and video camera of data queued | |
| CN106710142A (en) | Fall down behavior visual inspection system facing smart phone | |
| Boudouane et al. | Wearable camera for fall detection embedded system | |
| US20040237096A1 (en) | Automated in-home observation of user interactions with video devices | |
| KR20170141049A (en) | System and method for event alarm based on metadata and application therefor | |
| JP2020102657A (en) | Electronic apparatus, control unit, control method, and control program | |
| CN109040604A (en) | Shoot processing method, device, storage medium and the mobile terminal of image | |
| JP7241011B2 (en) | Information processing device, information processing method and program | |
| CN117935371A (en) | Night starting behavior distinguishing method, device, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |