CN107610135A - Quick finger dividing method based on geodesic curve distance algorithm and morphological recognition - Google Patents
Quick finger dividing method based on geodesic curve distance algorithm and morphological recognition Download PDFInfo
- Publication number
- CN107610135A CN107610135A CN201710862159.8A CN201710862159A CN107610135A CN 107610135 A CN107610135 A CN 107610135A CN 201710862159 A CN201710862159 A CN 201710862159A CN 107610135 A CN107610135 A CN 107610135A
- Authority
- CN
- China
- Prior art keywords
- finger
- pixel
- point
- geodesic curve
- hand
- 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
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of quick finger dividing method based on geodesic curve distance algorithm and morphological recognition, and using the depth image of gesture as input, hand Segmentation is carried out with the Threshold segmentation in depth and principal direction;And for motion blur and sensor noise, carry out the filtering of noise and smooth with medium filtering and closing operation of mathematical morphology;After being pre-processed, by image binaryzation, potential finger tip point is found apart from extreme value using the geodesic curve in the image FX after binaryzation for plane finger tip;Judge to differentiate true finger tip finally by morphologic feature, and the region for meeting feature is designated as finger pixel, complete finger segmentation.The present invention is found potential finger tip point using geodesic curve extreme value, the quick detection and finger split plot design of true finger tip is identified using morphological feature, blocked for finger part, finger identification and segmentation under the situation such as different background or illumination, there is very strong success rate, and can reach the effect of Real-time segmentation.
Description
Technical field:
The present invention relates to a kind of gesture processing method, more particularly to a kind of based on geodesic curve distance algorithm and morphological recognition
Quick finger dividing method.
Background technology:
With the rise of the new technologies such as VR, AR, the research based on Consumer's Experience turns into the method that all trades and professions ensure customer relationship
One of, man-machine interaction Consumer's Experience three-dimensional animation modeling in, it is desirable to the Gesture Recognition pinpoint accuracy of view-based access control model and
High efficiency.The detection and extraction of finger tip are a kind of solution methods conventional, technology is common.Traditional two dimensional image is based on the colour of skin
Gesture identification it is serious and can not solve the problems, such as that most of finger blocks by such environmental effects such as background and illumination.It is and near
The research that the depth information based on Kinect risen in year carries out gesture identification can improve this situation, but this method calculates
Amount is big, it is impossible to reaches good real-time.
The content of the invention:
The technical problems to be solved by the invention are:Overcome the deficiencies in the prior art, there is provided a kind of efficiently rapid, success rate it is high and
The quick finger dividing method based on geodesic curve distance algorithm and morphological recognition of Real-time segmentation effect can be reached.
The technical scheme is that:
1st, a kind of quick finger dividing method based on geodesic curve distance algorithm and morphological recognition, it is characterized in that:Using gesture
Depth image as input, with depth and principal direction Threshold segmentation carry out hand Segmentation;And for motion blur and
Sensor noise, the filtering of noise and smooth is carried out with medium filtering and closing operation of mathematical morphology;After being pre-processed, by image
Binaryzation, potential finger tip is found apart from extreme value using the geodesic curve in the image FX after binaryzation for plane finger tip
Point;Judge to differentiate true finger tip finally by morphologic feature, and the region for meeting feature is designated as finger pixel, complete hand
Refer to segmentation.
2nd, the quick finger segmentation side according to claim 1 based on geodesic curve distance algorithm and morphological recognition
Method, it is characterized in that:The step of hand Segmentation is:Since a point a nearest from video camera, obtained hand maximum picture will be tested
Plain depth β is assigned to 0 as segmentation threshold with pixel value of a points pixel difference more than β;Such operation is based on a hypothesis:Hand
It is the object nearest from video camera, such hypothesis is also commonly used for gesture identification, and after such operation, still having can
Arm pixel can be left, remaining pixel is considered as the point cloud in three dimensions further to analyze its feature;Use principal component
Analysis PCA algorithms calculate the principal direction of remaining point cloud, then the length β of a hand is partitioned into from a points along principal direction, so
Can completes eliminating for most of non-hand pixel;Image after so treating is referred to as hand images.
3rd, the quick finger segmentation side according to claim 1 based on geodesic curve distance algorithm and morphological recognition
Method, it is characterized in that:Due to the characteristic of depth image, finger-image can show projection feature, employ one it is more quick
Detection algorithm obtain potential finger tip:The characteristics of finger tip is the human body end points on two dimensional surface, according to based on geodesic curve away from
Detection algorithm from extreme point screening human body end points can obtain potential finger tip pixel, make use of improved double scannings
Euclidean distance transform algorithms, to try to achieve hand images all pixels point to the geodesic curve of extreme value point set
Distance, and constantly expand extreme value point set with recursive mode, so as to identify potential finger tip point.
4th, the quick finger segmentation side according to claim 3 based on geodesic curve distance algorithm and morphological recognition
Method, it is characterized in that:Algorithm steps are as follows:
(1)The hand images of segmentation portion are subjected to binaryzation, and does first expansion post-etching and operates to obtain a UNICOM domain;
(2)The substantially skeletal graph of hand is obtained using distance transform algorithms, and using extreme point as in morphology
The heart, this morphology center are initial extreme value point set;
(3)With geodesic distance transform algorithms calculate hand pixel to extreme value point set geodesic curve away from
From obtaining a geodesic curve distance map;
(4)Using the maximum point of geodesic curve distance in geodesic curve distance map as new extreme point, and this point is added to extreme value
In point set;
(5)Repeat above-mentioned(3)(4)Step, constantly update geodesic curve distance map;
By this processing, it is already possible to find most potential finger tip point, in order to reach high-efficient simple, take and detect
The first seven as finger tip point to be screened.
5th, the quick finger segmentation side according to claim 1 based on geodesic curve distance algorithm and morphological recognition
Method, it is characterized in that:Identify true finger tip and split finger pixel:Real finger is characterized in tall and thin, therefore utilizes that to limit finger tip attached
The length-width ratio of nearly pixel is judged;After carrying out binaryzation to hand images, respectively using each finger tip point as set of assigned points, together
Sample can obtain the geodesic curve distance map of each potential finger tip, Ran Houjing with geodesic distance transform algorithms
Cross limiting distance maximum and be worth to the pixel for belonging to the potential finger of this root, and this threshold value can be changed and obtain finger pixel
" growth ";During this growth, the length and width and length-width ratio of this section of pixel are constantly estimated, and by itself and general finger
Relevant parameter is compared, you can judges whether this section of pixel is real finger.
6th, the quick finger segmentation side according to claim 5 based on geodesic curve distance algorithm and morphological recognition
Method, it is characterized in that:Calculated as follows on long and wide estimation:
(1)It is long:The bianry image for the potential finger separated is subjected to center bone line drawing, in order to reduce algorithm complexity
, bone line drawing still uses the non-iterative thinning algorithm based on distance transform algorithms, then calculates bone
The length of bone line;
(2)It is wide:By the contour line extraction of potential finger out and compared with the contour line of whole hand images, the point of coincidence is gone
Fall, then computational length.
According to experiment test, it is limited in wide in the range of 2.5-8.0, is limited in 12.0-25.0 by long, length-width ratio limitation
More accurate result can be obtained in the range of 2.0-4.0, unit is the length in pixel meaning, meets the one of this limitation
Section pixel will be denoted as finger, and will not screen potential finger tip point again in the pixel for be designated as finger.
The beneficial effects of the invention are as follows:
1st, depth image data input microprocessor is contrasted and judged by gesture collecting device by the present invention, image up to standard
Data are shown by display device, can intuitively show gesture and the form of finger tip, accuracy height.
2nd, the present invention is for the different size of finger that can be identified on two dimensional surface, except partially due to motion blur and company
Outside finger together, there is very high success rate;And processing time is averagely no more than 2ms, there is very high processing speed.
3rd, the present invention makes simulation wire size be converted into data signal by A/D converter, is easy to microprocessor to carry out at data
Reason, efficiency high.
4th, present invention employs depth information as inputting, using pixel of the gesture after Threshold segmentation as hand figure
Picture, potential finger tip can be detected by geodesic curve distance algorithm by pretreated image, and where finger tip point investigates it
The length and width and length-width ratio of finger, finally screening obtain real finger tip and mark finger where it, and the degree of accuracy is high.
5th, the present invention has very strong robustness and very for the partial occlusion problem, different backgrounds or illumination of finger
Efficiently rapid, success rate is high, can reach the effect of Real-time segmentation, and its is applied widely, has good economic effect after popularization
Benefit.
Brief description of the drawings:
Fig. 1 is the binary map after the hand Segmentation of the first gesture in finger dividing method of the present invention;
Fig. 2 is the geodesic curve distance map after the hand Segmentation of the first gesture shown in Fig. 1;
Fig. 3 is the finger schematic diagram being partitioned into after the hand Segmentation of the first gesture shown in Fig. 1;
Fig. 4 is the binary map after the hand Segmentation of second of gesture in finger dividing method of the present invention;
Fig. 5 is the geodesic curve distance map after the hand Segmentation of second of gesture shown in Fig. 4;
Fig. 6 is the finger schematic diagram being partitioned into after the hand Segmentation of second of gesture shown in Fig. 4;
Fig. 7 is the binary map after the hand Segmentation of the third gesture in finger dividing method of the present invention;
Fig. 8 is the geodesic curve distance map after the hand Segmentation of the third gesture shown in Fig. 7;
Fig. 9 is the finger schematic diagram being partitioned into after the hand Segmentation of the third gesture shown in Fig. 7;
Figure 10 is the binary map after the hand Segmentation of the 4th kind of gesture in finger dividing method of the present invention;
Figure 11 is the geodesic curve distance map after the hand Segmentation of the 4th kind of gesture shown in Figure 10;
Figure 12 is the finger schematic diagram being partitioned into after the hand Segmentation of the 4th kind of gesture shown in Figure 10;
Figure 13 is the binary map after the hand Segmentation of the 5th kind of gesture in finger dividing method of the present invention;
Figure 14 is the geodesic curve distance map after the hand Segmentation of the 5th kind of gesture shown in Figure 13;
Figure 15 is the finger schematic diagram being partitioned into after the hand Segmentation of the 5th kind of gesture shown in Figure 13;
Figure 16 is the binary map after the hand Segmentation of the 6th kind of gesture in finger dividing method of the present invention;
Figure 17 is the geodesic curve distance map after the hand Segmentation of the 6th kind of gesture shown in Figure 16;
Figure 18 is the finger schematic diagram being partitioned into after the hand Segmentation of the 6th kind of gesture shown in Figure 16.
Embodiment:
Embodiment:Referring to Fig. 1-Figure 18.
Quick finger dividing method based on geodesic curve distance algorithm and morphological recognition, its technical scheme are:Using hand
The depth image of gesture carries out hand Segmentation as input with the Threshold segmentation in depth and principal direction;And for motion blur
And sensor noise, carry out the filtering of noise and smooth with medium filtering and closing operation of mathematical morphology;After being pre-processed, it will scheme
As binaryzation, potential finger is found apart from extreme value using the geodesic curve in the image FX after binaryzation for plane finger tip
Cusp;Judge to differentiate true finger tip finally by morphologic feature, and the region for meeting feature is designated as finger pixel, complete
Finger is split.
The step of hand Segmentation is:Since a point a nearest from video camera, the hand maximum pixel that experiment is obtained is deep
β is spent as segmentation threshold, and 0 is all assigned to pixel value of a points pixel difference more than β;Such operation is based on a hypothesis:Hand be from
The nearest object of video camera, such hypothesis are also commonly used for gesture identification, and after such operation, in this case it is still possible to stay
Lower arms pixel, remaining pixel is considered as the point cloud in three dimensions further to analyze its feature;Use principal component analysis
PCA algorithms calculate the principal direction of remaining point cloud, then the length β of a hand is partitioned into from a points along principal direction, thus may be used
To complete eliminating for most of non-hand pixel;Image after so treating is referred to as hand images.
Due to the characteristic of depth image, finger-image can show the feature of projection, employ a more quickly inspection
Method of determining and calculating obtains potential finger tip:The characteristics of finger tip is the human body end points on two dimensional surface, according to based on geodesic curve apart from pole
The detection algorithm of value point screening human body end points can obtain potential finger tip pixel, make use of improved double scanning euclidean
Distance transform algorithms, to try to achieve hand images all pixels point to the geodesic curve distance of extreme value point set, it is used in combination
Recursive mode constantly expands extreme value point set, so as to identify potential finger tip point.
Algorithm steps are as follows:
(1)The hand images of segmentation portion are subjected to binaryzation, and does first expansion post-etching and operates to obtain a UNICOM domain;
(2)The substantially skeletal graph of hand is obtained using distance transform algorithms, and using extreme point as in morphology
The heart, this morphology center are initial extreme value point set;
(3)With geodesic distance transform algorithms calculate hand pixel to extreme value point set geodesic curve away from
From obtaining a geodesic curve distance map;
(4)Using the maximum point of geodesic curve distance in geodesic curve distance map as new extreme point, and this point is added to extreme value
In point set;
(5)Repeat above-mentioned(3)(4)Step, constantly update geodesic curve distance map;
By this processing, it is already possible to find most potential finger tip point, in order to reach high-efficient simple, take and detect
The first seven as finger tip point to be screened.
Identify true finger tip and split finger pixel:Real finger be characterized in it is tall and thin, therefore utilize limit finger tip nearby picture
The length-width ratio of element is judged;After carrying out binaryzation to hand images, respectively using each finger tip point as set of assigned points, equally may be used
To obtain the geodesic curve distance map of each potential finger tip with geodesic distance transform algorithms, then by limit
System distance is maximum to be worth to the pixel for belonging to the potential finger of this root, and can change this threshold value and finger pixel is obtained " life
It is long ";During this growth, the length and width and length-width ratio of this section of pixel are constantly estimated, and it is corresponding to general finger
Parameter is compared, you can judges whether this section of pixel is real finger.
Calculated as follows on long and wide estimation:
(3)It is long:The bianry image for the potential finger separated is subjected to center bone line drawing, in order to reduce algorithm complexity
, bone line drawing still uses the non-iterative thinning algorithm based on distance transform algorithms, then calculates bone
The length of bone line;
(4)It is wide:By the contour line extraction of potential finger out and compared with the contour line of whole hand images, the point of coincidence is gone
Fall, then computational length.
According to experiment test, it is limited in wide in the range of 2.5-8.0, is limited in 12.0-25.0 by long, length-width ratio limitation
More accurate result can be obtained in the range of 2.0-4.0, unit is the length in pixel meaning, meets the one of this limitation
Section pixel will be denoted as finger, and will not screen potential finger tip point again in the pixel for be designated as finger.
The depth image data of collection is transferred to microprocessor, microprocessor and memory by gesture collecting device of the present invention
Connection, microprocessor is split depth image data and comparison is handled, and the view data up to standard isolated is shown by gesture
Show that equipment is shown, microprocessor and gesture display device are connected with power supply.
Preferably:Microprocessor is connected with A/D converter, and depth image data is transferred directly to micro- place when being data signal
Device is managed, depth image data first passes through converter conversion and is transmitted further to microprocessor when being analog signal.Microprocessor is
STM32F103VC, gesture collecting device are video camera or projecting apparatus, and gesture display device is liquid crystal display.
The present invention is found potential finger tip point using geodesic curve extreme value, the quick inspection of true finger tip is identified using morphological feature
Survey and finger segmentation, block, finger identification and segmentation under the situation such as different background or illumination, have very strong for finger part
Success rate, and can reach the effect of Real-time segmentation.
The above described is only a preferred embodiment of the present invention, any formal limitation not is made to the present invention, it is all
It is any simple modification, equivalent change and modification made according to the technical spirit of the present invention to above example, still falls within
In the range of technical solution of the present invention.
Claims (6)
1. a kind of quick finger dividing method based on geodesic curve distance algorithm and morphological recognition, it is characterized in that:Using gesture
Depth image as input, with depth and principal direction Threshold segmentation carry out hand Segmentation;And for motion blur and
Sensor noise, the filtering of noise and smooth is carried out with medium filtering and closing operation of mathematical morphology;After being pre-processed, by image
Binaryzation, potential finger tip is found apart from extreme value using the geodesic curve in the image FX after binaryzation for plane finger tip
Point;Judge to differentiate true finger tip finally by morphologic feature, and the region for meeting feature is designated as finger pixel, complete hand
Refer to segmentation.
2. the quick finger dividing method according to claim 1 based on geodesic curve distance algorithm and morphological recognition, its
It is characterized in:The step of hand Segmentation is:Since a point a nearest from video camera, the hand maximum pixel that experiment is obtained is deep
β is spent as segmentation threshold, and 0 is all assigned to pixel value of a points pixel difference more than β;Such operation is based on a hypothesis:Hand be from
The nearest object of video camera, such hypothesis are also commonly used for gesture identification, and after such operation, in this case it is still possible to stay
Lower arms pixel, remaining pixel is considered as the point cloud in three dimensions further to analyze its feature;Use principal component analysis
PCA algorithms calculate the principal direction of remaining point cloud, then the length β of a hand is partitioned into from a points along principal direction, thus may be used
To complete eliminating for most of non-hand pixel;Image after so treating is referred to as hand images.
3. the quick finger dividing method according to claim 1 based on geodesic curve distance algorithm and morphological recognition, its
It is characterized in:Due to the characteristic of depth image, finger-image can show the feature of projection, employ a more quickly detection
Algorithm obtains potential finger tip:The characteristics of finger tip is the human body end points on two dimensional surface, according to based on geodesic curve apart from extreme value
The detection algorithm of point screening human body end points can obtain potential finger tip pixel, make use of improved double scanning euclidean
Distance transform algorithms, to try to achieve hand images all pixels point to the geodesic curve distance of extreme value point set, it is used in combination
Recursive mode constantly expands extreme value point set, so as to identify potential finger tip point.
4. the quick finger dividing method according to claim 3 based on geodesic curve distance algorithm and morphological recognition, its
It is characterized in:Algorithm steps are as follows:
(1)The hand images of segmentation portion are subjected to binaryzation, and does first expansion post-etching and operates to obtain a UNICOM domain;
(2)The substantially skeletal graph of hand is obtained using distance transform algorithms, and using extreme point as in morphology
The heart, this morphology center are initial extreme value point set;
(3)With geodesic distance transform algorithms calculate hand pixel to extreme value point set geodesic curve away from
From obtaining a geodesic curve distance map;
(4)Using the maximum point of geodesic curve distance in geodesic curve distance map as new extreme point, and this point is added to extreme value
In point set;
(5)Repeat above-mentioned(3)(4)Step, constantly update geodesic curve distance map;
By this processing, it is already possible to find most potential finger tip point, in order to reach high-efficient simple, take and detect
The first seven as finger tip point to be screened.
5. the quick finger dividing method according to claim 1 based on geodesic curve distance algorithm and morphological recognition, its
It is characterized in:Identify true finger tip and split finger pixel:Real finger be characterized in it is tall and thin, therefore utilize limit finger tip nearby pixel
Length-width ratio judged;, equally can be with respectively using each finger tip point as set of assigned points after carrying out binaryzation to hand images
The geodesic curve distance map of each potential finger tip is obtained with geodesic distance transform algorithms, then by limitation
Distance is maximum to be worth to the pixel for belonging to the potential finger of this root, and can change this threshold value and finger pixel is obtained " life
It is long ";During this growth, the length and width and length-width ratio of this section of pixel are constantly estimated, and it is corresponding to general finger
Parameter is compared, you can judges whether this section of pixel is real finger.
6. the quick finger dividing method according to claim 5 based on geodesic curve distance algorithm and morphological recognition, its
It is characterized in:Calculated as follows on long and wide estimation:
(1) it is long:The bianry image for the potential finger separated is subjected to center bone line drawing, in order to reduce algorithm complexity
, bone line drawing still uses the non-iterative thinning algorithm based on distance transform algorithms, then calculates bone
The length of bone line;
(2) it is wide:By the contour line extraction of potential finger out and compared with the contour line of whole hand images, the point of coincidence is gone
Fall, then computational length;
According to experiment test, it is limited in wide in the range of 2.5-8.0, length is limited in 12.0-25.0, length-width ratio is limited in
More accurate result can be obtained in the range of 2.0-4.0, unit is the length in pixel meaning, meets one section of this limitation
Pixel will be denoted as finger, and will not screen potential finger tip point again in the pixel for be designated as finger.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710862159.8A CN107610135A (en) | 2017-09-21 | 2017-09-21 | Quick finger dividing method based on geodesic curve distance algorithm and morphological recognition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710862159.8A CN107610135A (en) | 2017-09-21 | 2017-09-21 | Quick finger dividing method based on geodesic curve distance algorithm and morphological recognition |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107610135A true CN107610135A (en) | 2018-01-19 |
Family
ID=61061846
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710862159.8A Pending CN107610135A (en) | 2017-09-21 | 2017-09-21 | Quick finger dividing method based on geodesic curve distance algorithm and morphological recognition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107610135A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108985191A (en) * | 2018-06-28 | 2018-12-11 | 广东技术师范学院 | A kind of contour extraction method based on mobile device gesture identification |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102622606A (en) * | 2010-02-03 | 2012-08-01 | 北京航空航天大学 | Human skeleton extraction and orientation judging method based on geodesic model |
CN103544472A (en) * | 2013-08-30 | 2014-01-29 | Tcl集团股份有限公司 | Processing method and processing device based on gesture images |
-
2017
- 2017-09-21 CN CN201710862159.8A patent/CN107610135A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102622606A (en) * | 2010-02-03 | 2012-08-01 | 北京航空航天大学 | Human skeleton extraction and orientation judging method based on geodesic model |
CN103544472A (en) * | 2013-08-30 | 2014-01-29 | Tcl集团股份有限公司 | Processing method and processing device based on gesture images |
Non-Patent Citations (2)
Title |
---|
孙骁: "基于深度图像的实时鲁棒手势跟踪研究", 《中国优秀硕士学位论文全文数据库 信息科技辑(月刊)》 * |
李长龙: "基于Kinect深度图像的手势识别研究", 《中国优秀硕士学位论文全文数据库 信息科技辑(月刊)》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108985191A (en) * | 2018-06-28 | 2018-12-11 | 广东技术师范学院 | A kind of contour extraction method based on mobile device gesture identification |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021208275A1 (en) | Traffic video background modelling method and system | |
CN106570486B (en) | Filtered target tracking is closed based on the nuclear phase of Fusion Features and Bayes's classification | |
CN109086724B (en) | Accelerated human face detection method and storage medium | |
CN109815865B (en) | Water level identification method and system based on virtual water gauge | |
CN107403436B (en) | Figure outline rapid detection and tracking method based on depth image | |
Oprisescu et al. | Automatic static hand gesture recognition using tof cameras | |
US20180122083A1 (en) | Method and device for straight line detection and image processing | |
CN110334762B (en) | Feature matching method based on quad tree combined with ORB and SIFT | |
CN108229475B (en) | Vehicle tracking method, system, computer device and readable storage medium | |
CN111160291B (en) | Human eye detection method based on depth information and CNN | |
CN109685045A (en) | A kind of Moving Targets Based on Video Streams tracking and system | |
CN106815578A (en) | A kind of gesture identification method based on Depth Motion figure Scale invariant features transform | |
CN108710879B (en) | Pedestrian candidate region generation method based on grid clustering algorithm | |
CN111027370A (en) | Multi-target tracking and behavior analysis detection method | |
CN106023249A (en) | Moving object detection method based on local binary similarity pattern | |
Yogeswaran et al. | 3d surface analysis for automated detection of deformations on automotive body panels | |
CN107610135A (en) | Quick finger dividing method based on geodesic curve distance algorithm and morphological recognition | |
JP2018180646A (en) | Object candidate area estimation device, object candidate area estimation method and object candidate area estimation program | |
Kuang et al. | An effective skeleton extraction method based on Kinect depth image | |
Getahun et al. | A robust lane marking extraction algorithm for self-driving vehicles | |
CN113689365B (en) | Target tracking and positioning method based on Azure Kinect | |
Wang et al. | Skin Color Weighted Disparity Competition for Hand Segmentation from Stereo Camera. | |
Varkonyi-Koczy | Fuzzy logic supported corner detection | |
JP2008084109A (en) | Eye opening/closing determination device and eye opening/closing determination method | |
CN111798506A (en) | Image processing method, control method, terminal and computer readable storage 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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180119 |