CN110781791B - Gait energy map optimization synthesis method based on gravity center alignment - Google Patents
Gait energy map optimization synthesis method based on gravity center alignment Download PDFInfo
- Publication number
- CN110781791B CN110781791B CN201910998018.8A CN201910998018A CN110781791B CN 110781791 B CN110781791 B CN 110781791B CN 201910998018 A CN201910998018 A CN 201910998018A CN 110781791 B CN110781791 B CN 110781791B
- Authority
- CN
- China
- Prior art keywords
- connected domain
- smaller
- gait
- picture
- value
- 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.)
- Active
Links
- 230000005021 gait Effects 0.000 title claims abstract description 96
- 230000005484 gravity Effects 0.000 title claims abstract description 49
- 238000005457 optimization Methods 0.000 title claims abstract description 14
- 238000001308 synthesis method Methods 0.000 title claims abstract description 14
- 238000000034 method Methods 0.000 claims description 26
- 230000014759 maintenance of location Effects 0.000 claims description 6
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000006872 improvement Effects 0.000 abstract description 2
- 230000002194 synthesizing effect Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 7
- 210000002683 foot Anatomy 0.000 description 5
- 238000006073 displacement reaction Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000003068 static effect Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 210000003371 toe Anatomy 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000037237 body shape Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 210000003414 extremity Anatomy 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 210000003127 knee Anatomy 0.000 description 1
- 210000003141 lower extremity Anatomy 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 210000000689 upper leg Anatomy 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
- G06V40/25—Recognition of walking or running movements, e.g. gait recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/24—Aligning, centring, orientation detection or correction of the image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a gait energy map optimization synthesis method based on center of gravity alignment, which comprises the steps of judging a binary image belonging to a gait cycle, firstly judging whether each frame of the binary image belonging to the gait cycle belongs to a broken-end image, acquiring a height value of a figure contour of each frame of the binary image according to a judgment result, judging and eliminating a headless image in the binary image belonging to the gait cycle according to the height value of the figure contour of each frame of the binary image, namely merging a head and a trunk of the broken-end image, finding and deleting the headless image, calculating accurate center of gravity coordinates of a reserved image, and finally synthesizing the gait energy map by using center of gravity alignment; compared with the traditional gait energy map synthesized by the gravity center alignment, the gait energy map optimization synthesis method based on the gravity center alignment has certain improvement, and particularly the ghost of the head is improved well.
Description
Technical Field
The invention relates to the field of image processing and computer vision research, in particular to a gait energy map optimization synthesis method based on gravity center alignment.
Background
In recent years, the increasing speed of information-based schedules makes information security and public security more and more prominent, and especially for public places with dense population, how to accurately identify the identity of a person and protect the information security becomes a critical problem which must be solved; the biometric identification technology is more and more popular among people due to the advantages of simplicity, rapidness, reliability and accuracy in identity identification, easiness in integration with safety, monitoring and management systems and the like; gait recognition is a member of biological recognition, and is attracting attention in this large background.
Compared with the traditional biological feature recognition, the gait recognition has the following three advantages:
the gait feature acquisition can be finished without being discovered due to the concealment without the cooperation of recognized people;
secondly, the gait characteristics can be acquired and identified in a long distance without close contact;
thirdly, the gait is difficult to disguise.
The gait recognition means that the extracted gait features are compared with sample features in a database, and a classifier decides the class to which the gait features belong and obtains a recognition result; for a long time, many gait recognition methods have been developed, and these gait recognition methods can be divided into two ways according to their strategies:
one gait identification method mainly based on a mode is to bring the static shape and kinetic energy movement of human gait into the gait feature extraction process; the static shape of gait is mostly described in an organization mode, such as a skeleton structure chart, ellipse approximation, a body trunk three-dimensional model and the like; or by measuring time variation parameters such as gait cycle, stance width value and stride; the kinetic energy movement of gait is described by movement modes, such as hip and thigh rotation modes, joint tracks, limb direction changes and the like; for example, xue et al use infrared images as the gait feature source for contour extraction to solve the problem of nighttime identification, and acquire images by the temperature sensing principle, so that the foreground is easier to be taken out from the background, the noise is less, and the image is not easily influenced by the illumination value or brightness value; in the feature representation part, xue et al extract behavior features from continuous images by wavelet transform, and obtain the height and the positions of the head, knee, left foot and right foot by a morphological image processing method, then calculate the ratio between different length values and the included angle between two feet (lower limbs) and the floor as physiological features, and finally use a support vector machine as a training and classifying method.
The other mode is a non-mode, namely a mode mainly based on appearance, and the contour features are extracted by taking a gait cycle as a unit; the gait cycle represents the movement of a person during walking, such as the periodicity of arm swing, step change and the like, and comprises a standing part and a swinging part; according to the analysis of the state of the sole of the foot in gait, the whole gait has several states, and the heel, the toes and the feet can land, leave the ground and hang in the air with different range values; according to the walking program of human body, the gait is summarized to include four states, wherein the heel and the sole are grounded, the toe is separated from the ground, and the sole is suspended.
A commonly used Gait representation method is a Gait energy map (GEI), wherein the GEI records the main shape and change of a binary contour in a one-step cycle, namely the average pixel value of each pixel point in the Gait cycle; the part with the pixel value of 1 in the gait energy image contains static information, the part with the pixel value of less than 1 contains dynamic information, the characteristics of the body shape, the stride, the pace and the like of a moving human body can be well represented, the extraction method is simple, the gait characteristics in one period can be comprehensively expressed, and the extracted gait characteristics are more continuous; furthermore, the gait energy map is relatively robust to noise, since random noise is suppressed during averaging over a period.
However, the person regions identified by using target detection for the image frames are not communicated, such as headless and broken ends, so that correct barycentric coordinates cannot be found, and further, the synthesized gait energy image is inaccurate; therefore, the method reasonably processes the headless and broken image frames and then synthesizes the gait energy map, and is particularly important for gait recognition.
The invention content is as follows:
the invention aims to provide a gait energy map optimization synthesis method based on gravity center alignment, which processes broken and headless image frames in a binary image belonging to a gait cycle, so that the synthesized gait energy map has a better subjective reconstruction effect.
In order to achieve the purpose, the invention adopts the following technical scheme:
a gait energy map optimization synthesis method based on gravity center alignment comprises the following steps:
step 1: preprocessing the collected human gait video, acquiring a binarization picture of each frame of image in the video, and selecting the binarization picture belonging to one gait cycle from the acquired binarization pictures; in particular, the method comprises the following steps of,
p＝(p 1 ,p2,...p i ,...,p n )；
wherein, P is a set of binary images belonging to a gait cycle, P i The method comprises the following steps of obtaining binary pictures of an ith frame picture belonging to one gait cycle, wherein i =1,2 and … n;
step 2: searching a corresponding connected domain for each binarization picture belonging to one gait cycle in the step 1; i.e. find p i A connected domain of (c);
and step 3: if said p is i If there is only one connected domain, then p is judged i Not belonging to the broken picture, define p i The contour is a human figure contour, and the step 5 is carried out;
and 4, step 4: if said p is i If a smaller connected domain is the lost head of the largest connected domain, then combining the smaller connected domain and the largest connected domain to form an integral region M, where M is defined as p i The figure outline of (1); if said p is i If all the smaller connected domains can not meet the judgment condition, discarding all the smaller connected domains, i.e. defining said p i Has a maximum connected component M' of p i The figure outline of (1);
wherein, M = M '+ M ", M' is the largest connected domain, M" is the smaller connected domain that satisfies all the determination conditions, i.e. the smaller connected domain;
and 5: calculating said p i Height value H of figure outline i There are three cases:
if said p is i If the contour is a human contour, then p is i Height value H of figure outline i I.e. is said p i A height value of the profile;
if M in step 4 is p i The figure outline of (1), then said p i The height value of the figure outline
Wherein M is y Ordinate, M, representing the top left corner of the maximum connected component h A height value representing a maximum connected domain>
A ordinate representing the upper left corner of the smaller connected domain;
if the maximum connected domain M' in the step 4 is p i The figure outline of (1), then said p i The height value H of the figure outline i ＝M h Wherein M is h A height value representing a maximum connected domain;
step 6: using p as described in step 5 i The height value H of the figure outline i Determining said p i If it belongs to headless picture, if said p i If the picture belongs to the headless picture, the p is abandoned i Keeping p ', p' as a first binary image belonging to a gait cycle;
the judging method comprises the following steps:
judgment of H i-1 And H i With respect to a threshold value D, wherein D =10,h i Is said p i Height value of the figure outline of (1), H i-1 Is p i-1 The height value of the figure outline of (1);
if | H i-1 -H i If | < D, then judge said p i Not belonging to headless pictures, retaining p' = (p) 1 ,p2,...p i ,...,p n )；
If | H i-1 -H i If > D, judging that the headless picture appears in the p, and dividing the two conditions into the following two conditions：
If H i-1 -H i If > D, judging that p is larger than i Belonging to headless pictures, discarding p i Retention of p' = (p) 1 ,p2,...p i-1 ,p i+1 ,...,p n )；
If H i -H i-1 If < D, p is judged 1 ,p2,...p i-1 All belong to headless pictures, discard p 1 ,p2,...p i-1 Retention of p' = (p) i+1 ,...,p n )；
And 7: carrying out height normalization processing on all the pictures in p 'to enable the height values of the binary image images of the human body contour in p' to be equal;
and 8: each frame picture p in p i Dividing into two parts of upper and lower parts according to the height value of the character outline, and calculating the corresponding gravity center M of the upper half part i s ；
And step 9: each frame picture p in p i According to its corresponding center of gravity M i s Sequentially aligning, and accumulating and averaging pixels of all reserved pictures to obtain a final gait energy image GEI;
wherein, G represents the finally generated gait energy image, and N represents that one gait cycle comprises N frames of images.
The judgment conditions in the step 4 are as follows:
condition 1: the ratio of the difference value of the abscissa of the gravity center point of the largest connected domain and the abscissa of the gravity center point of the smaller connected domain to the width value of the smaller connected domain is smaller than a threshold value A;
wherein M is cx Is the abscissa of the center of gravity point of the maximum connected domain,
is the abscissa, M, of the center of gravity point of the smaller connected domain w Is the width value of the smaller connected domain, where a =0.1;
condition 2: the ratio of the vertical direction distance value of the maximum connected domain and the smaller connected domain to the height value of the maximum connected domain is smaller than a threshold value B;
wherein,
is a height value of a smaller communicating volume>
Is the ordinate, M, of the upper left corner of the frame of the smaller connected domain y Is the ordinate, M, of the upper left corner of the frame of the maximum connected domain h A height value of the largest connected domain, wherein B =0.1;
condition 3: the vertical coordinate of the gravity center point of the smaller connected domain is higher than the upper boundary of the largest connected domain;
wherein,
is the ordinate of the center of gravity point of the smaller connected domain, M y Is the ordinate of the upper left corner of the frame of the maximum connected domain;
condition 4: the ratio of the area value of the maximum connected domain to the area value of the smaller connected domain is larger than a threshold value C 1 And is less than a threshold value C 2 ；
Wherein S is M Is the area value of the largest connected domain,
is the area value of the smaller connected component, where C 1 ＝7、C 2 ＝14。
The invention has the beneficial effects that:
the invention relates to a gait energy map optimization synthesis method based on gravity center alignment, which judges a binarization picture belonging to a gait cycle, firstly judges whether each frame of binarization picture belonging to the gait cycle belongs to a broken-end picture, acquires a height value of a figure outline of each frame of binarization picture according to a judgment result, judges and eliminates a headless picture in the binarization picture belonging to the gait cycle according to the height value of the figure outline of each frame of binarization picture, namely, combines a head part and a trunk part of the broken-end picture, finds and deletes the headless picture, calculates accurate gravity center coordinates of a reserved picture, and finally synthesizes a gait energy map by using gravity center alignment.
Description of the drawings:
in order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of a broken-end binary image;
FIG. 3 is a schematic diagram of a headless binary image;
FIG. 4 is a schematic diagram comparing GEIs generated using a conventional method and GEIs generated using the method of the present invention, where the first action is the GEI generated using the conventional method and the second action is the GEI generated using the method of the present invention.
The specific implementation mode is as follows:
the technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1: the invention relates to a gait energy map optimization synthesis method based on gravity center alignment, which comprises the following steps:
step 1: preprocessing the collected human gait video, acquiring a binarization picture of each frame of image in the video, and selecting the binarization picture belonging to one gait cycle from the acquired binarization pictures;
in particular, the method comprises the following steps of,
p＝(p 1 ,p2,...p i ,...,p n )；
wherein, P is a set of binary images belonging to a gait cycle, P i The method comprises the following steps of obtaining a binary image of an i frame image belonging to a gait cycle, wherein i =1,2 and … n;
the method specifically comprises the following steps:
acquiring a gait video of a person (experimental object) during normal walking by using a camera, and extracting a binary image of each frame of image in the video by using a target identification method; then selecting a binarization picture belonging to a gait cycle from the obtained binarization pictures;
and 2, step: searching a corresponding connected domain for each binarization picture belonging to one gait cycle in the step 1; i.e. find p i A connected domain of (c);
and 3, step 3: if said p is i If there is only one connected domain, then p is judged i Not belonging to the broken picture, define p i The contour is a figure contour, and the step 5 is carried out; the schematic diagram of the broken-end binary image is shown in FIG. 2;
if said p is i If a plurality of connected domains exist, judging whether a smaller connected domain (other connected domains except the maximum connected domain) is the lost head of the maximum connected domain or not by judging conditions; if and only if said p i A plurality of connected domains ofWhen the relation between a smaller connected domain and the largest connected domain meets all judgment conditions, judging that the smaller connected domain is the lost head of the largest connected domain;
it should be noted that the area of the binarized image formed by the human body is certainly much larger than that of the binarized image formed by the head of the human body, so that the default maximum connected domain is the binarized image formed by the human body;
the judgment conditions are as follows:
condition 1: the ratio of the difference value of the abscissa of the gravity center point of the largest connected domain and the abscissa of the gravity center point of the smaller connected domain to the width value of the smaller connected domain is smaller than a threshold value A;
wherein M is cx Is the abscissa of the center of gravity point of the maximum connected domain,
is the abscissa, M, of the center of gravity point of the smaller connected domain w Is the width value of the smaller connected domain, where a =0.1; when the conditions are met, the horizontal displacement of the maximum connected domain and the smaller connected domain can be ensured to be within a reasonable range, and the problem that the offset between the head and the body is not met when people walk when the horizontal displacement is large is avoided; the center of gravity coordinates are calculated according to the following formula:
wherein x is z(i) Is the abscissa, y, of the center of gravity point of the ith frame image z(i) Is the vertical coordinate, x, of the gravity center point of the ith frame image i 、y i Is the pixel point coordinate of the ith frame image, the target area is the smaller connected domain,n is the total number of pixel points in the connected domain;
condition 2: the ratio of the vertical direction distance value of the maximum connected domain and the smaller connected domain to the height value of the maximum connected domain is smaller than a threshold value B;
wherein,
a height value in the smaller connected domain>
Is the ordinate, M, of the upper left corner of the border of the smaller connected domain y Is the ordinate, M, of the upper left corner of the frame of the maximum connected domain h A height value of the largest connected domain, wherein B =0.1; when the conditions are met, the vertical displacement of the maximum connected domain and the smaller connected domain can be ensured to be within a reasonable range, and the problem that the distance between the head and the body is large when the walking of a person is not met when the vertical displacement is large is avoided;
condition 3: the vertical coordinate of the gravity center point of the smaller connected domain is higher than the upper boundary of the largest connected domain;
wherein,
is the ordinate of the center of gravity point of the smaller connected domain, M y The ordinate of the upper left corner of the frame of the maximum connected domain is shown; when the conditions are met, the smaller connected domain can be ensured to be above the largest connected domain, and the situation that the head is lower than the body when people walk is avoided;
condition 4: the ratio of the area value of the maximum connected domain to the area value of the smaller connected domain is larger than a threshold value C 1 And is less than a threshold value C 2 ；
Wherein S is M Is the area value of the largest connected domain,
is the area value of the smaller connected domain, wherein C 1 ＝7，C 2 =14; when the conditions are met, the ratio of the area value of the largest connected domain to the area value of the smaller connected domain can be ensured to be in a reasonable range, and the problem that the ratio of the head to the body of a person is not in accordance with common knowledge is avoided;
and 4, step 4: if said p is i If a smaller connected domain is the lost head of the largest connected domain, then combining the smaller connected domain and the largest connected domain to form an integral region M, where M is defined as p i The figure outline of (1); if said p is i If all the smaller connected domains can not meet the judgment condition, discarding all the smaller connected domains, i.e. defining said p i Is said p i The figure outline of (1);
wherein, M = M '+ M ", M' is the largest connected domain, M" is the smaller connected domain that satisfies all the determination conditions, i.e. the smaller connected domain;
and 5: calculating said p i Height value H of figure outline i The following three cases are distinguished:
if said p is i If the contour is a human contour, then p is i Height value H of figure outline i Is said p i A height value of the profile;
if M in step 4 is p i The figure outline of (1), then said p i Height value of the figure outline
Wherein M is y Ordinate, M, representing the top left corner of the maximum connected component h High representing maximum connected domainValue,. Or>
A ordinate representing the upper left corner of the smaller connected domain;
if the maximum connected domain M' in the step 4 is p i The figure outline of (1), then said p i The height value H of the figure outline i ＝M h Wherein M is h A height value representing a maximum connected domain;
step 6: using p as described in step 5 i Height value H of figure outline i Judging said p i If it belongs to headless picture, if said p i If the picture belongs to the headless picture, the p is abandoned i Keeping p ', p' as a first binary image belonging to a gait cycle; a schematic diagram of the headless binary image is shown in fig. 3;
the judging method comprises the following steps:
judgment of H i-1 And H i With respect to a threshold value D, wherein D =10,h i Is said p i Height value of the figure outline of (1), H i-1 Is p i-1 The height value of the figure outline of (1);
if | H i-1 -H i If | < D, then judge said p i Not belonging to headless pictures, keep p' = (p) 1 ,p2,...p i ,...,p n )；
If | H i-1 -H i If the picture is more than D, judging that the headless picture appears in the p, and dividing the two situations into the following two situations:
if H i-1 -H i If > D, judging p i Belonging to headless pictures, discarding p i Retention of p' = (p) 1 ,p2,...p i-1 ,p i+1 ,...,p n )；
If H is i -H i-1 If < D, p is judged 1 ,p2,...p i-1 All belong to headless pictures, discard p 1 ,p2,...p i-1 Retention of p' = (p) i+1 ,...,p n )；
The basis for making the above judgment is as follows:
because the camera for acquiring the gait video of the person is fixed and the person moves, the image of the person is larger when the person is close to the camera according to the principle of big or small distance; when the person is far from the camera, the image of the person is small, and therefore the previous frame p i-1 The height value of the figure outline and the current frame p i If the difference value is too large, judging that a headless picture appears in p, and when the headless picture appears in a certain frame in the middle of p, namely the picture sequence in p is a head picture, a headless picture and a head picture, only discarding the headless picture can eliminate the influence on the gait energy map; when all pictures from the first frame to the i-1 frame in p are headless pictures, the previous i-1 frame pictures may all satisfy | H | i-1 -H i D is less than or equal to | H appears until the ith frame shows the head picture i -H i-1 If the number D is less than D, judging that the previous i-1 frame picture is a headless picture, namely abandoning the previous i-1 frame picture, namely eliminating the influence on the gait energy picture;
and 7: performing height normalization processing on all the pictures in p 'to enable the height values of the binary image images of the human body contour in p' to be equal;
and 8: picture p of each frame in p i Dividing the figure contour into two parts of average upper and lower parts according to the height value of the figure contour, and calculating the corresponding gravity center M of the upper half part i s ；
And step 9: picture p of each frame in p i According to its corresponding center of gravity M i s Sequentially aligning, and accumulating and averaging pixels of all reserved pictures to obtain a final gait energy image GEI;
wherein, G represents the finally generated gait energy image, and N represents that one gait cycle comprises N frames of images.
The invention relates to a gait energy map optimization synthesis method based on gravity center alignment, which judges a binarization picture belonging to a gait cycle, firstly judges whether each frame of binarization picture belonging to the gait cycle belongs to a broken-end picture, acquires a height value of a figure outline of each frame of binarization picture according to a judgment result, judges and eliminates a headless picture in the binarization picture belonging to the gait cycle according to the height value of the figure outline of each frame of binarization picture, namely, combines a head part and a trunk part of the broken-end picture, finds and deletes the headless picture, calculates accurate gravity center coordinates of a reserved picture, and finally synthesizes a gait energy map by using gravity center alignment.
As shown in fig. 4, the gait energy map optimization synthesis method based on center of gravity alignment according to the invention has certain improvement compared with the traditional gait energy map synthesized by center of gravity alignment, and particularly the ghost of the head is improved; the reason why the ghost phenomenon occurs is that pictures without heads and broken heads are not distinguished, so that the correct barycentric coordinate cannot be calculated; particularly, under the condition of no head, the calculated barycentric coordinates are deviated from the real coordinates, so that the head generates ghost images; the gait energy diagram synthesized by the gait energy diagram optimization synthesis method based on the gravity center alignment overcomes the defects and has good alignment effect; an MT twin network is used as a training network, the recognition accuracy rates of different algorithms are shown in table 1, the first row is the recognition accuracy rate of the traditional gravity center alignment method, and the second row is the recognition accuracy rate of the method; compared with the traditional method, the method disclosed by the invention has the advantages that the identification accuracy is improved by 16.52 percentage points; the method of the invention not only has good alignment synthesis effect, but also has unusual expression in recognition accuracy.
Table 1: gait recognition accuracy comparison
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (2)
1. A gait energy map optimization synthesis method based on gravity center alignment is characterized by comprising the following steps:
step 1: preprocessing the collected human gait video, acquiring a binarization picture of each frame of image in the video, and selecting the binarization picture belonging to one gait cycle from the acquired binarization pictures; specifically, p = (p) 1 ,p2,...p i ,...,p n )；
Wherein, P is a set of binary images belonging to a gait cycle, P i The method comprises the following steps of obtaining binary pictures of an ith frame picture belonging to one gait cycle, wherein i =1,2 and … n;
step 2: searching a corresponding connected domain for each binarization picture belonging to one gait cycle in the step 1; i.e. find p i A connected domain of (c);
and step 3: if said p is i If there is only one connected domain, then p is judged i Not belonging to the broken picture, define p i The contour is a human figure contour, and the step 5 is carried out;
and 4, step 4: if said p is i If a smaller connected domain is the lost head of the largest connected domain, then combining the smaller connected domain and the largest connected domain to form an integral region M, where M is defined as p i The figure outline of (1); if said p is i If all the smaller connected domains can not meet the judgment condition, discarding all the smaller connected domains, i.e. defining said p i Maximum connected domain ofM' is said for p i The figure outline of (1);
wherein, M = M '+ M ", M' is the largest connected domain, M" is the smaller connected domain that satisfies all the determination conditions, i.e. the smaller connected domain;
and 5: calculating said p i The height value H of the figure outline i The following three cases are distinguished:
if said p is i If the contour is a human contour, then p is i The height value H of the figure outline i Is said p i A height value of the profile;
if M in step 4 is p i The figure outline of (1), then said p i Height value H of figure outline i ＝M y +M h -M’ y Wherein M is y Ordinate, M, representing the top left corner of the maximum connected component h Height value, M ', representing the maximum connected region' y The ordinate of the upper left corner of the smaller connected domain is represented;
if the maximum connected domain M' in the step 4 is p i The figure outline of (1), then said p i Height value H of figure outline i ＝M h Wherein M is h A height value representing a maximum connected domain;
step 6: using p as described in step 5 i Height value H of figure outline i Judging said p i If it belongs to headless picture, if said p i If the picture belongs to the headless picture, the p is abandoned i Keeping p ', p' as a first binary image belonging to a gait cycle;
the judging method comprises the following steps:
judgment of H i-1 And H i With respect to a threshold value D, wherein D =10,h i Is said p i Height value of the figure outline of (1), H i-1 Is p i-1 The height value of the figure outline of (1);
if | H i-1 -H i If | < D, then judge said p i Not belonging to headless pictures, retaining p' = (p) 1 ,p2,...p i ,...,p n )；
If | H i-1 -H i If > D, judging that the headless picture appears in the p, and dividing the following two conditions:
if H is i-1 -H i If > D, judging p i Belongs to headless pictures, abandons p i Retention of p' = (p) 1 ,p2,...p i-1 ,p i+1 ,...,p n )；
If H is i -H i-1 If < D, p is judged 1 ,p2,...p i-1 All belong to headless pictures, discard p 1 ,p2,...p i-1 Retention of p' = (p) i+1 ,...,p n )；
And 7: carrying out height normalization processing on all the pictures in p 'to enable the height values of the binary image images of the human body contour in p' to be equal;
and 8: picture p of each frame in p i Dividing the figure contour into two parts of average upper and lower parts according to the height value of the figure contour, and calculating the corresponding gravity center M of the upper half part i s ；
And step 9: picture p of each frame in p i According to its corresponding center of gravity M i s Sequentially aligning, and accumulating and averaging pixels of all reserved pictures to obtain a final gait energy image GEI;
wherein, G represents the finally generated gait energy image, and N represents that one gait cycle comprises N frames of images.
2. The gait energy map optimization synthesis method based on gravity center alignment as claimed in claim 1, characterized in that: the judgment conditions in the step 4 are as follows:
condition 1: the ratio of the difference value of the abscissa of the gravity center point of the largest connected domain and the abscissa of the gravity center point of the smaller connected domain to the width value of the smaller connected domain is smaller than a threshold value A;
((M cx -M’ cx )/M w )＜A；
wherein M is cx Is the barycentric point abscissa, M 'of the maximum connected domain' cx Is the abscissa, M, of the center of gravity point of the smaller connected domain w Is the width value of the smaller connected domain, where a =0.1;
condition 2: the ratio of the vertical direction distance value of the maximum connected domain and the smaller connected domain to the height value of the maximum connected domain is smaller than a threshold value B;
(M’ h +M’ y -M y )/M h ＜B；
wherein, M' h Is a height value of a smaller connected domain, M' y Is the ordinate, M, of the upper left corner of the frame of the smaller connected domain y Is the ordinate, M, of the upper left corner of the frame of the maximum connected domain h A height value of the largest connected domain, wherein B =0.1;
condition 3: the vertical coordinate of the gravity center point of the smaller connected domain is higher than the upper boundary of the largest connected domain;
M’ cy ＜M y ；
wherein, M' cy Is the ordinate of the center of gravity point of the smaller connected domain, M y The ordinate of the upper left corner of the frame of the maximum connected domain is shown;
condition 4: the ratio of the area value of the maximum connected domain to the area value of the smaller connected domain is larger than a threshold value C 1 And is less than a threshold value C 2 ；
Wherein S is M Is the area value of the maximum connected region, S' M Is the area value of the smaller connected domain, wherein C 1 ＝7、C 2 ＝14。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910998018.8A CN110781791B (en) | 2019-10-18 | 2019-10-18 | Gait energy map optimization synthesis method based on gravity center alignment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910998018.8A CN110781791B (en) | 2019-10-18 | 2019-10-18 | Gait energy map optimization synthesis method based on gravity center alignment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110781791A CN110781791A (en) | 2020-02-11 |
| CN110781791B true CN110781791B (en) | 2023-03-24 |
Family
ID=69386091
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910998018.8A Active CN110781791B (en) | 2019-10-18 | 2019-10-18 | Gait energy map optimization synthesis method based on gravity center alignment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110781791B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104834893A (en) * | 2015-03-13 | 2015-08-12 | 燕山大学 | Front-view pedestrian gait period detection method |
| JP2018026018A (en) * | 2016-08-10 | 2018-02-15 | 公立大学法人大阪市立大学 | Gait analyzer |
| CN108830259A (en) * | 2018-06-30 | 2018-11-16 | 天津大学 | Gait recognition method based on average difference image |
| WO2019041412A1 (en) * | 2017-08-31 | 2019-03-07 | 中国科学院微电子研究所 | Thermal imaging-based guan pulse identification method |
-
2019
- 2019-10-18 CN CN201910998018.8A patent/CN110781791B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104834893A (en) * | 2015-03-13 | 2015-08-12 | 燕山大学 | Front-view pedestrian gait period detection method |
| JP2018026018A (en) * | 2016-08-10 | 2018-02-15 | 公立大学法人大阪市立大学 | Gait analyzer |
| WO2019041412A1 (en) * | 2017-08-31 | 2019-03-07 | 中国科学院微电子研究所 | Thermal imaging-based guan pulse identification method |
| CN108830259A (en) * | 2018-06-30 | 2018-11-16 | 天津大学 | Gait recognition method based on average difference image |
Non-Patent Citations (2)
| Title |
|---|
| 基于HOG特征的步态能量图身份识别算法;崔雪红等;《电子测量技术》;20170715(第07期);全文 * |
| 基于下肢轮廓角度距离和步幅变化的步态识别算法研究;王希娟;《计算机应用与软件》;20180215(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110781791A (en) | 2020-02-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Singh et al. | Vision-based gait recognition: A survey | |
| CN105279483B (en) | A kind of tumble behavior real-time detection method based on depth image | |
| Tafazzoli et al. | Model-based human gait recognition using leg and arm movements | |
| Yoo et al. | Extracting human gait signatures by body segment properties | |
| US20170243057A1 (en) | System and method for automatic gait cycle segmentation | |
| US20020028003A1 (en) | Methods and systems for distinguishing individuals utilizing anatomy and gait parameters | |
| RU2546327C1 (en) | Human tracking device, human tracking method and non-temporary computer-readable medium storing human tracking programme | |
| CN109949341A (en) | A kind of pedestrian target tracking based on human skeleton structured features | |
| CN111027417B (en) | Gait recognition method and gait evaluation system based on human body key point detection algorithm | |
| CN109271918B (en) | Method for distinguishing people with balance ability disorder based on gravity center shift model | |
| Isaac et al. | Trait of gait: A survey on gait biometrics | |
| JP7531168B2 (en) | Method and system for detecting a child's sitting posture based on child's face recognition | |
| CN109919137A (en) | A kind of pedestrian's structured features expression | |
| Bora et al. | Understanding human gait: A survey of traits for biometrics and biomedical applications | |
| CN111243230A (en) | Human body falling detection device and method based on two depth cameras | |
| Liu et al. | Human body fall detection based on the Kinect sensor | |
| CN104331705B (en) | Automatic detection method for gait cycle through fusion of spatiotemporal information | |
| CN110781791B (en) | Gait energy map optimization synthesis method based on gravity center alignment | |
| CN110765925A (en) | Carrier detection and gait recognition method based on improved twin neural network | |
| CN107007285B (en) | Fall detection method based on pressure and acceleration transducer | |
| CN111144167A (en) | Gait information identification optimization method, system and storage medium | |
| Singh et al. | Biometric gait recognition with carrying and clothing variants | |
| CN112613430B (en) | Gait recognition method based on deep migration learning | |
| Madden et al. | A framework for track matching across disjoint cameras using robust shape and appearance features | |
| JP2005074075A (en) | Walking cycle measuring device, periodic image obtaining device, compression processor for moving object outline, and moving object discrimination system |
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 |