CN112950788B - Human body reconstruction and garment customization data acquisition method based on single image - Google Patents
Human body reconstruction and garment customization data acquisition method based on single image Download PDFInfo
- Publication number
- CN112950788B CN112950788B CN202110301963.5A CN202110301963A CN112950788B CN 112950788 B CN112950788 B CN 112950788B CN 202110301963 A CN202110301963 A CN 202110301963A CN 112950788 B CN112950788 B CN 112950788B
- Authority
- CN
- China
- Prior art keywords
- human body
- point
- points
- shoulder
- reconstruction
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2210/00—Indexing scheme for image generation or computer graphics
- G06T2210/12—Bounding box
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2210/00—Indexing scheme for image generation or computer graphics
- G06T2210/16—Cloth
Abstract
The invention discloses a human body reconstruction and garment customization data acquisition method based on a single image, which comprises the steps of inputting a single color image into a human body reconstruction network to reconstruct a human body mesh model; extracting vertexes from the human body mesh model to obtain a human body surface point cloud model after down-sampling; quickly selecting human body key points from a human body surface point cloud model through geometric constraint information; solving the human body clothing customization data by using the human body key points; by combining a deep learning strategy, the method realizes the reconstruction of the human body three-dimensional model from a single image, solves the problem of hardware dependence in three-dimensional reconstruction, and reduces the human body three-dimensional reconstruction threshold; meanwhile, the human body key points determined based on the geometric constraint are used for selecting the measurement object, so that the precision and the speed of the human body measurement data are improved.
Description
Technical Field
The invention relates to the technical field of three-dimensional human body reconstruction, anthropometry and point cloud data processing, in particular to a single image-based human body reconstruction and garment customization data acquisition method.
Background
At present, for the related achievements of a human body measurement data acquisition system, a special human body data acquisition system is often required to be built, and the system comprises a bracket, a rotary platform, various sensors and the like; it is difficult to directly estimate the measurement parameters corresponding to the three-dimensional human body model from a single input color picture. For example, a method of dimensional measurement and fitting based on a three-dimensional human body model is mainly used for processing a three-dimensional model collected from Kinect, and regressing measurement data of each part of a human body using a neural network; however, it is difficult to use for processing common image input.
The human body size measurement method based on three-dimensional scanning uses binocular stereo vision to realize human body point cloud reconstruction and realizes human body data measurement on the basis; however, the self-developed binocular stereo vision based three-dimensional scanning device is used in the text and is difficult to popularize.
Disclosure of Invention
This section is for the purpose of summarizing some aspects of embodiments of the invention and to briefly introduce some preferred embodiments. In this section, as well as in the abstract and the title of the invention of this application, simplifications or omissions may be made to avoid obscuring the purpose of the section, the abstract and the title, and such simplifications or omissions are not intended to limit the scope of the invention.
The present invention has been made in view of the above-mentioned conventional problems.
Therefore, the invention provides a single image-based human body reconstruction and garment customization data acquisition method, which can effectively solve the problems of time consumption and high cost in building a large human body scanning system.
In order to solve the technical problems, the invention provides the following technical scheme: the method comprises the steps of inputting a single color image to a human body reconstruction network to reconstruct a human body mesh model; extracting vertexes from the human body mesh model to obtain a downsampled human body surface point cloud model; quickly selecting human body key points from the human body surface point cloud model through geometric constraint information; and solving the human body clothing customization data by using the human body key points.
As a preferable scheme of the method for human body reconstruction and garment customization data acquisition based on a single image, the method comprises the following steps: the human reconstruction network includes a set of 3D codecs and a set of 2D codecs.
As a preferable scheme of the method for human body reconstruction and garment customization data acquisition based on a single image, the method comprises the following steps: the method also comprises the steps of extracting implicit pixel characteristics in the two-dimensional image through the 2D codec, and then mapping the implicit pixel characteristics to an implicit voxel space through the 3D codec.
As a preferable scheme of the method for human body reconstruction and garment customization data acquisition based on a single image, the method comprises the following steps: reconstructing the human body grid model comprises aligning and fusing implicit voxel space characteristics and the implicit pixel characteristics, and transmitting an aligning and fusing result to an implicit surface function; and judging the effective segmentation of the human body surface in the three-dimensional space through the implicit surface function so as to generate a human body mesh model containing human figure details.
As a preferable scheme of the method for human body reconstruction and garment customization data acquisition based on a single image, the method comprises the following steps: the key points of the human body comprise a shoulder point, a back neck point, a breast tip point and a back waist point; and determining the human body key points by using the three-dimensional coordinates of the human body key points as the geometric constraint information and combining the point cloud model after the posture adjustment.
As a preferable scheme of the method for human body reconstruction and garment customization data acquisition based on a single image, the method comprises the following steps: the posture adjustment comprises the steps of controlling the scale of the output human body surface point cloud through complete random sampling, and completing the down-sampling of the human body surface point cloud; creating a direction bounding box for the surface point cloud after the down-sampling; and moving the surface point cloud to a world coordinate origin by using the direction bounding box, ensuring that the direction of a coordinate system of the bounding box is consistent with that of the world coordinate system, and finishing the down-sampling and posture adjustment of the human body surface point cloud model.
As a preferable scheme of the method for human body reconstruction and garment customization data acquisition based on a single image, the method comprises the following steps: the three-dimensional coordinates of the human body key points comprise that the shoulder points are solved: screening a candidate point set corresponding to the shoulder area by using the shoulder joint points and the point cloud, then counting the Euclidean distances from each candidate point of the candidate point set to the shoulder joint points, and selecting the point with the largest Euclidean distance as the shoulder point; solving the back neck point: screening a candidate point set corresponding to a neck region by using neck joint points and shoulder points, then constructing a plane passing through left and right shoulder points and the neck joint points, counting the distance from each candidate point of the candidate point set to the plane, and taking the point with the shortest plane distance as the nape point; solving the breast cusp: performing four-neighborhood search operation on a candidate point set in a chest range, solving the maximum value of local coordinates, and screening the breast apex candidate point set according to the characteristic that a breast apex connecting line segment and a shoulder point connecting line segment are approximately parallel to each other to obtain a breast apex; solving the waist point: and taking a longitudinal cutting plane passing through the center point of the cube surrounded by the human body, calculating the distance from each candidate point in the back candidate point set to the longitudinal cutting plane, and taking the point closest to the longitudinal cutting plane as the waist point.
As a preferable scheme of the method for human body reconstruction and garment customization data acquisition based on a single image, the method comprises the following steps: solving the human body garment customization data comprises projecting neighboring spatial points onto the same cutting plane determined by the human body key points, and obtaining a smooth approximate curve through curve fitting; and sampling the approximate curve to obtain a plurality of groups of sampling points, wherein the sum of continuous Euclidean distances among the plurality of groups of sampling points is the customized data of the human body garment.
The invention has the beneficial effects that: the method is combined with a deep learning strategy, so that the reconstruction of the three-dimensional model of the human body from a single image is realized, the problem of hardware dependence in three-dimensional reconstruction is solved, and the three-dimensional reconstruction threshold of the human body is reduced; meanwhile, the human body key points determined based on the geometric constraint are used for selecting the measurement object, so that the precision and the speed of the human body measurement data are improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise. Wherein:
fig. 1 is a schematic flow chart of a human body reconstruction and garment customization data acquisition method based on a single image according to a first embodiment of the present invention;
fig. 2 is a schematic structural diagram of a human body reconstruction network of a method for acquiring data of human body reconstruction and garment customization based on a single image according to a first embodiment of the present invention;
fig. 3 is a single color human body whole body schematic diagram of a human body reconstruction and garment customization data acquisition method based on a single image according to a first embodiment of the present invention;
fig. 4 is a schematic diagram of a three-dimensional mesh generated after reconstruction in a front view angle by a single image-based human body reconstruction and garment customization data acquisition method according to a first embodiment of the present invention;
fig. 5 is a schematic diagram of a three-dimensional mesh generated after reconstruction at a side rear view angle of a method for human body reconstruction and garment customization data acquisition based on a single image according to a first embodiment of the present invention;
fig. 6 is a schematic view of visualization of key points and measurement objects of a human body at different viewpoints of a method for acquiring data of human body reconstruction and garment customization based on a single image according to a first embodiment of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, specific embodiments accompanied with figures are described in detail below, and it is apparent that the described embodiments are a part of the embodiments of the present invention, not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, shall fall within the protection scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways than those specifically described and will be readily apparent to those of ordinary skill in the art without departing from the spirit of the present invention, and therefore the present invention is not limited to the specific embodiments disclosed below.
Furthermore, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
The present invention will be described in detail with reference to the drawings, wherein the cross-sectional views illustrating the structure of the device are not enlarged partially in general scale for convenience of illustration, and the drawings are only exemplary and should not be construed as limiting the scope of the present invention. In addition, the three-dimensional dimensions of length, width and depth should be included in the actual fabrication.
Meanwhile, in the description of the present invention, it should be noted that the terms "upper, lower, inner and outer" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of describing the present invention and simplifying the description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation and operate, and thus, cannot be construed as limiting the present invention. Furthermore, the terms first, second, or third are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
The terms "mounted, connected" and "connected" in the present invention are to be construed broadly, unless otherwise explicitly specified or limited, for example: can be fixedly connected, detachably connected or integrally connected; they may be mechanically, electrically, or directly connected, or indirectly connected through intervening media, or may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in a specific case to those of ordinary skill in the art.
Example 1
Referring to fig. 1 to 6, a first embodiment of the present invention provides a method for human body reconstruction and garment customization data acquisition based on a single image, including:
s1: and a human body mesh model is reconstructed by inputting a single color image into the human body reconstruction network.
As shown in fig. 2, the human reconstruction network includes a set of 3D codecs and a set of 2D codecs; the network can effectively process a single input image and restore the space geometric structure of a human body from the image; the network has the main idea that the spatial features and the pixel features extracted from the image are integrated, and a group of multilayer perceptrons is used for fitting an implicit surface function to obtain the probability that the pixels in the current projection space are located in the surface of a human body; and further, effectively refining the voxel space to obtain a complete three-dimensional model of the human body.
The 2D codec is responsible for extracting implicit pixel features in the two-dimensional image, and the 3D codec is responsible for mapping the implicit pixel features to an implicit voxel space.
Specifically, the steps of reconstructing the human mesh model are as follows:
(1) aligning and fusing the implicit voxel space characteristic and the implicit pixel characteristic, and transmitting a fusion result to an implicit surface function;
it should be noted that the implicit surface function is fitted in the network by using a set of multi-layer perceptrons, for the input feature K, there is a set of implicit surface functions M (×), and the probability P ═ M (K) that the corresponding position under the feature is located in the surface of the human body can be obtained, and since the deep neural network can fit any function, this surface function is fitted by using a set of multi-layer perceptrons.
The implicit surface function is transmitted with the result of fusion of two features (similar to feature splicing, m + n-dimensional features after n-dimensional voxel features and m-dimensional pixel features are spliced are jointly input into the implicit surface function by combining camera parameters).
(2) And judging the effective segmentation of the human body surface in the three-dimensional space through an implicit surface function so as to generate a human body mesh model containing human figure details.
As shown in fig. 3, 4, and 5, fig. 3 is an input single color image, and fig. 4 and 5 are three-dimensional meshes (human mesh models) generated by reconstruction at different viewing angles, respectively.
S2: and extracting vertexes from the human body mesh model to obtain a downsampled human body surface point cloud model.
The grid model is easy to generate holes, so that the data acquisition precision is influenced, and the point cloud model has the characteristics of simplicity and easiness in processing.
And extracting mesh vertexes of the generated human body mesh model, acquiring a human body surface point cloud model, and controlling the scale of the output human body surface point cloud through complete random sampling to realize the down-sampling of the human body surface point cloud.
S3: and quickly selecting human body key points from the human body surface point cloud model through geometric constraint information.
And creating a direction bounding box for the human body surface point cloud subjected to downsampling, moving the point cloud to the world coordinate origin by using the direction bounding box, and ensuring that the direction of a coordinate system of the bounding box is consistent with that of the world coordinate system, so that downsampling and posture adjustment of a human body surface point cloud model are realized.
Then, using the three-dimensional coordinates of the human body joint points as geometric constraint information, and determining human body key points by combining the point cloud model after the posture adjustment; wherein, the key points of the human body to be solved comprise a shoulder point, a back neck point, a breast tip point and a back waist point.
Specifically, (1) when solving the shoulder point, firstly screening a candidate point set corresponding to the shoulder area by using the shoulder joint point and the point cloud, then counting the Euclidean distances from each point to the shoulder joint point, and selecting the point with the maximum Euclidean distance as the shoulder point.
(2) When solving the back neck point, screening a candidate point set by using the neck joint point and the shoulder points, then constructing a plane passing through the left and right shoulder points and the neck joint point, counting the distance from the candidate point to the plane, and taking the point with the shortest distance as the back neck point.
(3) When the nipple point is solved, four-neighborhood search operation is carried out on a candidate point set in the chest range, the maximum value of a local coordinate is solved, a left nipple point and a right nipple point are connected to obtain a line segment L1, a left shoulder point and a right shoulder point are connected to obtain a line segment L2, and the characteristic that L1 is approximately parallel to L2 exists according to the space characteristics of a human body when the human body is normally erected; the characteristic is utilized to screen the breast apex candidate point set to obtain the breast apex points.
(4) And when solving the waist point, taking a longitudinal cutting plane passing through the center point of the cube surrounded by the human body, calculating the distance from each element in the back candidate point set to the longitudinal cutting plane, and taking the point closest to the longitudinal cutting plane as the waist point.
Preferably, compared with the method for directly solving the circumference of the human body to be measured by using the circular cutting method, the method for determining the key points of the human body can more accurately find the circular cutting plane of the human body which most meets the measurement specification.
S4: and solving the human body clothing customization data by using the human body key points.
When the body clothing customization data such as the chest circumference, the waist circumference and the like are predicted, the adjacent space points are projected to the same cutting plane determined by the key points, a smooth approximate curve is obtained by curve fitting, and the sum of continuous Euclidean distances among a plurality of groups of sampling points obtained by curve sampling can be used for representing the body circumference measurement value.
Specifically, when the shoulder width is solved, a plane on a three-dimensional space is constructed by using a left shoulder point, a right shoulder point and a back neck point, a group of surface point sets adjacent to the segmented section are obtained by using threshold screening in a candidate point set according to a path of the left shoulder point, the back neck point and the right shoulder point, curve fitting is performed on the group of point sets, and then a group of three-dimensional coordinate point sets are obtained by using equidistant sampling, so that the shoulder width solving can be degraded into the sum of continuous Euclidean distances of fitted curve sampling points.
Fig. 6 shows the human body key point and the measured object visualization, wherein the continuous curve represents the human body data object to be measured. The triangular points are selected key points of the human body. The pentagonal points are the human body joint points.
According to the method, a voxel model predicted from a deep neural network is refined by using an implicit function, a fine three-dimensional grid is generated, and then a method based on geometric constraint is used for determining key points of a human body from surface point cloud of grid model downsampling; compared with the traditional method for slicing the point cloud model layer by layer, the method for solving the circumference by using the human body key points has lower computational complexity and more excellent solving precision, and can realize the acquisition of the clothing customization data with high quality.
Example 2
In order to verify and explain the technical effect adopted in the method, the embodiment selects a data acquisition algorithm and adopts the method to perform comparison test, and compares the test results by means of scientific demonstration to verify the real effect of the method.
Data acquisition algorithm (image-based human reconstruction and data acquisition algorithm): (to illustrate existing deficiencies, such as accuracy, effectiveness, endurance, errors, etc.).
In order to verify that the method has higher measurement accuracy compared with a data acquisition algorithm, in the embodiment, the data acquisition algorithm and the method are adopted to respectively carry out human body data acquisition work comparison on four male experimental subjects with obvious body type differences.
In order to avoid subjective errors of manual measurement, the experiment firstly carries out manual measurement on human body data, takes the manual measurement as a reference true value, and respectively compares the proposed method and a comparison algorithm with a manual measurement true value; according to the standard of human body tailoring, a special flexible rule for a garment design major is used for respectively measuring the upper body data of the four experimental subjects in normal standing posture, such as shoulder width, chest circumference, waist circumference, height and the like, so as to obtain experimental results; by adopting the method, a single human body upright high-definition image is collected from the forward direction to be used as input, and the human body measurement data under the viewpoint is predicted; under the same human body posture, the forward and lateral images of the human body are simultaneously acquired, contour extraction and binarization are carried out, the obtained result is input into a data acquisition algorithm, and comparison test data are obtained, wherein the specific experimental data are as shown in the following table.
Table 1: the experimental results are shown in a comparison table.
According to experimental results, the accuracy of the obtained measurement result is higher than that of the human body data predicted by directly using the binary contour feature points through segmenting the reconstructed model by the three-dimensional key points; meanwhile, the invention can be applied to the human body data acquisition of only a single input image, and the human body data is calculated by using the binary contour, so that at least two characteristic postures of shot images are required.
It should be noted that the above-mentioned embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, which should be covered by the claims of the present invention.
Claims (2)
1. A human body reconstruction and garment customization data acquisition method based on a single image is characterized by comprising the following steps:
reconstructing a human body mesh model by inputting a single color image to a human body reconstruction network;
the human body reconstruction network comprises a group of 3D codecs and a group of 2D codecs;
extracting implicit pixel characteristics in the two-dimensional image through the 2D coder, and then mapping the implicit pixel characteristics to an implicit voxel space through the 3D coder;
reconstructing the human mesh model includes:
carrying out alignment fusion on the implicit voxel space characteristic and the implicit pixel characteristic, and transmitting an alignment fusion result to an implicit surface function;
judging effective segmentation of the human body surface in the three-dimensional space through the implicit surface function so as to generate a human body mesh model containing human figure details;
extracting vertexes from the human body mesh model to obtain a human body surface point cloud model after down-sampling;
quickly selecting human body key points from the human body surface point cloud model through geometric constraint information;
the key points of the human body comprise a shoulder point, a back neck point, a breast tip point and a back waist point;
determining human body key points by using three-dimensional coordinates of the human body key points as the geometric constraint information and combining the point cloud model after the posture adjustment;
the attitude adjustment includes:
controlling the scale of outputting the human body surface point cloud through complete random sampling to finish the downsampling of the human body surface point cloud;
creating a direction bounding box for the surface point cloud after the down-sampling;
moving the surface point cloud to a world coordinate origin by using the direction bounding box, ensuring that the direction of a coordinate system of the bounding box is consistent with that of the world coordinate system, and finishing the down-sampling and the posture adjustment of a human body surface point cloud model;
the three-dimensional coordinates of the human body key points comprise:
solving the shoulder points: screening a candidate point set corresponding to the shoulder area by using the shoulder joint points and the point cloud, then counting the Euclidean distances from each candidate point of the candidate point set to the shoulder joint points, and selecting the point with the largest Euclidean distance as the shoulder point;
solving the back neck point: screening a candidate point set corresponding to a neck region by using neck joint points and shoulder points, then constructing a plane passing through left and right shoulder points and the neck joint points, counting the distance from each candidate point of the candidate point set to the plane, and taking the point with the shortest plane distance as the nape point;
solving the breast cusp: performing four-neighborhood search operation on a candidate point set in a chest range, solving the maximum value of local coordinates, and screening the breast apex candidate point set according to the characteristic that a breast apex connecting line segment and a shoulder point connecting line segment are approximately parallel to each other to obtain a breast apex;
solving the waist point: taking a longitudinal cutting plane passing through the center point of the cube surrounded by the human body, calculating the distance from each candidate point in the back candidate point set to the longitudinal cutting plane, and taking the point closest to the longitudinal cutting plane as the waist point;
and solving the human body clothing customization data by using the human body key points.
2. The single image-based human body reconstruction and garment customization data acquisition method according to claim 1, wherein solving the human body garment customization data comprises:
projecting the neighboring space points to the same cutting plane determined by the human body key points, and obtaining a smooth approximate curve through curve fitting;
and sampling the approximate curve to obtain a plurality of groups of sampling points, wherein the sum of continuous Euclidean distances among the plurality of groups of sampling points is the customized data of the human body garment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110301963.5A CN112950788B (en) | 2021-03-22 | 2021-03-22 | Human body reconstruction and garment customization data acquisition method based on single image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110301963.5A CN112950788B (en) | 2021-03-22 | 2021-03-22 | Human body reconstruction and garment customization data acquisition method based on single image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112950788A CN112950788A (en) | 2021-06-11 |
| CN112950788B true CN112950788B (en) | 2022-07-19 |
Family
ID=76227533
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110301963.5A Active CN112950788B (en) | 2021-03-22 | 2021-03-22 | Human body reconstruction and garment customization data acquisition method based on single image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112950788B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108010134A (en) * | 2017-11-29 | 2018-05-08 | 湘潭大学 | A kind of real-time three-dimensional virtual fit method based on mobile terminal |
| CN110175897A (en) * | 2019-06-03 | 2019-08-27 | 广东元一科技实业有限公司 | A kind of 3D synthesis fitting method and system |
| CN111340944A (en) * | 2020-02-26 | 2020-06-26 | 清华大学 | Single-image human body three-dimensional reconstruction method based on implicit function and human body template |
| CN112330795A (en) * | 2020-10-10 | 2021-02-05 | 清华大学 | Human body three-dimensional reconstruction method and system based on single RGBD image |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10885708B2 (en) * | 2018-10-16 | 2021-01-05 | Disney Enterprises, Inc. | Automated costume augmentation using shape estimation |
-
2021
- 2021-03-22 CN CN202110301963.5A patent/CN112950788B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108010134A (en) * | 2017-11-29 | 2018-05-08 | 湘潭大学 | A kind of real-time three-dimensional virtual fit method based on mobile terminal |
| CN110175897A (en) * | 2019-06-03 | 2019-08-27 | 广东元一科技实业有限公司 | A kind of 3D synthesis fitting method and system |
| CN111340944A (en) * | 2020-02-26 | 2020-06-26 | 清华大学 | Single-image human body three-dimensional reconstruction method based on implicit function and human body template |
| CN112330795A (en) * | 2020-10-10 | 2021-02-05 | 清华大学 | Human body three-dimensional reconstruction method and system based on single RGBD image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112950788A (en) | 2021-06-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110874864B (en) | Method, device, electronic equipment and system for obtaining three-dimensional model of object | |
| CN110782524B (en) | Indoor three-dimensional reconstruction method based on panoramic image | |
| CN103971408B (en) | Three-dimensional facial model generating system and method | |
| CN108027984B (en) | Method and system for detecting and combining structural features in 3D reconstruction | |
| CN104504671B (en) | Method for generating virtual-real fusion image for stereo display | |
| Hong et al. | Stereopifu: Depth aware clothed human digitization via stereo vision | |
| CN102697508B (en) | Method for performing gait recognition by adopting three-dimensional reconstruction of monocular vision | |
| Furukawa et al. | Accurate, dense, and robust multiview stereopsis | |
| CN109242954B (en) | Multi-view three-dimensional human body reconstruction method based on template deformation | |
| US7085409B2 (en) | Method and apparatus for synthesizing new video and/or still imagery from a collection of real video and/or still imagery | |
| CN107767442A (en) | A kind of foot type three-dimensional reconstruction and measuring method based on Kinect and binocular vision | |
| WO2015188684A1 (en) | Three-dimensional model reconstruction method and system | |
| CN109427007B (en) | Virtual fitting method based on multiple visual angles | |
| CN103021017A (en) | Three-dimensional scene rebuilding method based on GPU acceleration | |
| CN110148217A (en) | A kind of real-time three-dimensional method for reconstructing, device and equipment | |
| CN113366491B (en) | Eyeball tracking method, device and storage medium | |
| CN108053476A (en) | A kind of human parameters measuring system and method rebuild based on segmented three-dimensional | |
| EP4036863A1 (en) | Human body model reconstruction method and reconstruction system, and storage medium | |
| Oswald et al. | A convex relaxation approach to space time multi-view 3d reconstruction | |
| CN115546442A (en) | Multi-view stereo matching reconstruction method and system based on perception consistency loss | |
| Enciso et al. | Synthesis of 3D faces | |
| Li et al. | Three-dimensional motion estimation via matrix completion | |
| CN112927348A (en) | High-resolution human body three-dimensional reconstruction method based on multi-viewpoint RGBD camera | |
| Lee et al. | Inference of segmented overlapping surfaces from binocular stereo | |
| CN115761116B (en) | Three-dimensional face reconstruction method based on perspective projection of monocular camera |
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 |