CN110334576A - A kind of hand method for tracing and device - Google Patents
A kind of hand method for tracing and device Download PDFInfo
- Publication number
- CN110334576A CN110334576A CN201910359991.5A CN201910359991A CN110334576A CN 110334576 A CN110334576 A CN 110334576A CN 201910359991 A CN201910359991 A CN 201910359991A CN 110334576 A CN110334576 A CN 110334576A
- Authority
- CN
- China
- Prior art keywords
- video frame
- hand
- hand images
- detection
- images
- 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.)
- Granted
Links
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
- G06V40/107—Static hand or arm
-
- 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
- G06V40/107—Static hand or arm
- G06V40/117—Biometrics derived from hands
Abstract
Aspect of the invention is related to the present invention relates to technical field of hand gesture recognition, more particularly to a kind of hand method for tracing and device.Hand method for tracing includes: to obtain video frame step, obtains continuous video frame；Judgment step judges whether detect hand images in previous video frame according to the mark for whether having hand images in previous video frame；Single scale part detecting step has hand images according in previous video frame, carries out hand detection to the subrange of current video frame, and mark position of the hand images in current video frame；Multiple dimensioned global detection step carries out hand detection to current video frame full figure range, and mark position of the hand images in current video frame according to not having hand images in previous video frame.It by using the hand position in two kinds of detection mode detection videos and is labeled, for tracking the continuous motion profile of hand images, helps to improve hand tracking efficiency, reduce the difficulty of hand detection difficulty and real time execution.
Description
Technical field
This invention relates generally to technical field of hand gesture recognition, more particularly to a kind of hand method for tracing and device.
Background technique
Gesture identification is not by the mainstay of the oncontacting human-computer interaction of the mechanical equipments such as touch screen, and people can be used
Simple gesture is controlled or is interacted with equipment, allows the behavior of the computer understanding mankind, is that important in man-machine interaction mode is ground
Study carefully one of direction.
However the continuous of hand position determines that most difficult in i.e. hand tracking exactly gesture identification is also most time-consuming link.
It is different from generic object detection and tracking, the difficulty of hand tracking is mainly reflected in three aspects in gesture identification: in a first aspect,
The deformability of hand is high, and detection difficulty is big；Second aspect, movement speed of the hand in camera view is fast, and tracking difficulty is big；
The third aspect, computing resource is limited, and data run difficulty is big.
Summary of the invention
In order to solve the above-mentioned problems in the prior art, the present invention provides a kind of hand method for tracing and device.
In a first aspect, the embodiment of the present invention provides a kind of hand method for tracing, wherein include: to obtain video frame step, obtain
Take continuous video frame, including previous video frame and current video frame；Whether judgment step has hand according in previous video frame
The mark of image judges whether detect hand images in previous video frame；Single scale part detecting step, when judging previous view
There are hand images in frequency frame, then by single scale, locally detection carries out hand detection to the subrange of current video frame, and
Position of the one or more hand images detected in current video frame is marked in current video frame；Multiple dimensioned global inspection
Step is surveyed, when judging do not have hand images in previous video frame, then by multiple dimensioned global detection to current video frame full figure
Range carries out hand detection, and the one or more hand images detected are marked in current video frame in current video frame
Position.
In one embodiment, hand method for tracing further include: selection main body hand images step is locally examined based on single scale
Survey one or more hand images that step or multiple dimensioned global detection step detect, select one of hand images as
Main body hand images, and delete the mark of remaining hand images.
In another embodiment, main body hand images step is selected further include: according to multiple hand images in current video
Size in frame selects the maximum hand images of area as main body hand images.
In another embodiment, main body hand images step is selected further include: match in selection video frame with default gesture
Identical hand images are as main body hand images.
In one embodiment, hand method for tracing further include: hard recognition step, based on one in current video frame or
Multiple hand images judge whether hand images are true hand, are such as judged as not it is true hand, then delete hand images
Mark.
In one embodiment, hard recognition step further include: according to the position of hand images, current video frame is cut
It takes, obtains interception image, interception image is zoomed in a fixed identification size range and is judged.
In one embodiment, single scale part detecting step further include: the position based on hand images in previous video frame,
Expansion interception is carried out to current video frame, obtains the topography of subrange, hand detection is carried out in topography.
In another embodiment, single scale part detecting step further include: topography is zoomed in and out to a fixation
Hand detection is carried out within the scope of detecting size.
Second aspect, the embodiment of the present invention provide a kind of hand follow-up mechanism, wherein include: to obtain video frame module, use
In the continuous video frame of acquisition, including previous video frame and current video frame；Judgment module, for being according in previous video frame
The no mark for having hand images judges whether detect hand images in previous video frame；Single scale part detection module, is used for
When judging that there are hand images in previous video frame, hand detection is carried out to the subrange of current video frame, and working as forward sight
Position of the one or more hand images detected in current video frame is marked in frequency frame；Multiple dimensioned global detection module,
For carrying out hand detection to current video frame full figure range, and working as when judging do not have hand images in previous video frame
Position of the one or more hand images detected in current video frame is marked in preceding video frame.
The third aspect, the embodiment of the present invention provide a kind of electronic equipment, wherein electronic equipment includes: memory, for depositing
Storage instruction；And processor, the hand method for tracing of the instruction execution first aspect for calling memory to store.
Fourth aspect, the embodiment of the present invention provide a kind of computer readable storage medium, wherein computer-readable storage medium
Matter is stored with computer executable instructions, and computer executable instructions when executed by the processor, execute the hand of first aspect
Method for tracing.
A kind of hand method for tracing and device provided by the invention are locally detected and the multiple dimensioned overall situation by using single scale
The hand position in two kinds of detection mode detection videos is detected, and the hand images that will test out are labeled, for tracking hand
Image continuous motion profile in portion's helps to improve hand tracking efficiency, reduces the difficulty of hand detection difficulty and real time execution.
Detailed description of the invention
The following detailed description is read with reference to the accompanying drawings, above-mentioned and other purposes, the feature of embodiment of the present invention
It will become prone to understand with advantage.In the accompanying drawings, several implementations of the invention are shown by way of example rather than limitation
Mode, in which:
Fig. 1 shows a kind of hand method for tracing schematic diagram provided in an embodiment of the present invention；
Fig. 2 shows another hand method for tracing schematic diagrames provided in an embodiment of the present invention；
Fig. 3 shows another hand method for tracing schematic diagram provided in an embodiment of the present invention；
Fig. 4 shows another hand method for tracing schematic diagram provided in an embodiment of the present invention；
Fig. 5 shows a kind of hand follow-up mechanism schematic diagram provided in an embodiment of the present invention；
Fig. 6 shows a kind of electronic equipment schematic diagram provided in an embodiment of the present invention；
In the accompanying drawings, identical or corresponding label indicates identical or corresponding part.
Specific embodiment
The principle and spirit of the invention are described below with reference to several illustrative embodiments.It should be appreciated that providing this
A little embodiments are used for the purpose of making those skilled in the art can better understand that realizing the present invention in turn, and be not with any
Mode limits the scope of the invention.
Although being noted that the statements such as " first " used herein, " second " to describe embodiments of the present invention not
Same module, step and data etc., still the statement such as " first ", " second " is merely in different modules, step and data etc.
Between distinguish, and be not offered as specific sequence or significance level.In fact, the statements such as " first ", " second " are complete
It may be used interchangeably.
Fig. 1 is the flow diagram of one embodiment of hand method for tracing 10.As shown in Figure 1, the embodiment method packet
It includes: obtaining video frame step 110, judgment step 120, single scale part detecting step 130, multiple dimensioned global detection step 140.
Each step in Fig. 1 is described in detail below.
Video frame step 110 is obtained, continuous video frame, including previous video frame and current video frame are obtained.
In the present embodiment, video frame can carry out obtaining real-time continuous video frame, e.g., mobile phone by image capture device
Camera, computer camera can also obtain continuous video by carrying out transferring one section of video in local data base or cloud
Frame.In one example, acquisition image is carried out using image capture device, opens preview video stream, obtain real-time video frame.Another
In example, video frame is obtained from the video stored in local data base or cloud.For acquisition continuous video frame into
The tracking of row hand.
Judgment step 120, according to the mark for whether having hand images in previous video frame, judge in previous video frame whether
Detect hand images.
In the present embodiment, according in the previous video frame of acquisition whether with hand images mark as judge according to
According to judging in previous video frame whether there is hand images.It selects different detection means to be detected according to judging result, drops
Low detection difficulty.
Single scale part detecting step 130 then passes through single scale part when judging there is hand images in previous video frame
It detects and hand detection is carried out to the subrange of current video frame, and mark the one or more detected in current video frame
Position of the hand images in current video frame.
Locally detection is able to detect a certain range of scale i.e. hand images of pixel to single scale.In the present embodiment,
Current video frame is detected according to the result judged in previous video frame with hand images.Mould is detected by using single scale hand
Type is locally detected to local single scale is carried out in current video frame, and wherein single scale hand detection model can be based on convolution mind
Target detection model through network participates in the jump of hand frame only relatively fixed range in pixel scale of the model training
It is dynamic, such as: 56X56 to 96X96, therefore the prototype network structure is simple, it is few to be related to detection data amount, can quickly detect currently
Whether there is hand images in the subrange of video frame, if detecting hand images in subrange, by rectangular or
All hand images that the positions such as person's circle frame will test on current video frame are labeled；If hand figure is not detected
Picture, then without mark.Image of the current video frame after single scale locally detection as next video frame judgment basis, when
The image of preceding video frame after tested be next video frame previous video frame, for being judged, next video frame according to
The result selection detection mode of image after judging current video frame detection.Speed is detected using single scale part detection mode
Fastly, real time execution difficulty is reduced.
Multiple dimensioned global detection step 140, when judging do not have hand images in previous video frame, then by multiple dimensioned complete
Office's detection carries out hand detection to current video frame full figure range, and the one or more detected is marked in current video frame
Position of the hand images in current video frame.
Multiple dimensioned global detection is able to detect the hand images of all scales in video frame.In the present embodiment, according to sentencing
Result in previous video frame of breaking without hand images detects current video frame, can detect mould by using multiple dimensioned hand
Type carries out carrying out the multiple dimensioned global detection of full figure range to current video frame, wherein multiple dimensioned hand detection model can be and be based on
The target detection model of convolutional neural networks, the hand frame for participating in the model training vary in pixel scale, that is, have big
Hand also has small hand, therefore has the ability for the hand that can handle a variety of scales simultaneously.It is detected by multiple dimensioned hand detection model
Whether contain hand images in current video frame full figure range, if detecting hand images, the side of passing through in current video frame
All hand images that the positions such as shape or circle frame will test on current video frame are labeled；If in current video frame
In hand images are not detected, then without mark.Image of the current video frame after multiple dimensioned global detection is as next
Video frame judgment basis, the image of current video frame after tested is the previous video frame of next video frame, for being sentenced
Disconnected, next video frame selects detection mode according to the result of the image after judging current video frame detection.By using multiple dimensioned
Global detection can more fully detect the whole hand images contained in current video frame.In one example, because of the company of acquisition
In continuous video frame, first frame video frame does not have previous video frame, therefore default judges the previous video of first video frame
Frame does not have hand images, and the first video frame is detected by multiple dimensioned global detection.
Above-described embodiment detects hand position in video and right by the way that the different detection mode of complexity is used alternatingly
It is labeled, and for tracking the continuous motion profile of hand images, helps to improve hand tracking efficiency, reduces hand detection
The difficulty of difficulty and real time execution.
Fig. 2 shows the flow diagrams of another embodiment of hand method for tracing 10.As shown in Fig. 2, the embodiment
Method further include: selection main body hand images step 150 is based on single scale part detecting step 130 or multiple dimensioned global detection
One or more hand images that step 140 detects, select one of hand images as main body hand images, and delete
The mark of remaining hand images.
In some cases, by the detection of single scale part or multiple dimensioned global detection, in current video frame simultaneously
Multiple hand images are marked out, such as occur multiple hand images in the subrange that single scale locally detects, or in more rulers
It spends in the video frame full figure of global detection and multiple hand images occurs, for the ease of tracking hand images, select current video
A hand in frame retains the position frame in main body hand images and is tracked as main body hand images, deletes remaining hand
The mark of portion's image avoids interference tracking main body hand images.
In one embodiment, main body hand images step 150 is selected further include: according to multiple hand images in current video
Size in frame selects the maximum hand images of area as main body hand images.By single scale locally detection or
Multiple dimensioned global detection marks out multiple hand images simultaneously in current video frame, it is generally the case that the maximum hand of area is
Hand to be detected, the small hand of area is background, therefore selects the maximum hand images of hand images area in current video frame
It is tracked as main body hand images, can satisfy usual demand, and can be convenient for observation hand images fortune by the method
Dynamic trail change, facilitates tracking.
In another embodiment, select main body hand images step 150 further include: selection video frame in default gesture
With identical hand images as main body hand images.Select to match identical hand images with preset gesture as master
Body hand images are simultaneously tracked, and are conducive in the application for needing gesture control, accurate to track required hand images, are mentioned
High control efficiency.Such as: a kind of coded lock is unlocked according to one section of preset gesture transformation, and coded lock is in scanning unlock
In the process, there is the poster with hand images in picture background, and selection matches identical in the video frame with preset gesture
Hand images are tracked as main body hand images.
Fig. 3 and Fig. 4 shows the flow diagram of other embodiments of hand method for tracing 10.Such as Fig. 3 and Fig. 4 institute
Show, the embodiment method further include: hard recognition step 160 is sentenced based on one or more hand images in current video frame
Whether disconnected hand images are true hand, are such as judged as not it is true hand, then delete the mark of hand images.By using hand
Part class model carries out hard recognition, and wherein hand disaggregated model can be the disaggregated model based on convolutional neural networks, be used for
Identify single scale locally detection or multiple dimensioned global detection to hand images whether be true manpower, if it is true
Manpower then retains the mark of hand images in video frame, if not true manpower, then deletes the mark of hand images.It is logical
Hard recognition is crossed, the artificial hand for helping avoid keeping track error detection falls into endless loop.
In one example, hard recognition step is for before selecting main body hand step, identifying single scale locally detection or more
Scale global detection to all hand images whether be true manpower, will not be that true manpower excludes, delete
The hand images of artificial hand mark, and only retain true hand images, the number of hand images in current video frame is if identifying
Zero, without true manpower, then determine that current video frame does not have hand images.By hard recognition, be conducive in selection main body
During hand, the selection time is reduced, improves hard recognition accuracy rate.
In another example, hard recognition step detects main body hand images for facilitating after selecting main body hand step
It whether is true hand images.When identifying that main body hand is not true manpower, then determine that current video frame does not have hand images.
By being identified to main body hand images, avoids being tracked artificial hand, improve hand tracking efficiency and accuracy.
In one embodiment, hard recognition step 160 further include: according to the position of hand images, to current video frame into
Row interception, obtains interception image, zooms in a fixed identification size range to interception image and judges.Hard recognition is
It is completed by hand disaggregated model, interception image is zoomed to the identification size range of unified fixation, such as: by interception image
Pixel zoom in 56*56 to 64*64 pixel coverage, facilitate save hand classification time, improve hard recognition efficiency.
In one embodiment, single scale part detecting step 130 further include: the position based on hand images in previous video frame
It sets, expansion interception is carried out to current video frame, obtains the topography of subrange, hand detection is carried out in topography.
According to the position that previous video frame hand images mark, the topography of current video frame is intercepted, because in video, the short time
Mobile variation can occur for interior hand, but the moving distance of hand is limited again in the short time, so in interception current video frame
Before topography, first carries out the range of topography to be extended to single scale and locally detect the preset range being able to detect that, into
Row interception, the position after detecting the mobile variation of hand in current video frame, such as: the range 128*128 of default interception image, on
The picture size of main body hand in the video frame is 56*56 in one video frame, then 128*128 progress is expanded in current video frame
Interception, avoids and detects to whole figure, save detection time, helps to improve tracking efficiency, and it is difficult to reduce real time execution
Degree.
In another embodiment, single scale part detecting step 130 further include: topography is zoomed in and out to one and is fixed
Detecting size range carry out hand detection.Topography after interception is uniformly zoomed to the detecting size model of same fixation
It encloses, for example, the pixel of topography is zoomed in 56*56 to 96*96 pixel coverage, the present embodiment makes hand after interception
Topography in pixel size it is relatively fixed, facilitate further mitigate detection pressure, save detection time.
Fig. 5 shows the exemplary structure schematic diagram of hand follow-up mechanism 20.As shown in figure 5, the embodiment hand is tracked
Device includes: to obtain video frame module 210, for obtaining continuous video frame, including previous video frame and current video frame；Sentence
Disconnected module 220, for judging whether detect in previous video frame according to the mark for whether having hand images in previous video frame
Hand images；Single scale part detection module 230 judges there is hand images in previous video frame for working as, to current video
The subrange of frame carries out hand detection, and the one or more hand images detected are marked in current video frame current
Position in video frame；Multiple dimensioned global detection module 240 judges not having in previous video frame hand images for working as, right
Current video frame full figure range carries out hand detection, and the one or more hand images detected are marked in current video frame
Position in current video frame.
In one embodiment, hand follow-up mechanism further include: selection main body hand images module, for being based on single scale office
One or more hand images that portion's detection module or multiple dimensioned global detection module detect, select one of hand images
As main body hand images, and delete the mark of remaining hand images.
In one embodiment, selection main body hand images module is also used to according to multiple hand images in current video frame
Size, select the maximum hand images of area as main body hand images.
In one embodiment, hand follow-up mechanism further include: hard recognition module, for one in current video frame or
Multiple hand images judge whether hand images are true hand, are such as judged as not it is true hand, then delete hand images
Mark.
In one embodiment, hard recognition module is also used to the position according to hand images, cuts to current video frame
It takes, obtains interception image, interception image is zoomed in a fixed identification size range and is judged.
In one embodiment, single scale part detection module is also used to the position according to hand images in previous video frame,
Expansion interception is carried out to current video frame, obtains the topography of subrange, hand detection is carried out in topography.
In one embodiment, single scale part detection module is also used to zoom in and out topography to a fixed detection
Hand detection is carried out in size range.
The function that modules in device are realized is corresponding with the step in method as described above, specific implementation
The description for method and step above is referred to technical effect, details are not described herein.
As shown in fig. 6, an embodiment of the invention provides a kind of electronic equipment 30.Wherein, the electronic equipment 30
Including memory 310, processor 320, input/output (Input/Output, I/O) interface 330.Wherein, memory 310 are used
In store instruction.Processor 320, the instruction execution embodiment of the present invention for calling memory 310 to store are chased after for hand
Track method.Wherein, processor 320 is connect with memory 310, I/O interface 330 respectively, for example, can by bus system and/or its
He is attached bindiny mechanism's (not shown) of form.Memory 310 can be used for storing program and data, including the present invention is implemented
Program involved in example for hand tracking, processor 320 are stored in the program of memory 310 by operation thereby executing electricity
The various function application and data processing of sub- equipment 30.
Processor 320 can use digital signal processor (Digital Signal in the embodiment of the present invention
Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), programmable patrol
At least one of volume array (Programmable Logic Array, PLA) example, in hardware realizes, the processor 320
It can be central processing unit (Central Processing Unit, CPU) or there is data-handling capacity and/or instruction
The combination of one or more of the processing unit of other forms of executive capability.
Memory 310 in the embodiment of the present invention may include one or more computer program products, the computer
Program product may include various forms of computer readable storage mediums, such as volatile memory and/or non-volatile deposit
Reservoir.The volatile memory for example may include random access memory (Random Access Memory, RAM) and/
Or cache memory (cache) etc..The nonvolatile memory for example may include read-only memory (Read-Only
Memory, ROM), flash memory (Flash Memory), hard disk (Hard Disk Drive, HDD) or solid state hard disk
(Solid-State Drive, SSD) etc..
In the embodiment of the present invention, I/O interface 330 can be used for receiving input instruction (such as number or character information, and
Generate key signals input related with the user setting of electronic equipment 30 and function control etc.), it can also be output to the outside various
Information (for example, image or sound etc.).In the embodiment of the present invention I/O interface 330 may include physical keyboard, function button (such as
Volume control button, switch key etc.), mouse, operating stick, trace ball, microphone, one in loudspeaker and touch panel etc.
It is a or multiple.
In some embodiments, the present invention provides a kind of computer readable storage medium, the computer-readable storages
Media storage has computer executable instructions, and computer executable instructions when executed by the processor, execute described above appoint
Where method.
Although description operation in a particular order in the accompanying drawings should not be construed as requiring specific shown in
Sequence or serial order operate to execute these operations, or shown in requirement execution whole to obtain desired result.?
In specific environment, multitask and parallel processing be may be advantageous.
Methods and apparatus of the present invention can be completed using standard programming technology, using rule-based logic or its
His logic realizes various method and steps.It should also be noted that herein and the terms used in the claims " device "
" module " is intended to include using the realization of a line or multirow software code and/or hardware realization and/or for receiving input
Equipment.
One or more combined individually or with other equipment can be used in any step, operation or program described herein
A hardware or software module are executed or are realized.In one embodiment, it includes packet tool computer program that software module, which uses,
The computer program product of the computer-readable medium of code is realized, can be executed by computer processor any for executing
Or whole described step, operation or programs.
For the purpose of example and description, the preceding description that the present invention is implemented is had been presented for.Preceding description is not poor
Also not the really wanting of act property limits the invention to exact form disclosed, according to the above instruction there is likely to be various modifications and
Modification, or various changes and modifications may be obtained from the practice of the present invention.Select and describe these embodiments and be in order to
Illustrate the principle of the present invention and its practical application, so that those skilled in the art can be to be suitable for the special-purpose conceived
Come in a variety of embodiments with various modifications and utilize the present invention.
Claims (11)
1. a kind of hand method for tracing, wherein include:
Video frame step is obtained, continuous video frame, including previous video frame and current video frame are obtained；
Judgment step, according to the mark for whether having hand images in the previous video frame, judge be in the previous video frame
It is no to detect hand images；
Single scale part detecting step then passes through single ruler when judging there is the hand images in the previous video frame
The detection of degree part carries out hand detection to the subrange of the current video frame, and detection is marked in the current video frame
To position of the one or more hand images in the current video frame；
Multiple dimensioned global detection step, when judging do not have the hand images in the previous video frame, then by described more
Scale global detection carries out hand detection to the current video frame full figure range, and detection is marked in the current video frame
To position of the one or more hand images in the current video frame.
2. according to the method described in claim 1, wherein, the method also includes:
Main body hand images step is selected, based on single scale part detecting step or the multiple dimensioned global detection step inspection
The one or more hand images measured, select one of them described hand images as main body hand images, and delete
The mark of remaining hand images.
3. according to the method described in claim 2, wherein, the selection main body hand images step further include: according to multiple institutes
Size of the hand images in the current video frame is stated, selects the maximum hand images of area as the main body
Hand images.
4. according to the method described in claim 2, wherein, the selection main body hand images step further include: select the view
The identical hand images are matched as the main body hand images with default gesture in frequency frame.
5. method according to any one of claims 1 to 4, wherein the method also includes: hard recognition step is based on
Hand images described in one or more of described current video frame judge whether the hand images are true hand, are such as sentenced
Break not to be true hand, then deletes the mark of the hand images.
6. according to the method described in claim 5, wherein, the hard recognition step further include: according to the hand images
The position intercepts the current video frame, obtains interception image, zooms to a fixed knowledge to the interception image
Judged in other size range.
7. according to the method described in claim 1, wherein, single scale part detecting step further include: based on described previous
The position of hand images described in video frame carries out expansion interception to the current video frame, obtains the office of the subrange
Portion's image carries out the hand detection in the topography.
8. according to the method described in claim 7, wherein, single scale part detecting step further include: to the Local map
The hand detection is carried out as zooming in and out within the scope of the detecting size fixed to one.
9. a kind of hand follow-up mechanism, wherein include:
Video frame module is obtained, for obtaining continuous video frame, including previous video frame and current video frame；
Judgment module, for judging the previous video frame according to the mark for whether having hand images in the previous video frame
In whether detect hand images；
Single scale part detection module judges there is the hand images in the previous video frame for working as, to described current
The subrange of video frame carries out hand detection, and the one or more hand that mark detects in the current video frame
Portion's image is in the position of the current video frame；
Multiple dimensioned global detection module, for working as to described when judging do not have the hand images in the previous video frame
Preceding video frame full figure range carries out hand detection, and the one or more hand that mark detects in the current video frame
Portion's image is in the position of the current video frame.
10. a kind of electronic equipment, wherein the electronic equipment includes:
Memory, for storing instruction；And
Processor, for calling the instruction execution hand tracking of any of claims 1-8 of the memory storage
Method.
11. a kind of computer readable storage medium, wherein the computer-readable recording medium storage has computer is executable to refer to
It enables, when executed by the processor, perform claim requires hand tracking described in any one of 1-8 to the computer executable instructions
Method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910359991.5A CN110334576B (en) | 2019-04-30 | 2019-04-30 | Hand tracking method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910359991.5A CN110334576B (en) | 2019-04-30 | 2019-04-30 | Hand tracking method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110334576A true CN110334576A (en) | 2019-10-15 |
| CN110334576B CN110334576B (en) | 2021-09-24 |
Family
ID=68139906
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910359991.5A Active CN110334576B (en) | 2019-04-30 | 2019-04-30 | Hand tracking method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110334576B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112068699A (en) * | 2020-08-31 | 2020-12-11 | 北京市商汤科技开发有限公司 | Interaction method, interaction device, electronic equipment and storage medium |
| CN112818320A (en) * | 2021-02-26 | 2021-05-18 | 上海交通大学 | Smartphone pattern password conjecture method based on video |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100027845A1 (en) * | 2008-07-31 | 2010-02-04 | Samsung Electronics Co., Ltd. | System and method for motion detection based on object trajectory |
| CN102722886A (en) * | 2012-05-21 | 2012-10-10 | 浙江捷尚视觉科技有限公司 | Video speed measurement method based on three-dimensional calibration and feature point matching |
| CN103839036A (en) * | 2012-11-22 | 2014-06-04 | 富士通株式会社 | Hand tracking method and device |
| CN104408760A (en) * | 2014-10-28 | 2015-03-11 | 燕山大学 | Binocular-vision-based high-precision virtual assembling system algorithm |
| CN107239727A (en) * | 2016-12-07 | 2017-10-10 | 北京深鉴智能科技有限公司 | Gesture identification method and system |
| CN109410249A (en) * | 2018-11-13 | 2019-03-01 | 深圳龙岗智能视听研究院 | A kind of method for tracing of combination depth characteristic and Freehandhand-drawing feature adaptive targets |
-
2019
- 2019-04-30 CN CN201910359991.5A patent/CN110334576B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100027845A1 (en) * | 2008-07-31 | 2010-02-04 | Samsung Electronics Co., Ltd. | System and method for motion detection based on object trajectory |
| CN102722886A (en) * | 2012-05-21 | 2012-10-10 | 浙江捷尚视觉科技有限公司 | Video speed measurement method based on three-dimensional calibration and feature point matching |
| CN103839036A (en) * | 2012-11-22 | 2014-06-04 | 富士通株式会社 | Hand tracking method and device |
| CN104408760A (en) * | 2014-10-28 | 2015-03-11 | 燕山大学 | Binocular-vision-based high-precision virtual assembling system algorithm |
| CN107239727A (en) * | 2016-12-07 | 2017-10-10 | 北京深鉴智能科技有限公司 | Gesture identification method and system |
| CN109410249A (en) * | 2018-11-13 | 2019-03-01 | 深圳龙岗智能视听研究院 | A kind of method for tracing of combination depth characteristic and Freehandhand-drawing feature adaptive targets |
Non-Patent Citations (3)
| Title |
|---|
| LIU ZHENYU 等: "CU Partition Mode Decision for HEVC Hardwired Intra Encoder Using Convolution Neural Network", 《IEEE TRANSACTIONS ON IMAGE PROCESSING》 * |
| SHIYANG YAN 等: "Multi-Scale Convolutional Neural Networks for Hand Detection", 《APPLIED COMPUTATIONAL INTELLIGENCE AND SOFT COMPUTING》 * |
| 杨燕平: "基于深度学习的手势跟踪方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112068699A (en) * | 2020-08-31 | 2020-12-11 | 北京市商汤科技开发有限公司 | Interaction method, interaction device, electronic equipment and storage medium |
| CN112818320A (en) * | 2021-02-26 | 2021-05-18 | 上海交通大学 | Smartphone pattern password conjecture method based on video |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110334576B (en) | 2021-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103324937B (en) | The method and apparatus of label target | |
| US10902056B2 (en) | Method and apparatus for processing image | |
| CN103630244B (en) | Power system device method for diagnosing faults and system | |
| CN110189285A (en) | A kind of frames fusion method and device | |
| CN110163248B (en) | Visualization method, visualization device, computer equipment and storage medium for model evaluation | |
| CN108932053A (en) | Drawing practice, device, storage medium and computer equipment based on gesture | |
| CN103383731A (en) | Projection interactive method and system based on fingertip positioning and computing device | |
| CN109460267A (en) | Mobile robot offline map saves and real-time method for relocating | |
| CN108762505A (en) | Virtual object control method, device, storage medium based on gesture and equipment | |
| CN110210285A (en) | Face tracking method, face tracking device and computer storage medium | |
| Sarkar et al. | Dynamic hand gesture recognition for mobile systems using deep LSTM | |
| CN110334576A (en) | A kind of hand method for tracing and device | |
| CN110443241A (en) | Car license recognition model training method, licence plate recognition method and device | |
| CN109190648A (en) | Simulated environment generation method, device, mobile terminal and computer-readable storage medium | |
| CN102799273B (en) | Interaction control system and method | |
| CN106537423A (en) | Adaptive featurization as service | |
| US11715260B2 (en) | Offline teaching device, measurement control device, and program | |
| Zeng et al. | A novel tensor decomposition-based efficient detector for low-altitude aerial objects with knowledge distillation scheme | |
| CN111814733A (en) | Concentration degree detection method and device based on head posture | |
| CN110334716A (en) | Characteristic pattern processing method, image processing method and device | |
| CN110046277A (en) | More video merging mask methods and device | |
| CN111652168B (en) | Group detection method, device, equipment and storage medium based on artificial intelligence | |
| CN109785362A (en) | Target object tracking, device and storage medium based on target object detection | |
| CN110084157A (en) | The data processing method and device that image identifies again | |
| CN103309434A (en) | Instruction identification method and electronic equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |