CN107270829B

CN107270829B - Human body three-dimensional measurement method based on depth image

Info

Publication number: CN107270829B
Application number: CN201710430782.6A
Authority: CN
Inventors: 周晓军; 王行; 盛赞; 李朔; 李骊; 杨高峰
Original assignee: Nanjing Huajie Imi Technology Co ltd
Current assignee: Nanjing Huajie Imi Technology Co ltd
Priority date: 2017-06-08
Filing date: 2017-06-08
Publication date: 2020-06-19
Anticipated expiration: 2037-06-08
Also published as: CN107270829A

Abstract

The invention discloses a human body three-dimensional measurement method based on a depth image, which comprises the following steps: acquiring a preset number of human body three-dimensional data samples, and respectively fitting linear regression equations of chest circumference, chest width, chest thickness, waist circumference, waist width, waist thickness and hip circumference, hip width and hip thickness to the preset number of human body three-dimensional data samples by using a linear regression method; acquiring a human body front depth image, a human body side depth image and human body skeleton information of a user to be measured; processing the human body front depth image to obtain a human body front foreground image; processing the depth image of the side face of the human body to obtain a foreground image of the side face of the human body; and finally calculating to obtain the chest width and chest thickness, the waist width and waist thickness and the hip width and hip thickness of the human body based on the human body side foreground image and the human body skeleton information, and respectively substituting the chest width and chest thickness, waist width and waist thickness and hip width and hip thickness into corresponding linear regression equations to obtain the three-dimensional data of the human body. The on-line and accurate human body three-dimensional data calculation device has the advantages of simple structure and convenience in operation.

Description

Human body three-dimensional measurement method based on depth image

Technical Field

The invention relates to the technical field of human body three-dimensional measurement, in particular to a human body three-dimensional measurement method based on a depth image.

Background

In general, as computer technology and internet technology have developed, more and more computer applications use 3D graphic image technology. In some application fields, such as online fitting, somatosensory games, three-dimensional human body reconstruction, clothing design, game character animation and the like, the demand for online acquisition of the human body three-dimensional data is increasing. However, due to various reasons such as human body differences and complexity of application scenarios, it becomes a difficult topic to accurately calculate the human body three-dimensional data on line.

Chinese patent application CN105222738A discloses a "method for measuring the size of human body 3D model data", which is a measuring method performed on the basis of obtaining a human body 3D model data file by 3D scanning technology, firstly, a scale is set corresponding to a complete human body 3D model through a system foreground, after an operator clicks the height of a three-dimensional object to be measured on the scale, the height is converted into a height proportional value by the system foreground, and the proportional value is transmitted to a server, the server retrieves the human body 3D model data, and finds out a set of all points representing the height to be measured in the file; and after the coordinates of each point in the point set are obtained, the perimeter of the geometric figure around which each point in the point set is wound is obtained through a wrapping algorithm, and the perimeter is the dimension of the three-dimensional position of the human body model at the height to be measured. The method has the following defects: on one hand, a 3D scanner is required to scan a human body to obtain a 3D model data file of the human body before measuring the three-dimensional data, and meanwhile, the data file is required to be placed at a server end, so that the data file cannot be used in a conventional application scene and cannot be measured on line in real time; on the other hand, the method needs interaction between an operator and a system foreground to carry out measurement, increases the complexity of operation, and brings potential errors to a calculation result.

Chinese patent application CN103767219A discloses a non-contact human body three-dimensional size measuring method, which comprises the steps of firstly obtaining a human body digital picture, then measuring pixels of a plurality of body type characteristic points of a human body in the digital picture, and obtaining a first size group of the body type characteristics of the human body in the digital picture through calculation; and calculating a second size group corresponding to the actual body shape characteristics of the human body in the digital picture according to the first size group. The method has the following defects: on one hand, after the human body digital picture is obtained, the method needs to preprocess the digital picture, and carries out scaling, angle adjustment and turning over on a plurality of pictures so as to keep the proportion and the angle of the plurality of pictures consistent, thereby increasing the complexity of the method undoubtedly, causing calculation errors at the same time and having higher difficulty in realizing the method; on the other hand, the method is inevitably influenced by illumination and environment due to the use of digital pictures, so that certain requirements are required for application scenes, and the method is limited from being widely applied.

Chinese patent application CN103535960A discloses a 'human body three-dimensional measurement method based on digital images', which is realized based on a human body three-dimensional cross section database and two-dimensional digital images, firstly, a high-precision scale data table is made through the digital images, and human body plane two-dimensional data is measured through the digital images and the scale table on the front and the side of the human body; the human body three-dimensional cross section is obtained through the three-dimensional scanner, the human body three-dimensional cross section and the three-dimensional scanner are fitted to obtain the outline of the human body measured conventionally, and then the three-dimensional data of the human body are obtained. The method has the following defects: on one hand, the method has complex steps, is difficult to realize and cannot carry out online real-time measurement; on the other hand, the method needs to be supported by hardware equipment of a digital camera and a three-dimensional scanner, and simultaneously needs vector software and three-dimensional software for data processing, so that deviation accumulation occurs in the processing process, and finally the obtained human body three-dimensional data has larger deviation; also, the method uses digital images, which are also affected by the environment and light, and cannot be widely used.

Therefore, how to overcome the above-mentioned deficiencies in the prior art becomes a technical problem to be solved in the field.

Disclosure of Invention

The invention aims to at least solve one of the technical problems in the prior art and provides a human body three-dimensional measuring method based on a depth image.

In order to achieve the above object, the present invention provides a method for measuring a three-dimensional circumference of a human body based on a depth image, the method comprising:

acquiring a preset number of human body three-dimensional data samples, and fitting a linear regression equation of chest circumference, chest width and chest thickness, a linear regression equation of waist circumference, waist width and waist thickness and a linear regression equation of hip circumference, hip width and hip thickness to the preset number of human body three-dimensional data samples by using a linear regression method;

acquiring a human body front depth image, a human body side depth image and human body skeleton information of a user to be measured;

processing the human body front depth image to obtain a human body front foreground image; processing the depth image of the side face of the human body to obtain a foreground image of the side face of the human body;

calculating front and rear two edge point pixels of a chest, front and rear two edge point pixels of a waist and front and rear two edge point pixels of a hip of the human body based on the human body side foreground image and the human body skeleton information; calculating to obtain the Euclidean distance corresponding to the coordinates under a world coordinate system according to the front and rear edge point pixels of the chest so as to obtain the chest thickness of the human body; calculating to obtain the Euclidean distance corresponding to the coordinates under the world coordinate system according to the front and rear edge point pixels of the waist so as to obtain the waist thickness of the human body; calculating to obtain the Euclidean distance corresponding to the coordinates under a world coordinate system according to the front and rear edge point pixels of the hip so as to obtain the thickness of the hip of the human body;

calculating left and right edge point pixels of a chest, left and right edge point pixels of a waist and left and right edge point pixels of a hip of the human body based on the front foreground image of the human body and the skeleton information of the human body; calculating to obtain Euclidean distance corresponding to coordinates under a world coordinate system according to the left and right edge point pixels of the chest so as to obtain the chest width of the human body; calculating to obtain Euclidean distance corresponding to coordinates under a world coordinate system according to the left and right edge point pixels of the waist so as to obtain the waist width of the human body; calculating to obtain the Euclidean distance corresponding to the coordinates under a world coordinate system according to the left and right edge point pixels of the hip so as to obtain the width of the hip of the human body;

and substituting the calculated chest thickness and the calculated chest width of the human body into a linear regression equation of the chest circumference, the chest width and the chest thickness, substituting the calculated waist thickness and the calculated waist width of the human body into a linear regression equation of the waist circumference, the waist width and the waist thickness, and substituting the calculated hip thickness and the calculated hip width of the human body into a linear regression equation of the hip circumference, the hip width and the hip thickness to obtain the three-circumference data of the human body.

Preferably, the preset number of the human body three-dimensional data samples comprise human body three-dimensional data samples with different ages, different sexes and/or different body types.

Preferably, the linear regression equation of the circumference to the width and thickness of the chest is:

BC＝x_b*BW+y_b*BT+z_b(ii) a And the number of the first and second groups,

the linear regression equation of waist circumference and waist width and thickness:

WC＝x_w*WW+y_w*WT+z_w(ii) a And the number of the first and second groups,

the linear regression equation of the hip circumference, the hip width and the hip thickness is as follows:

HC＝x_h*HW+y_h*HT+z_h；

wherein BC is the chest circumference size, BW is the chest width size, BT is the chest thickness size, x_bWeight coefficient of chest width in linear regression equation of chest circumference, y_bWeight coefficient for chest thickness in the linear regression equation for chest circumference, z_bError compensation constants in the chest circumference linear regression equation;

WC is waist size, WW is waist width size, WT is waist thickness size, x_wIs the weight coefficient of waist width in the waist circumference linear regression equation, y_wIs the weight coefficient of waist thickness in the waist circumference linear regression equation, z_wAn error compensation constant in the waist linear regression equation is obtained;

HC is hip circumference size, HW is hip width size, HT is hip thickness size, x_hThe weight coefficient of the width of the buttocks in the linear regression equation of the circumference of the buttocks, the weight coefficient of the thickness of the buttocks in the linear regression equation of the circumference of the buttocks, the z_hAnd (4) an error compensation constant in the hip circumference linear regression equation.

Preferably, the human body front depth image comprises human body front pixels and human body front background pixels; and the number of the first and second groups,

the human body side depth image comprises human body side pixels and human body side background pixels;

the pixel values of the human body front pixels and the human body front background pixels are depth values, and the pixel values of the human body side pixels and the human body side background pixels are depth values.

Preferably, the front depth image of the human body is processed to obtain a front foreground image of the human body; and processing the depth image of the side face of the human body to obtain a foreground image of the side face of the human body, wherein the step of processing the depth image of the side face of the human body to obtain the foreground image of the side face of the human body comprises the following steps:

extracting human body front pixels in the human body front depth image;

the human body front foreground graph meets the following formula:

wherein, F is_f(x_f，y_f) Is a front foreground map of the human body, I (x)_f，y_f) The coordinate in the human body front depth image is (x)_f，y_f) A pixel of (a); said H_fThe human body front pixel set is the human body front depth image; b is_fThe human body front background pixel set in the human body front depth image is obtained; and the number of the first and second groups,

extracting human body side pixels in the human body side depth image;

the human body side face foreground image meets the following formula:

wherein, the I (x)_s，y_s) The coordinate in the depth image of the side face of the human body is (x)_s，y_s) A pixel of (a); said H_sA set of the human body side pixels that are the human body side depth image; b is_sAnd the image is a human body side background pixel set in the human body side depth image.

Preferably, the step of calculating front and rear two edge point pixels of the human chest based on the human side foreground image and the human skeleton information, and calculating the euclidean distance corresponding to the coordinates in the world coordinate system according to the front and rear two edge point pixels of the chest to obtain the human chest thickness includes:

extracting pixel coordinates of left and right shoulder joint points of the human body skeleton information corresponding to the human body side foreground image, and recording the pixel coordinates of the left shoulder joint point as (x)_ls，y_ls) And the pixel coordinate of the right shoulder joint point is marked as (x)_rs，y_rs)；

Calculating the pixel coordinates of the centers of the left and right shoulder joint points corresponding to the human body side foreground map, and recording the pixel coordinates of the centers of the left and right shoulder joint points as ((x)_ls+x_rs)/2，(y_ls+y_rs)/2)；

Extracting pixel coordinates of the spine central joint of the human body skeleton information corresponding to the human body side foreground image, and recording the pixel coordinates of the spine central joint as (x)_sp，y_sp)；

Based on pixel coordinates (x) of the left shoulder joint point_ls，y_ls) Pixel coordinate (x) of the right joint point_rs，y_rs) Pixel coordinates ((x) of centers of the left and right shoulder joint points_ls+x_rs)/2，(y_ls+y_rs) /2) and the pixel coordinates (x) of the central joint of the spine_sp，y_sp) Determining two front and back edge point pixels of the chest, and recording the front edge point pixel of the chest as I (x)_bf，y_b) The posterior border point pixel of the thorax is denoted as I (x)_bb，y_b)；

According to the anterior border point pixel I (x) of the chest_bf，y_b) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_wbf，y_wbf，z_wbf) (ii) a According to the posterior edge point pixel I (x) of the chest_bb，y_b) Calculating to obtain world coordinates under the corresponding world coordinate system, and recording the world coordinates as (x)_wbb，y_wbb，z_wbb)；

According to the chestThe leading edge point pixel of (a) corresponds to a world coordinate (x)_wbf，y_wbf，z_wbf) And world coordinates (x) corresponding to the posterior border point pixels of the thorax_wbb，y_wbb，z_wbb) Obtaining the chest thickness of the human body; wherein the human body chest thickness BT satisfies the following formula:

and the number of the first and second groups,

the step of calculating front and rear two edge point pixels of the human waist based on the human side foreground image and the human skeleton information, and calculating the Euclidean distance corresponding to the coordinates under the world coordinate system according to the front and rear two edge point pixels of the waist to obtain the human waist thickness comprises the following steps:

Extracting pixel coordinates of left and right hip joint points of the human body skeleton information corresponding to the human body side foreground image, and recording the pixel coordinates of the left hip joint points as (x)_lh，y_lh) The pixel coordinate of the right hip joint point is (x)_rh，y_rh)；

Calculating the pixel coordinates of the centers of the left hip joint point and the right hip joint point corresponding to the human body side foreground image, and recording the pixel coordinates of the centers of the left hip joint point and the right hip joint point as ((x)_lh+x_rh)/2，(y_lh+y_rh)/2)；

The pixel coordinate based on the spine center joint is noted as (x)_sp，y_sp) Pixel coordinate (x) of the left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) /2) determining two front and rear edge point pixels of the waist, and recording the front edge point pixel of the waist as I (x)_wf，y_w) The back edge point pixel of the waist is denoted as I (x)_wb，y_w)；

According to the front edge point pixel I (x) of the waist_wf，y_w) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_wwf，y_wwf，z_wwf) (ii) a According to the back edge point pixel I (x) of the waist_wb，y_w) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_wwb，y_wwb，z_wwb)；

According to the world coordinate (x) corresponding to the front edge point pixel of the waist_wwf，y_wwf，z_wwf) And world coordinates (x) corresponding to the back edge point pixels of the waist_wwb，y_wwb，z_wwb) Obtaining the waist thickness of the human body; wherein the human waist thickness WT satisfies the following formula:

and the number of the first and second groups,

the method comprises the following steps of calculating front and rear two edge point pixels of the hip of a human body based on the side foreground image of the human body and the skeleton information of the human body, and calculating to obtain the Euclidean distance corresponding to the coordinates under a world coordinate system according to the front and rear two edge point pixels of the hip so as to obtain the thickness of the hip of the human body:

Pixel coordinates (x) based on the left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) And/2) determining two front and rear edge point pixels of the hip, and recording the front edge point pixel of the hip as I (x)_hf，y_h) The posterior edge point pixel of the hip is denoted as I (x)_hb，y_h)；

From the leading edge point pixel I (x) of the hip_hf，y_h) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_whf，y_whf，z_whf) (ii) a According to the posterior edge point pixel I (x) of the hip_hby_h) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_whb，y_whb，z_whb)；

According to the corresponding world coordinate (x) of the front edge point pixel of the hip_whf，y_whf，z_whf) And world coordinates (x) corresponding to posterior edge point pixels of the buttocks_whb，y_whb，z_whb) Obtaining the thickness of the hip of the human body; wherein the human body hip thickness HT satisfies the following formula:

preferably, said pixel coordinates (x) based on said left shoulder joint point_ls，y_ls) Pixel coordinate (x) of the right joint point_rs，y_rs) Pixel coordinates ((x) of centers of the left and right shoulder joint points_ls+x_rs)/2，(y_ls+y_rs) /2) and the pixel coordinates (x) of the central joint of the spine_sp，y_sp) Determining two front and back edge point pixels of the chest, and recording the front edge point pixel of the chest as I (x)_bf，y_b) The posterior border point pixel of the thorax is denoted as I (x)_bb，y_b) Comprises the following steps:

(y) th from the side foreground map of the human body_ls+y_rs) Line by line,/2 to y_spLines and calculating the left and right edges of the side pixels of the human body of each lineEdge pixel points to determine the line with the maximum distance between the left edge pixel point and the right edge pixel point, wherein the line is the line where the front edge pixel point and the rear edge pixel point of the chest are located, and the coordinates of the line are recorded as y_b(ii) a And the number of the first and second groups,

the pixel coordinate based on the spine central joint is marked as (x)_sp，y_sp) Pixel coordinate (x) of the left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) /2) determining two front and rear edge point pixels of the waist, and recording the front edge point pixel of the waist as I (x)_wf，y_w) The back edge point pixel of the waist is denoted as I (x)_wb，y_w) Comprises the following steps:

(y) th of the human body side foreground map_sp+(y_lh+y_rh) row/2)/2-TH and row (y)_sp+(y_lh+y_rh) Line-by-line scanning is carried out between the lines of/2)/2 + TH, and left and right edge pixel points of the side pixels of the human body in each line are calculated to determine the line with the maximum distance between the left and right edge pixel points, wherein the line is the line where the front and rear edge pixel points of the waist are located, and the line coordinate is recorded as y_wWherein, the TH is a threshold value of a search line; and the number of the first and second groups,

pixel coordinates (x) based on the left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) And/2) determining two front and rear edge point pixels of the hip, and recording the front edge point pixel of the hip as I (x)_hf，y_h) The posterior edge point pixel of the hip is denoted as I (x)_hb，y_h) Comprises the following steps:

(y) th of the human body side foreground map_lh+y_rh) the/2-TH row and the (y) TH_lh+y_rh) Line-by-line scanning is carried out between the/2 + TH lines, and the left and right edge pixel points of the side pixels of the human body of each line are calculated to determine the left and right twoThe line with the maximum distance between the edge pixels is the line where the two edge pixels in front of and behind the hip are located, and the coordinates of the line are recorded as y_hAnd TH is a threshold value of a search line.

Preferably, the step of calculating left and right edge point pixels of the human chest based on the human front foreground image and the human skeleton information, and calculating the euclidean distance corresponding to the coordinates in the world coordinate system according to the left and right edge point pixels of the chest to obtain the human chest width includes:

acquiring the line coordinate y of the front and the rear edge point pixels of the chest_b；

Based on the line coordinate y_bCalculating the pixel coordinates of the left and right edge point pixels of the chest corresponding to the front foreground image of the human body, and recording the pixel coordinates of the left edge point of the chest as I (x)_bl，y_b) The pixel coordinate of the right edge point of the chest is marked as I (x)_br，y_b)；

According to the left edge point pixel coordinate I (x) of the chest_bl，y_b) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_wbl，y_wbl，z_wbl) (ii) a According to the right edge point pixel coordinate I (x) of the chest_bl，y_b) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_wbr，y_wbr，z_wbr)；

According to the world coordinate (x) corresponding to the pixel coordinate of the left edge point of the chest_wbl，y_wbl，z_wbl) And world coordinates (x) corresponding to pixel coordinates of right edge point of chest_wbr，y_wbr，z_wbr) Obtaining the chest width of the human body; wherein the human body chest width BW satisfies the following formula:

and the number of the first and second groups,

the step of calculating left and right edge point pixels of the human waist based on the human body front foreground image and the human body skeleton information, and calculating the Euclidean distance corresponding to the coordinates under a world coordinate system according to the left and right edge point pixels of the waist to obtain the human waist width comprises the following steps:

acquiring the line coordinate y of the front and the rear edge point pixels of the waist_w；

Based on the line coordinate y_wCalculating pixel coordinates of the left and right edge point pixels of the waist corresponding to the front foreground image of the human body, and recording the pixel coordinates of the left edge point of the waist as I (x)_wl，y_w) The right edge point pixel coordinate of the waist is denoted as I (x)_wr，y_w)；

According to the left edge point pixel coordinate I (x) of the waist_wl，y_w) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_wwl，y_wwl，z_wwl) (ii) a According to the right edge point pixel coordinate I (x) of the waist_wr，y_w) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_wwr，y_wwr，z_wwr)；

According to the world coordinate (x) corresponding to the pixel coordinate of the left edge point of the waist_wwl，y_wwl，z_wwl) And world coordinate (x) corresponding to right edge point pixel coordinate of waist_wwr，y_wwr，z_wwr) Obtaining the waist width of the human body; wherein the human waist width WW satisfies the following formula:

and the number of the first and second groups,

the method comprises the following steps of calculating left and right edge point pixels of the hip of a human body based on the front foreground image of the human body and the skeleton information of the human body, and calculating the Euclidean distance corresponding to the coordinates under a world coordinate system according to the left and right edge point pixels of the hip so as to obtain the width of the hip of the human body:

acquiring a line coordinate y of the front edge point pixel and the rear edge point pixel of the hip_h；

Based on the line coordinate y_hMeter for measuringCalculating pixel coordinates of the left and right edge point pixels of the hip corresponding to the front foreground image of the human body, and recording the pixel coordinates of the left edge point of the hip as I (x)_hl，y_h) The pixel coordinate of the right edge point of the hip is denoted as I (x)_hr，y_h)；

According to the pixel coordinate I (x) of the left edge point of the hip_hl，y_h) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_whl，y_whl，z_whl) (ii) a According to the pixel coordinate I (x) of the right edge point of the hip_hr，y_h) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_whr，y_whr，z_whr)；

According to the world coordinate (x) corresponding to the pixel coordinate of the left edge point of the hip_whl，y_whl，z_whl) And world coordinates (x) corresponding to pixel coordinates of the right edge point of the hip_whr，y_whr，z_whr) Obtaining the width of the human body hip; wherein the human body hip width HW satisfies the following formula:

preferably, the step of acquiring the front depth image, the side depth image and the skeleton information of the human body of the user to be measured includes:

respectively acquiring a human body front depth image, a human body side depth image and human body skeleton information to be measured by using a somatosensory device; the motion sensing device comprises a collecting device capable of obtaining a depth map and human skeleton information.

According to the human body three-dimensional measurement method based on the depth image, a linear regression equation of chest circumference, chest width and chest thickness, a linear regression equation of waist circumference, waist width and waist thickness and a linear regression equation of hip circumference, hip width and hip thickness are fitted to a preset number of human body three-dimensional data samples by a linear regression method through obtaining a preset number of human body three-dimensional data samples. Then acquiring a human body front depth image, a human body side depth image and human body skeleton information of a user to be measured, and processing the human body front depth image to obtain a human body front foreground image; and processing the human body side depth image to obtain a human body side foreground image, finally calculating to obtain the chest width and chest thickness, the waist width and waist thickness and the hip width and hip thickness of the human body, and respectively bringing the chest width and chest thickness, waist width and waist thickness and hip width and hip thickness to corresponding linear regression equations to obtain the human body three-dimensional data. Therefore, the human body three-dimensional measurement method based on the depth image can accurately calculate the human body three-dimensional data on line, has a simple structure and is convenient to operate, and the human body three-dimensional measurement method based on the depth image is not influenced by illumination and environment, so that the application scene is basically not limited, and the application range of the human body three-dimensional measurement method based on the depth image is widened.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:

FIG. 1 is a schematic flow chart of a method for measuring the three dimensions of a human body based on depth images according to a first embodiment of the present invention;

fig. 2 is a schematic structural diagram illustrating a second embodiment of the present invention, in which a motion sensing device is used to obtain a depth image of a side of a human body and skeleton information of the human body of a user to be measured;

fig. 3 is a schematic structural diagram illustrating a third embodiment of the present invention, in which a motion sensing device is used to obtain a front depth image of a human body and skeleton information of a user to be measured;

FIG. 4 is a diagram illustrating a side view of a human body in a fourth embodiment of the present invention;

fig. 5 is a front perspective view of a human body in a fifth embodiment of the present invention.

Description of the reference numerals

201: a user to be measured;

202: a motion sensing device;

401: the centers of the left and right shoulder joint points are corresponding pixels in a human body side foreground image;

402: corresponding pixels of the spine central joint in the human body side foreground image;

403: the centers of the left hip joint point and the right hip joint point are corresponding pixels in a foreground image of the side surface of the human body;

501: the corresponding pixel of the left shoulder joint point in the front foreground image of the human body;

502: the right shoulder joint point is corresponding to a pixel in a front foreground image of the human body;

503: the spine central joint corresponds to pixels in a front foreground image of a human body;

504: the corresponding pixel of the left hip joint point in the front foreground image of the human body;

505: and the right hip joint point corresponds to a pixel in the front foreground image of the human body.

Detailed Description

The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present invention, are given by way of illustration and explanation only, not limitation.

Referring to fig. 1 to 5, the present invention relates to a method S100 for measuring a three-dimensional circumference of a human body based on a depth image, the method S100 comprising:

s110, obtaining a preset number of human body three-dimensional data samples, and fitting a linear regression equation of chest circumference, chest width and chest thickness, a linear regression equation of waist circumference, waist width and waist thickness and a linear regression equation of hip circumference, hip width and hip thickness to the preset number of human body three-dimensional data samples by using a linear regression method.

In this step, the preset number of human body three-dimensional data samples is not limited, and in order to make the method for measuring human body three-dimensional based on depth images of the present invention more accurate, the human body three-dimensional data samples should preferably include human body three-dimensional data samples of different ages, different sexes, and/or different body types. In addition, the number of samples is not limited, as long as the number of samples can satisfy the linear regression equation of chest circumference, chest width and chest thickness, the linear regression equation of waist circumference, waist width and waist thickness, and the linear regression equation of hip circumference, hip width and hip thickness by using a linear regression method.

Preferably, in this step, the linear regression equation of the bust size to the chest width and the chest thickness is:

BC＝x_b*BW+y_b*BT+z_b(ii) a And the number of the first and second groups,

WC＝x_w*WW+y_w*WT+z_w(ii) a And the number of the first and second groups,

HC＝x_h*HW+y_h*HT+z_h；

HC is hip circumference size, HW is hip width size, HT is hip thickness size, x_hWeight coefficient of hip width in hip circumference linear regression equation, y_hWeight coefficient for hip thickness in hip circumference linear regression equation, z_hAnd (4) an error compensation constant in the hip circumference linear regression equation.

S120, acquiring a human body front depth image, a human body side depth image and human body skeleton information of the user 201 to be measured.

In this step, there is no limitation on how to acquire the human body front depth image, the human body side depth image, and the human body skeleton information of the user 201 to be measured.

Preferably, in this step, referring to fig. 2 and fig. 3, a front depth image of a human body, a side depth image of the human body, and skeleton information of the human body of the user 201 to be measured may be acquired by using the motion sensing device 202. The motion sensing device 202 comprises a collecting device capable of obtaining a depth map and human skeleton information and corresponding supporting software.

S130, processing the front depth image of the human body to obtain a front foreground image of the human body; and processing the depth image of the side face of the human body to obtain a foreground image of the side face of the human body.

In this step, the human body front depth image may include human body front pixels and human body front background pixels. Accordingly, the number of the first and second electrodes,

the human body side depth image comprises human body side pixels and human body side background pixels.

extracting human body front pixels in the human body front depth image;

the human body front foreground graph meets the following formula:

extracting human body side pixels in the human body side depth image;

the human body side face foreground image meets the following formula:

S140, calculating front and rear two edge point pixels of a chest, front and rear two edge point pixels of a waist and front and rear two edge point pixels of a hip of the human body based on the human body side foreground image and the human body skeleton information; calculating to obtain the Euclidean distance corresponding to the coordinates under a world coordinate system according to the front and rear edge point pixels of the chest so as to obtain the chest thickness of the human body; calculating to obtain the Euclidean distance corresponding to the coordinates under the world coordinate system according to the front and rear edge point pixels of the waist so as to obtain the waist thickness of the human body; and calculating to obtain the Euclidean distance corresponding to the coordinates under the world coordinate system according to the front and rear edge point pixels of the hip so as to obtain the thickness of the hip of the human body.

In this step, the step specifically includes:

the human body chest thickness calculation method comprises the following steps:

As shown in fig. 4, the coordinates of the pixel 401 corresponding to the foreground map of the centers of the left and right shoulder joint points on the side of the human body are calculated, and the coordinates of the pixel 401 of the centers of the left and right shoulder joint points are expressed as ((x)_ls+x_rs)/2，(y_ls+y_rs)/2)；

Extracting the pixel 402 coordinate of the spine center joint of the human body skeleton information corresponding to the human body side foreground image, and recording the pixel coordinate of the spine center joint as (x)_sp，y_sp)；

According to the world coordinate (x) corresponding to the anterior edge point pixel of the chest_wbf，y_wbf，2_wbf) And world coordinates (x) corresponding to the posterior border point pixels of the thorax_wbb，y_wbb，2_wbb) Obtaining the chest thickness of the human body; wherein the human body chest thickness BT satisfies the following formula:

preferably, said pixel coordinates (x) based on said left shoulder joint point_ls，y_ls) Pixel coordinate (x) of the right joint point_rs，y_rs) Pixel coordinates ((x) of centers of the left and right shoulder joint points_ls+x_rs)/2，(y_ls+y_rs) /2) and images of the central joints of the spineElement coordinate (x)_sp，y_sp) Determining two front and back edge point pixels of the chest, and recording the front edge point pixel of the chest as I (x)_bf，y_b) The posterior border point pixel of the thorax is denoted as I (x)_bb，y_b) Comprises the following steps:

(y) th from the side foreground map of the human body_ls+y_rs) Line by line,/2 to y_spAnd lines are calculated, the left edge pixel point and the right edge pixel point of the side pixel of the human body in each line are calculated to determine the line with the maximum distance between the left edge pixel point and the right edge pixel point, the line is the line where the front edge pixel point and the rear edge pixel point of the chest are located, and the line coordinate is recorded as y_b。

The human waist thickness calculation method comprises the following steps:

extracting pixel coordinates of the spine central joint of the human body skeleton information corresponding to the human body side foreground image, and recording the pixel coordinates of the spine central joint as (x)_sp，y_sp)。

Extracting the coordinates of pixels 403 corresponding to the left hip joint point and the right hip joint point of the human body skeleton information on the human body side foreground image, and recording the coordinates of the pixels of the left hip joint point as (x)_lh，y_lh) The pixel coordinate of the right hip joint point is (x)_rh，y_rh)。

Calculating the pixel coordinates of the centers of the left hip joint point and the right hip joint point corresponding to the human body side foreground image, and recording the pixel coordinates of the centers of the left hip joint point and the right hip joint point as ((x)_lh+x_rh)/2，(y_lh+y_rh)/2)。

The pixel coordinate based on the spine center joint is noted as (x)_sp，y_sp) Pixel coordinate (x) of the left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) /2) determining two front and rear edge point pixels of the waist, and recording the front edge point pixel of the waist as I (x)_wf，y_w) The back edge point pixel of the waist is denoted as I (x)_wb，y_w)。

According to the front edge point pixel I (x) of the waist_wf，y_w) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_wwf，y_wwf，z_wwf) (ii) a According to the back edge point pixel I (x) of the waist_wb，y_w) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_wwb，y_wwb，z_wwb)。

preferably, the pixel coordinate based on the central joint of the spine is expressed as (x)_sp，y_sp) Pixel coordinate (x) of the left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) /2) determining two front and rear edge point pixels of the waist, and recording the front edge point pixel of the waist as I (x)_wf，y_w) The back edge point pixel of the waist is denoted as I (x)_wb，y_w) Comprises the following steps:

(y) th of the human body side foreground map_sp+(y_lh+y_rh) row/2)/2-TH and row (y)_sp+(y_lh+y_rh) Line-by-line scanning is carried out between the lines of/2)/2 + TH, and left and right edge pixel points of the side pixels of the human body in each line are calculated to determine the line with the maximum distance between the left and right edge pixel points, wherein the line is the line where the front and rear edge pixel points of the waist are located, and the line coordinate is recorded as y_wAnd TH is a threshold value of a search line.

The method for calculating the hip thickness of the human body comprises the following steps:

preferably, said pixel coordinates (x) based on said left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) And/2) determining two front and rear edge point pixels of the hip, and recording the front edge point pixel of the hip as I (x)_hf，y_h) The posterior edge point pixel of the hip is denoted as I (x)_hb，y_h) Comprises the following steps:

(y) th of the human body side foreground map_lh+y_rh) the/2-TH row and the (y) TH_lh+y_rh) Scanning line by line between the/2 + TH lines, calculating left and right edge pixel points of the side pixels of the human body in each line to determine a line with the maximum distance between the left and right edge pixel points, wherein the line is the line where the front and rear edge pixel points of the hip are located, and recording the line coordinate as y_hAnd TH is a threshold value of a search line.

S150, calculating left and right edge point pixels of a chest, left and right edge point pixels of a waist and left and right edge point pixels of a hip of the human body based on the front foreground image of the human body and the skeleton information of the human body; calculating to obtain Euclidean distance corresponding to coordinates under a world coordinate system according to the left and right edge point pixels of the chest so as to obtain the chest width of the human body; calculating to obtain Euclidean distance corresponding to coordinates under a world coordinate system according to the left and right edge point pixels of the waist so as to obtain the waist width of the human body; and calculating to obtain the Euclidean distance corresponding to the coordinates under the world coordinate system according to the left and right edge point pixels of the hip so as to obtain the width of the human hip.

In this step, as shown in fig. 5, the calculation method may be specifically as follows:

calculating the chest width:

Based on the line coordinate y_bCalculating the pixel coordinates of the left and right edge point pixels of the chest corresponding to the front foreground image of the human body, and calculating the left edge of the chestEdge point pixel coordinate is denoted as I (x)_bl，y_b) The pixel coordinate of the right edge point of the chest is marked as I (x)_br，y_b)；

it should be noted that, for how to determine the left and right edge point pixels of the chest, reference may be made to the foregoing calculation method related to the human body side foreground map and fig. 5, where fig. 5 shows a human body front foreground map diagram, where there are a pixel 501 corresponding to the left shoulder joint point in the human body front foreground map, a pixel 502 corresponding to the right shoulder joint point in the human body front foreground map, a pixel 503 corresponding to the spine center joint in the human body front foreground map, a pixel 504 corresponding to the left hip joint point in the human body front foreground map, and a pixel 505 corresponding to the right hip joint point in the human body front foreground map, respectively.

Calculating the waist width:

calculating the hip width:

Based on the line coordinate y_hCalculating pixel coordinates of the left and right edge point pixels of the hip corresponding to the front foreground image of the human body, and recording the pixel coordinates of the left edge point of the hip as I (x)_hl，y_h) The pixel coordinate of the right edge point of the hip is denoted as I (x)_hr，y_h)；

According to the left of the hipWorld coordinate (x) corresponding to edge point pixel coordinate_whl，y_whl，z_whl) And world coordinates (x) corresponding to pixel coordinates of the right edge point of the hip_whr，y_whr，z_whr) Obtaining the width of the human body hip; wherein the human body hip width HW satisfies the following formula:

s160, substituting the calculated chest thickness and the calculated chest width of the human body into a linear regression equation of the chest circumference, the chest width and the chest thickness, substituting the calculated waist thickness and the calculated waist width of the human body into a linear regression equation of the waist circumference, the waist width and the waist thickness, and substituting the calculated hip thickness and the calculated hip width of the human body into a linear regression equation of the hip circumference, the hip width and the hip thickness to obtain the three-circumference data of the human body.

In this step, the chest thickness BT, waist thickness WT and hip thickness HT of the user to be measured obtained in step S140, and the chest width BW, waist width WW and hip width HW of the user to be measured obtained in step S150 may be respectively substituted into the linear regression equation in step S110 to obtain the current circumference of the user to be measured.

It will be understood that the above embodiments are merely exemplary embodiments taken to illustrate the principles of the present invention, which is not limited thereto. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and substance of the invention, and these modifications and improvements are also considered to be within the scope of the invention.

Claims

1. A human body three-dimensional measurement method based on a depth image is characterized by comprising the following steps:

the method comprises the following steps of calculating front and rear two edge point pixels of a human chest based on the human body side foreground image and the human body skeleton information, and calculating to obtain the Euclidean distance of coordinates under a corresponding world coordinate system according to the front and rear two edge point pixels of the chest so as to obtain the thickness of the human chest:

Based on pixel coordinates (x) of the left shoulder joint point_ls，y_ls) Pixel coordinate (x) of the right shoulder joint point_rs，y_rs) Pixel coordinates ((x) of centers of the left and right shoulder joint points_ls+x_rs)/2，(y_ls+y_rs) /2) and the pixel coordinates (x) of the central joint of the spine_sp，y_sp) Determining two front and back edge point pixels of the chest, and recording the front edge point pixel of the chest as I (x)_bf，y_b) The posterior border point pixel of the thorax is denoted as I (x)_bb，y_b)；

According to the anterior border point pixel I (x) of the chest_bf，y_b) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_wbf，y_wbf，z_wbf) (ii) a According to the chestTrailing edge point pixel I (x)_bb，y_b) Calculating to obtain world coordinates under the corresponding world coordinate system, and recording the world coordinates as (x)_wbb，y_wbb，z_wbb)；

According to the world coordinate (x) corresponding to the anterior edge point pixel of the chest_wbf，y_wbf，z_wbf) And world coordinates (x) corresponding to the posterior border point pixels of the thorax_wbb，y_wbb，z_wbb) Obtaining the chest thickness of the human body; wherein the human body chest thickness BT satisfies the following formula:

and the number of the first and second groups,

The pixel coordinate based on the spine center joint is noted as (x)_sp，y_sp) Pixel coordinate (x) of the left hip joint point_lh，y_lh) Pixel coordinate (x) of the right hip joint point_rh，y_rh) And the above-mentionedPixel coordinates ((x) of centers of right and left hip joint points_lh+x_rh)/2，(y_lh+y_rh) /2) determining two front and rear edge point pixels of the waist, and recording the front edge point pixel of the waist as I (x)_wf，y_w) The back edge point pixel of the waist is denoted as I (x)_wb，y_w)；

and the number of the first and second groups,

From the leading edge point pixel I (x) of the hip_hf，y_h) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_whf，y_whf，z_whf) (ii) a According to the posterior edge point pixel I (x) of the hip_hb，y_h) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_whb，y_whb，z_whb)；

substituting the calculated chest thickness and chest width of the human body into a linear regression equation of the chest circumference, the chest width and the chest thickness, substituting the calculated waist thickness and waist width of the human body into a linear regression equation of the waist circumference, the waist width and the waist thickness, substituting the calculated hip thickness and hip width of the human body into a linear regression equation of the hip circumference, the hip width and the hip thickness, and obtaining the three-circumference data of the human body.

2. The method of claim 1, wherein the predetermined number of body volume data samples comprises body volume data samples of different ages, different sexes, and/or different sizes.

3. The measurement method according to claim 2, wherein the linear regression equation of the bust size with the chest width and the chest thickness is:

BC＝x_b*BW+y_b*BT+z_b(ii) a And the number of the first and second groups,

WC＝x_w*WW+y_w*WT+z_w(ii) a And the number of the first and second groups,

HC＝x_h*HW+y_h*HT+z_h；

WC is waist size, WW is waist width size, WT is waist thickness size, x_wIs the weight coefficient of waist width in the waist circumference linear regression equation, y_wIs the weight coefficient of waist thickness in the waist circumference linear regression equationz_wAn error compensation constant in the waist linear regression equation is obtained;

4. The measurement method according to any one of claims 1 to 3, wherein the human body front depth image comprises human body front pixels and human body front background pixels; and the number of the first and second groups,

5. The measurement method according to claim 4, wherein the front depth image of the human body is processed to obtain a front foreground image of the human body; and processing the depth image of the side face of the human body to obtain a foreground image of the side face of the human body, wherein the step of processing the depth image of the side face of the human body to obtain the foreground image of the side face of the human body comprises the following steps:

extracting human body front pixels in the human body front depth image;

the human body front foreground graph meets the following formula:

wherein, F is_f(x_f，y_f) Is a front foreground map of the human body, I (x)_f，y_f) The coordinate in the human body front depth image is (x)_f，y_f) A pixel of (a); said H_fThe human body front pixel set is the human body front depth image; b is_fIs the front of the human bodyThe human body front background pixel set in the depth image; and the number of the first and second groups,

extracting human body side pixels in the human body side depth image;

the human body side face foreground image meets the following formula:

6. The measurement method according to claim 5, wherein the pixel coordinates (x) based on the left shoulder joint point_ls，y_ls) Pixel coordinate (x) of the right shoulder joint point_rs，y_rs) Pixel coordinates ((x) of centers of the left and right shoulder joint points_ls+x_rs)/2，(y_ls+y_rs) /2) and the pixel coordinates (x) of the central joint of the spine_sp，y_sp) Determining two front and back edge point pixels of the chest, and recording the front edge point pixel of the chest as I (x)_bf，y_b) The posterior border point pixel of the thorax is denoted as I (x)_bb，y_b) Comprises the following steps:

(y) th from the side foreground map of the human body_ls+y_rs) Line by line,/2 to y_spAnd lines are calculated, the left edge pixel point and the right edge pixel point of the side pixel of the human body in each line are calculated to determine the line with the maximum distance between the left edge pixel point and the right edge pixel point, the line is the line where the front edge pixel point and the rear edge pixel point of the chest are located, and the line coordinate is recorded as y_b(ii) a And the number of the first and second groups,

the pixel coordinate based on the spine central joint is marked as (x)_sp，y_sp) Pixel coordinate (x) of the left hip joint point_lh，y_lh)、Pixel coordinate (x) of the right hip joint point_rh，y_rh) And pixel coordinates ((x) of centers of the left and right hip joint points_lh+x_rh)/2，(y_lh+y_rh) /2) determining two front and rear edge point pixels of the waist, and recording the front edge point pixel of the waist as I (x)_wf，y_w) The back edge point pixel of the waist is denoted as I (x)_wb，y_w) Comprises the following steps:

7. The measurement method according to claim 6, wherein the step of calculating left and right edge point pixels of the human chest based on the human front foreground map and the human skeleton information, and calculating the Euclidean distance corresponding to coordinates in a world coordinate system according to the left and right edge point pixels of the chest to obtain the human chest width comprises:

and the number of the first and second groups,

and the number of the first and second groups,

According toPixel coordinate I (x) of the left edge point of the hip_hl，y_h) Calculating to obtain world coordinates in the corresponding world coordinate system, and recording as (x)_whl，y_whl，z_whl) (ii) a According to the pixel coordinate I (x) of the right edge point of the hip_hr，y_h) The world coordinates of the corresponding world coordinate system are calculated and are marked as (x)_whr，y_whr，z_whr)；

8. the measurement method according to any one of claims 1 to 3, wherein the step of acquiring the front depth image, the side depth image and the skeleton information of the human body of the user to be measured comprises: