CN108520264A - A kind of hand contour feature optimization method based on depth image - Google Patents

A kind of hand contour feature optimization method based on depth image Download PDF

Info

Publication number
CN108520264A
CN108520264A CN201810285718.8A CN201810285718A CN108520264A CN 108520264 A CN108520264 A CN 108520264A CN 201810285718 A CN201810285718 A CN 201810285718A CN 108520264 A CN108520264 A CN 108520264A
Authority
CN
China
Prior art keywords
profile
iteration
hand
polygon
hand region
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
Application number
CN201810285718.8A
Other languages
Chinese (zh)
Inventor
刘佩林
金珂
葛昊
应忍冬
邹耀
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Digital Intelligent Technology Co Ltd
Original Assignee
Shanghai Digital Intelligent Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shanghai Digital Intelligent Technology Co Ltd filed Critical Shanghai Digital Intelligent Technology Co Ltd
Priority to CN201810285718.8A priority Critical patent/CN108520264A/en
Publication of CN108520264A publication Critical patent/CN108520264A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/181—Segmentation; Edge detection involving edge growing; involving edge linking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/66—Analysis of geometric attributes of image moments or centre of gravity
    • 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
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20068—Projection on vertical or horizontal image axis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20112—Image segmentation details
    • G06T2207/20164—Salient point detection; Corner detection

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Geometry (AREA)
  • Human Computer Interaction (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a kind of hand contour feature optimization method based on depth image, it is characterized in that：This method comprises the following steps：(1) depth image is read in, the hand region figure of redundancy profile is pre-processed and obtain including；(2) primary iteration point and iteration radius are calculated, primary iteration point is close proximity to finger tip；(3) it is iterated using mean shift algorithm, finds the border circular areas closest to palm, return to the center of circle and radius；(4) hand region profile is updated, redundancy profile is rejected, accurate hand region profile is obtained；(5) the external convex closure profile of hand region is obtained using convex closure detection algorithm, and combines the Hu of Internal periphery and outer convex closure not bending moment information, obtain 14 dimension Outline Feature Vectors.The optimization method can effectively remove other redundancy profile informations such as forearm, coat-sleeve using the hand contour feature optimization of average drifting and convex closure detection algorithm, improve the stability of final gesture recognition system.

Description

A kind of hand contour feature optimization method based on depth image
Technical field
The present invention relates to computer vision and human-computer interaction technique field, specially a kind of hand based on depth image Contour feature optimization method.
Background technology
The Chinese invention patent of Patent No. 201510282688.1 discloses a kind of hand-characteristic based on depth image Point detecting method comprising following steps：(1) hand Segmentation：Human motion video sequence is collected using Kinect to extract Hand obtains human hands location information by depth image using Openni, by setting region of search and depth threshold side Method, it is preliminary to obtain palm of the hand point；Utilize the find_contours function calls contouring in one's hands of Opencv；By finding in handwheel exterior feature The maximum inscribed circle center of circle, it is accurate to determine hand palm of the hand point, by calculating all hand internal points to the most short distance between profile point From m, the hand internal point representated by the maximizing M in the shortest distance, M is palm of the hand point, inradius R=M；(2) Feature point extraction：Gaussian smoothing is carried out by continuous opponent's contouring, and combines curvature threshold to obtain CSS curvature charts, root Hand finger tip point is obtained according to CSS edge analysis limiting value in figure and refers to valley point coordinate, while needing completion according to CSS curvature charts The hand being unable to get refers to valley point；(3) completion lacks finger, by angle threshold and depth jump in conjunction in the way of come completion Finger is lacked, to find the finger tip point of bending finger.
However, it is this based on setting region of search and depth threshold method extraction hand contour feature can be attached to forearm, The profile of other objects such as coat-sleeve, these redundancy profiles can cause contour feature extraction second-rate, in feature vector process In be not inconsistent with expected feature, cause the accuracy of identification of final gesture recognition system relatively low.
For this purpose, applicant carried out beneficial exploration and trial, result of the above problems is had found, will be detailed below The technical solution of introduction generates in this background.
Invention content
The purpose of the present invention is to provide a kind of hand contour feature optimization method based on depth image, in solution State the problem of being proposed in background technology.
To achieve the above object, the present invention provides the following technical solutions：A kind of hand contour feature based on depth image Optimization method, this method comprises the following steps：
(6) depth image is read in, the hand region figure of redundancy profile is pre-processed and obtain including；
(7) primary iteration point and iteration radius are calculated, primary iteration point is close proximity to finger tip；
(8) it is iterated using mean shift algorithm, finds the border circular areas closest to palm, return to the center of circle and radius；
(9) hand region profile is updated, redundancy profile is rejected, accurate hand region profile is obtained；
(10) the external convex closure profile of hand region is obtained using convex closure detection algorithm, and combines Internal periphery and outer convex closure Hu not bending moment information, obtain 14 dimension Outline Feature Vectors.
Preferably, the hand contour feature optimization method based on depth image, it is deep in the step (1) Degree image pre-processing module carries out threshold cutting to depth image, graphics filters and calculates largest connected region, including following Step：
(a) hand region is obtained using depth threshold, and maps that bianry image, including redundancy profile, before white Scape indicates hand region, and background is black；
(b) graphics operation is utilized, opening operation is first done, smoothed image profile simultaneously removes ambient noise, then does closed operation, Fill the said minuscule hole in target；
(c) maximum area profile is found, and thinks that the profile is hand region profile, including redundancy profile, fills up the wheel Hole in exterior feature.
Preferably, the hand contour feature optimization method based on depth image, in the step (2), including The step of primary iteration point and iteration radius are chosen from the hand region tentatively extracted, which includes the following steps：
(a) by the hand region profile of extraction with Polygons Representation, and the polygon situation containing inner ring is repaired；
(b) minimum enclosed rectangle of polygon is calculated, and is compared with image boundary, the number of edges overlapped according to the two Carry out following Taxonomic discussion：
If (a) hand is excessively close from camera lens, image can not show that complete hand region, algorithm terminate；
If (b) two sides overlapped are parallel edges, show that hand laterally or longitudinally runs through camera lens, image can not have been shown Whole hand region, algorithm terminate；
If (c) intersecting with image boundary without arm segment profile, redundancy profile is not present, and what is returned at this time initially changes Generation point is the barycenter of hand region, and primary iteration radius is chosen according to practical palm empirical value；
If (d) two sides overlapped are intersection edges, in four vertex of the minimum enclosed rectangle of polygon, calculate The vertex nearest apart from polygon, and to ensure that the vertex is effective, do not intersect with image boundary, then the top is calculated Subpoint of the point on polygon, i.e., the nearest point in the distance vertex on polygon；
Finally take the line midpoint of subpoint and polygon barycenter as primary iteration point, the half conduct of wire length Primary iteration radius, if primary iteration point in outside of polygon, takes subpoint of this on polygon as newly initial Iteration point.
Preferably, the hand contour feature optimization method based on depth image also wraps in the step (3) The iterative process based on average drifting is included, which includes following sub-step
(a) according to primary iteration point c0With primary iteration radius r0Obtain the prime area of iteration circle C；
(b) the intersecting area I of iteration circle C and hand region polygon P is found, and calculates the barycenter c in the regioni；
(c) compare barycenter ciWith the center of circle c of iteration circle S34 is entered step if the two differs by more than iteration threshold；If The two differs in iteration threshold, then enters step (e)；
(d), the center of circle c of adjustment iteration circle is the barycenter c of intersecting areai, radius r is the barycenter c of intersecting areaiIn hand The borderline projections of area polygonal P, and return to step (b)；
(e), if S (I)/S (C) ＞ 1.1sth, then increase iteration radius of circle r, and return to the step (2)；If S (I)/ S (C) ＜ 0.9sth, then reduce iteration radius of circle r, and return to step (b)；Otherwise iteration, changing at the end of output iteration are terminated The center location c and radius r, above-mentioned S (I) of Dai Yuan, S (C) is respectively intersecting area and the iteration area of a circle, sthFor significant surface Product pixel threshold.
Preferably, the hand contour feature optimization method based on depth image also wraps in the step (4) The process that hand region profile is updated according to the iteration result of average drifting is included, which includes following sub-step：
(a) according to the intersection situation of iteration circle and hand region polygon, hand region polygon is divided into intersecting area I and disjoint rangeWherein disjoint range Q may be by multiple independent polygon { q1, q2..., qnConstitute 's；
(b) the independent polygon q of each of disjoint range Q are directed to, the length for overlapping line segment d of q and image boundary are calculated Degree.IfThen q is wiped out in former hand region polygon Pi, and enter step S44；IfThen enter Step S43；Above-mentioned dthLine segment threshold value is overlapped for image boundary；
(c) polygon q independent to each of disjoint range Q, calculates the barycenter c of qq, it is with the center of circle c that iteration is justified Starting point makees extended line cq until image boundary, calculates the length for overlapping line segment m of q and cq extended lines.IfThen q is wiped out in former hand region polygon Pi；
(d) judge whether current hand region polygon P includes the polygon segments of multiple separation, if so, by it The middle maximum polygon P of areamaxAs finally obtained accurate hand region Pa, and return to PaProfile as accurate hand Profile.
Preferably, the hand contour feature optimization method based on depth image combines in the step (5) The Outline Feature Vector process of convex closure detection, which includes following sub-step：
(a) Graham scanning methods is used to calculate the external convex closure profile of accurate hand region；
(b) it calculates the 7 of accurate hand region profile and ties up Hu not bending moment vectors；
(c) it calculates the 7 of external convex closure profile and ties up Hu not bending moment vectors；
(d) the hand contour feature of 14 dimension Hu invariant rectangles Cheng Xin of combination hand region profile and external convex closure profile Vector.
Compared with prior art, the beneficial effects of the invention are as follows：The hand contour feature optimization side based on depth image Method utilizes the hand contour feature of average drifting and convex closure detection algorithm compared to traditional hand contour feature extraction algorithm Optimization can effectively remove other redundancy profile informations such as forearm, coat-sleeve, ensure wheel as far as possible during feature vector The accuracy of wide feature and comprehensive more accurately enters data source for providing for subsequent classification learning, improves The stability of final gesture recognition system.
Description of the drawings
Fig. 1 is overall algorithm flow chart of the present invention.
Fig. 2 is the depth image pretreatment process figure of the present invention.
Fig. 3 is the flow chart of calculating the primary iteration point and primary iteration radius of the present invention.
Fig. 4 is the flow chart that palm area is found according to mean shift algorithm iteration of the present invention.
Fig. 5 is the newer flow chart of hand region profile of the present invention.
Fig. 6 is the flow chart of the Outline Feature Vector process of the combination convex closure detection of the present invention.
Specific implementation mode
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete Site preparation describes, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its His embodiment, shall fall within the protection scope of the present invention.
- Fig. 6 is please referred to Fig.1, the present invention provides a kind of technical solution：
Embodiment 1：
A kind of hand contour feature optimization method based on depth image, this method comprises the following steps：
(11) depth image is read in, the hand region figure of redundancy profile is pre-processed and obtain including；
(12) primary iteration point and iteration radius are calculated, primary iteration point is close proximity to finger tip；
(13) it is iterated using mean shift algorithm, finds the border circular areas closest to palm, return to the center of circle and half Diameter；
(14) hand region profile is updated, redundancy profile is rejected, accurate hand region profile is obtained；
(15) the external convex closure profile of hand region is obtained using convex closure detection algorithm, and combines Internal periphery and outer convex closure Hu not bending moment information, obtain 14 dimension Outline Feature Vectors.
Embodiment 2：
According to the hand contour feature optimization method described in embodiment 1 based on depth image, in the step (1) In, depth image preprocessing module carries out threshold cutting to depth image, graphics filters and calculate largest connected region, wraps Include following steps：
(a) hand region is obtained using depth threshold, and maps that bianry image, including redundancy profile, before white Scape indicates hand region, and background is black；
(b) graphics operation is utilized, opening operation is first done, smoothed image profile simultaneously removes ambient noise, then does closed operation, Fill the said minuscule hole in target；
(c) maximum area profile is found, and thinks that the profile is hand region profile, including redundancy profile, fills up the wheel Hole in exterior feature.
Embodiment 3；
The hand contour feature optimization method based on depth image according to embodiment 1 or 2, in the step (2) In, include the steps that selection primary iteration point and iteration radius, the step include following step from the hand region tentatively extracted Suddenly：
(a) by the hand region profile of extraction with Polygons Representation, and the polygon situation containing inner ring is repaired；
(b) minimum enclosed rectangle of polygon is calculated, and is compared with image boundary, the number of edges overlapped according to the two Carry out following Taxonomic discussion：
If (a) hand is excessively close from camera lens, image can not show that complete hand region, algorithm terminate；
If (b) two sides overlapped are parallel edges, show that hand laterally or longitudinally runs through camera lens, image can not have been shown Whole hand region, algorithm terminate；
If (c) intersecting with image boundary without arm segment profile, redundancy profile is not present, and what is returned at this time initially changes Generation point is the barycenter of hand region, and primary iteration radius is chosen according to practical palm empirical value；
If (d) two sides overlapped are intersection edges, in four vertex of the minimum enclosed rectangle of polygon, calculate The vertex nearest apart from polygon, and to ensure that the vertex is effective, do not intersect with image boundary, then the top is calculated Subpoint of the point on polygon, i.e., the nearest point in the distance vertex on polygon；
Finally take the line midpoint of subpoint and polygon barycenter as primary iteration point, the half conduct of wire length Primary iteration radius, if primary iteration point in outside of polygon, takes subpoint of this on polygon as newly initial Iteration point.
Embodiment 4：
The hand contour feature optimization method based on depth image according to embodiment 1 or 2 or 3, in the step (3) further include the iterative process based on average drifting in, which includes following sub-step
(a) according to primary iteration point c0With primary iteration radius r0Obtain the prime area of iteration circle C；
(b) the intersecting area I of iteration circle C and hand region polygon P is found, and calculates the barycenter c in the regioni；
(c) compare barycenter ciWith the center of circle c of iteration circle S34 is entered step if the two differs by more than iteration threshold；If The two differs in iteration threshold, then enters step (e)；
(d), the center of circle c of adjustment iteration circle is the barycenter c of intersecting areai, radius r is the barycenter c of intersecting areaiIn hand The borderline projections of area polygonal P, and return to step (b)；
(e), if S (I)/S (C) ＞ 1.1sth, then increase iteration radius of circle r, and return to the step (2)；If S (I)/ S (C) ＜ 0.9sth, then reduce iteration radius of circle r, and return to step (b)；Otherwise iteration, changing at the end of output iteration are terminated The center location c and radius r, above-mentioned S (I) of Dai Yuan, S (C) is respectively intersecting area and the iteration area of a circle, sthFor significant surface Product pixel threshold.
Embodiment 5：
The hand contour feature optimization method based on depth image according to embodiment 1 or 2 or 3 or 4, described Further include the process that hand region profile is updated according to the iteration result of average drifting, which includes following in step (4) Sub-step：
(a) according to the intersection situation of iteration circle and hand region polygon, hand region polygon is divided into intersecting area I and disjoint rangeWherein disjoint range Q may be by multiple independent polygon { q1, q2..., qnConstitute 's；
(b) the independent polygon q of each of disjoint range Q are directed to, the length for overlapping line segment d of q and image boundary are calculated Degree.IfThen q is wiped out in former hand region polygon Pi, and enter step S44；IfThen enter Step S43；Above-mentioned dthLine segment threshold value is overlapped for image boundary；
(c) polygon q independent to each of disjoint range Q, calculates the barycenter c of qq, it is with the center of circle c that iteration is justified Point makees extended line cq until image boundary, calculates the length for overlapping line segment m of q and cq extended lines.IfThen Q is wiped out in former hand region polygon Pi；
(d) judge whether current hand region polygon P includes the polygon segments of multiple separation, if so, by it The middle maximum polygon P of areamaxAs finally obtained accurate hand region Pa, and return to PaProfile as accurate hand Profile.
Embodiment 6：
The hand contour feature optimization method based on depth image according to embodiment 1 or 2 or 3 or 4 or 5, in institute The Outline Feature Vector process that convex closure detection is combined in step (5) is stated, which includes following sub-step：
(a) Graham scanning methods is used to calculate the external convex closure profile of accurate hand region；
(b) it calculates the 7 of accurate hand region profile and ties up Hu not bending moment vectors；
(c) it calculates the 7 of external convex closure profile and ties up Hu not bending moment vectors；
(d) the hand contour feature of 14 dimension Hu invariant rectangles Cheng Xin of combination hand region profile and external convex closure profile Vector.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with Understanding without departing from the principles and spirit of the present invention can carry out these embodiments a variety of variations, modification, replace And modification, the scope of the present invention is defined by the appended.

Claims (6)

1. a kind of hand contour feature optimization method based on depth image, it is characterized in that：This method comprises the following steps：
(1) depth image is read in, the hand region figure of redundancy profile is pre-processed and obtain including；
(2) primary iteration point and iteration radius are calculated, primary iteration point is close proximity to finger tip；
(3) it is iterated using mean shift algorithm, finds the border circular areas closest to palm, return to the center of circle and radius；
(4) hand region profile is updated, redundancy profile is rejected, accurate hand region profile is obtained；
(5) the external convex closure profile of hand region is obtained using convex closure detection algorithm, and combines the Hu of Internal periphery and outer convex closure not Bending moment information obtains 14 dimension Outline Feature Vectors.
2. the hand contour feature optimization method according to claim 1 based on depth image, it is characterized in that：Described In step (1), depth image preprocessing module carries out threshold cutting to depth image, graphics filters and calculates largest connected area Domain includes the following steps：
(a) hand region is obtained using depth threshold, and maps that bianry image, including redundancy profile, white Foreground table Show that hand region, background are black；
(b) graphics operation is utilized, opening operation is first done, smoothed image profile simultaneously removes ambient noise, then does closed operation, fills mesh Said minuscule hole in mark；
(c) maximum area profile is found, and thinks that the profile is hand region profile, including redundancy profile, is filled up in the profile Hole.
3. the hand contour feature optimization method according to claim 1 based on depth image, it is characterized in that：In the step Suddenly include the steps that primary iteration point and iteration radius are chosen from the hand region tentatively extracted in (2), the step include with Lower step：
(a) by the hand region profile of extraction with Polygons Representation, and the polygon situation containing inner ring is repaired；
(b) minimum enclosed rectangle of polygon is calculated, and is compared with image boundary, is carried out such as according to the number of edges that the two overlaps Lower Taxonomic discussion：
If (a) hand is excessively close from camera lens, image can not show that complete hand region, algorithm terminate；
If (b) two sides overlapped are parallel edges, show that hand laterally or longitudinally runs through camera lens, image can not show complete hand Portion region, algorithm terminate；
If (c) intersecting with image boundary without arm segment profile, redundancy profile is not present, and the primary iteration point returned at this time is The barycenter of hand region, primary iteration radius are chosen according to practical palm empirical value；
If (d) two sides overlapped are intersection edges, in four vertex of the minimum enclosed rectangle of polygon, it is more to calculate distance The nearest vertex of side shape, and to ensure that the vertex is effective, do not intersect with image boundary, then the vertex is calculated polygon Subpoint in shape, i.e., the nearest point in the distance vertex on polygon；
Finally take the line midpoint of subpoint and polygon barycenter as primary iteration point, the half of wire length changes as initial For radius, if primary iteration point in outside of polygon, takes subpoint of this on polygon as new primary iteration point.
4. the hand contour feature optimization method according to claim 1 based on depth image, it is characterized in that：In the step Suddenly further include the iterative process based on average drifting in (3), which includes following sub-step
(a) according to primary iteration point c0With primary iteration radius r0Obtain the prime area of iteration circle C；
(b) the intersecting area I of iteration circle C and hand region polygon P is found, and calculates the barycenter c in the regioni；
(c) compare barycenter ciWith the center of circle c of iteration circle S34 is entered step if the two differs by more than iteration threshold；If the two phase Difference then enters step (e) in iteration threshold；
(d), the center of circle c of adjustment iteration circle is the barycenter c of intersecting areai, radius r is the barycenter c of intersecting areaiIn hand region The borderline projections of polygon P, and return to step (b)；
(e), if S (I)/S (C) ＞ 1.1sth, then increase iteration radius of circle r, and return to the step (2)；If S (I)/S (C) ＜ 0.9sth, then reduce iteration radius of circle r, and return to step (b)；Otherwise iteration is terminated, the iteration circle at the end of output iteration Center location c and radius r, above-mentioned S (I), S (C) are respectively intersecting area and the iteration area of a circle, sthFor effective area pixel threshold Value.
5. the hand contour feature optimization method according to claim 1 based on depth image, it is characterized in that：In the step Suddenly further include the process that hand region profile is updated according to the iteration result of average drifting, which includes following sub-step in (4) Suddenly：
(a) according to iteration circle and the intersection situation of hand region polygon, hand region polygon is divided into intersecting area I and not Intersecting areaWherein disjoint range Q may be by multiple independent polygon { q1, q2..., qnConstitute；
(b) the independent polygon q of each of disjoint range Q are directed to, the length for overlapping line segment d of q and image boundary are calculated.IfThen q is wiped out in former hand region polygon Pi, and enter step S44；IfThen enter step S43；Above-mentioned dthLine segment threshold value is overlapped for image boundary；
(c) polygon q independent to each of disjoint range Q, calculates the barycenter c of qq, using the center of circle c of iteration circle as starting point, make Extended line cq calculates the length for overlapping line segment m of q and cq extended lines until image boundary.IfThen in original Q is wiped out in hand region polygon Pi；
(d) judge whether current hand region polygon P includes the polygon segments of multiple separation, if so, will wherein face The maximum polygon P of productmaxAs finally obtained accurate hand region Pa, and return to PaProfile as accurate hand profile.
6. the hand contour feature optimization method according to claim 1 based on depth image, it is characterized in that：In the step Suddenly the Outline Feature Vector process of convex closure detection is combined in (5), which includes following sub-step：
(a) Graham scanning methods is used to calculate the external convex closure profile of accurate hand region；
(b) it calculates the 7 of accurate hand region profile and ties up Hu not bending moment vectors；
(c) it calculates the 7 of external convex closure profile and ties up Hu not bending moment vectors；
(d) the hand Outline Feature Vector of 14 dimension Hu invariant rectangles Cheng Xin of combination hand region profile and external convex closure profile.
CN201810285718.8A 2018-03-23 2018-03-23 A kind of hand contour feature optimization method based on depth image Pending CN108520264A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810285718.8A CN108520264A (en) 2018-03-23 2018-03-23 A kind of hand contour feature optimization method based on depth image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810285718.8A CN108520264A (en) 2018-03-23 2018-03-23 A kind of hand contour feature optimization method based on depth image

Publications (1)

Publication Number Publication Date
CN108520264A true CN108520264A (en) 2018-09-11

Family

ID=63431664

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810285718.8A Pending CN108520264A (en) 2018-03-23 2018-03-23 A kind of hand contour feature optimization method based on depth image

Country Status (1)

Country Link
CN (1) CN108520264A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109914532A (en) * 2017-11-27 2019-06-21 仁宝电脑工业股份有限公司 Intelligent water supply method and intelligent water supply device
CN110189276A (en) * 2019-05-31 2019-08-30 华东理工大学 A kind of facial image restorative procedure based on very big radius circle domain
CN110717874A (en) * 2019-10-10 2020-01-21 徐庆 Image contour line smoothing method
CN111753462A (en) * 2020-05-22 2020-10-09 北京邮电大学 Method and device for determining environmental signal value

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130058565A1 (en) * 2002-02-15 2013-03-07 Microsoft Corporation Gesture recognition system using depth perceptive sensors
CN103472916A (en) * 2013-09-06 2013-12-25 东华大学 Man-machine interaction method based on human body gesture recognition
EP2860969A1 (en) * 2012-06-11 2015-04-15 Center of Human-Centered Interaction for Coexistence Three-dimensional video conference device capable of enabling eye contact and method using same
CN104899600A (en) * 2015-05-28 2015-09-09 北京工业大学 Depth map based hand feature point detection method
CN107341811A (en) * 2017-06-20 2017-11-10 上海数迹智能科技有限公司 The method that hand region segmentation is carried out using MeanShift algorithms based on depth image

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130058565A1 (en) * 2002-02-15 2013-03-07 Microsoft Corporation Gesture recognition system using depth perceptive sensors
EP2860969A1 (en) * 2012-06-11 2015-04-15 Center of Human-Centered Interaction for Coexistence Three-dimensional video conference device capable of enabling eye contact and method using same
CN103472916A (en) * 2013-09-06 2013-12-25 东华大学 Man-machine interaction method based on human body gesture recognition
CN104899600A (en) * 2015-05-28 2015-09-09 北京工业大学 Depth map based hand feature point detection method
CN107341811A (en) * 2017-06-20 2017-11-10 上海数迹智能科技有限公司 The method that hand region segmentation is carried out using MeanShift algorithms based on depth image

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
蒲兴成等: "基于改进Hu矩算法的Kinect手势识别", 《计算机工程》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109914532A (en) * 2017-11-27 2019-06-21 仁宝电脑工业股份有限公司 Intelligent water supply method and intelligent water supply device
CN110189276A (en) * 2019-05-31 2019-08-30 华东理工大学 A kind of facial image restorative procedure based on very big radius circle domain
CN110717874A (en) * 2019-10-10 2020-01-21 徐庆 Image contour line smoothing method
CN110717874B (en) * 2019-10-10 2022-11-04 徐庆 Image contour line smoothing method
CN111753462A (en) * 2020-05-22 2020-10-09 北京邮电大学 Method and device for determining environmental signal value
CN111753462B (en) * 2020-05-22 2023-02-03 北京邮电大学 Method and device for determining environmental signal value

Similar Documents

Publication Publication Date Title
CN108520264A (en) A kind of hand contour feature optimization method based on depth image
CN111723585B (en) Style-controllable image text real-time translation and conversion method
CN107341811B (en) Method for segmenting hand region by using MeanShift algorithm based on depth image
CN104978012B (en) One kind points to exchange method, apparatus and system
CN106407985B (en) A kind of three-dimensional human head point cloud feature extracting method and its device
CN104809446B (en) Palmmprint area-of-interest rapid extracting method based on correction volar direction
WO2019174276A1 (en) Method, device, equipment and medium for locating center of target object region
CN103971102A (en) Static gesture recognition method based on finger contour and decision-making trees
Zhu et al. Vision based hand gesture recognition using 3D shape context
CN109389121A (en) A kind of nameplate recognition methods and system based on deep learning
CN110287952B (en) Method and system for recognizing characters of dimension picture
CN104850847B (en) Image optimization system and method with automatic thin face function
CN103870808A (en) Finger vein identification method
CN103413271A (en) Document image rectifying method based on local information
CN111414837A (en) Gesture recognition method and device, computer equipment and storage medium
CN105759967A (en) Global hand gesture detecting method based on depth data
CN104408462A (en) Quick positioning method of facial feature points
CN109948461B (en) Sign language image segmentation method based on centroid positioning and distance transformation
CN106503619B (en) Gesture recognition method based on BP neural network
CN110794956A (en) Gesture tracking and accurate fingertip positioning system based on Kinect
CN114863492B (en) Method and device for repairing low-quality fingerprint image
CN110895683A (en) Kinect-based single-viewpoint gesture and posture recognition method
WO2021196013A1 (en) Word recognition method and device, and storage medium
CN114445853A (en) Visual gesture recognition system recognition method
CN110427909B (en) Mobile terminal driving license detection method and system, electronic equipment and 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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20180911