Three-dimensional human body measurement method and device based on 3D structured light and double-layer surface model

Abstract

The invention discloses a three-dimensional human body measurement method and device based on 3D structured light and a double-layer surface model. In the double-layer surface model, the inner-layer model captures the basic form and posture of the human body, and the outer-layer model captures the detail features of non-rigid objects on the surface of the human body. The 3D structured light camera module emits light with structural characteristics to the surface of a to-be-measured human body, captures light reflected by the surface of the human body to generate depth map data of the to-be-measured human body, and further converts the depth map data into a three-dimensional point cloud data model. In the matching process of the three-dimensional point cloud data model and the double-layer surface model, the double-layer surface model is adjusted based on the data difference and the double-layer node diagram of the double-layer surface model. And finally, obtaining three-dimensional measurement data of the human body to be measured according to the adjusted double-layer surface model. The 3D structured light and the double-layer surface model are used for accurately reconstructing the three-dimensional form and the additional object of the human body, three-dimensional accurate measurement is achieved, and the measurement cost is remarkably reduced compared with 3D scanning equipment.

Description

Technical Field

[0001] The present application relates to the field of computer vision technology, and in particular to a three-dimensional human body measurement method and device based on 3D structured light and a double-layer surface model. Background Art

[0002] Traditional anthropometric methods, including manual measurement or two-dimensional image-based technology, have problems of insufficient accuracy and cumbersome operation. Manual measurement methods rely on the skills and experience of the operator, which is prone to human errors. Two-dimensional measurement technology cannot accurately capture the three-dimensional form of the human body, which limits the accuracy and credibility of the measurement results. These limitations are particularly prominent in industries such as medicine, sports, and clothing design, because these fields have extremely high requirements for measurement accuracy to ensure that customized products and services can be provided. For example, in the medical field, personalized orthopedic braces and prostheses are customized for patients based on measured human body dimensions. In the field of sports science, based on the measured body size data of athletes, athletes' training plans and competitive performance are optimized, such as providing shaping suggestions or developing personalized sportswear. In the field of clothing design, relying on body shape data, personalized clothing design and customization needs are supported, and the comfort and fit of clothing are improved. Although 3D scanning equipment can provide high-precision three-dimensional anthropometric data, such equipment is usually expensive. Summary of the invention

[0003] Based on the above problems, the present application provides a three-dimensional human body measurement method and device based on 3D structured light and double-layer surface model, aiming to achieve low-cost three-dimensional accurate measurement of the human body by accurately reconstructing the three-dimensional shape of the human body and attached objects using 3D structured light and double-layer surface model.

[0004] The embodiments of the present application disclose the following technical solutions:

[0005] In a first aspect, the present application provides a three-dimensional human body measurement method based on 3D structured light and a double-layer surface model, wherein the double-layer surface model includes an inner layer model and an outer layer model, wherein the inner layer model is used to capture the basic form and posture of the human body, and the outer layer model is used to capture the detailed features of non-rigid objects on the surface of the human body; the method comprises:

[0006] The 3D structured light camera module is used to emit light with structural features to the surface of the human body to be measured, and the light reflected by the surface of the human body to be measured is captured, and the depth map data of the human body to be measured is generated according to the light reflected by the surface of the human body to be measured;

[0007] Establishing a three-dimensional point cloud data model of the human body to be measured according to the depth map data;

[0008] Matching the three-dimensional point cloud data model with the double-layer surface model, adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process, and obtaining an adjusted double-layer surface model after the matching process is completed; the double-layer node graph includes: in-body nodes and far-body nodes, wherein the in-body nodes correspond to the inner-layer model, and the far-body nodes correspond to the outer-layer model;

[0009] The three-dimensional measurement data of the human body to be measured is obtained according to the adjusted double-layer surface model.

[0010] In an optional implementation of the first aspect, adjusting the double-layer surface model based on a difference between the double-layer surface model and the three-dimensional point cloud data model and a double-layer node graph of the double-layer surface model during the matching process includes:

[0011] Matching the inner layer model with the three-dimensional point cloud data model to obtain an error between the inner layer model after parameter adjustment and the three-dimensional point cloud data model, and adjusting the parameters of the inner layer model based on the error;

[0012] Based on the matched inner layer model, the point cloud data in the three-dimensional point cloud data model is integrated to match the outer layer model, and the parameters of the outer layer model are adjusted during the matching process;

[0013] Synchronously binding and data fusing the matched inner layer model and the matched outer layer model, dynamically adjusting the double-layer node graph based on the motion characteristics of the three-dimensional point cloud data model, so as to perform joint motion tracking on the double-layer surface model, and optimizing the parameters of the double-layer surface model based on the effect of the joint motion tracking;

[0014] It is determined whether the current double-layer surface model meets the model optimization cutoff condition. If so, the adjustment of the model parameters is stopped to obtain the adjusted double-layer surface model.

[0015] In an optional implementation of the first aspect, matching the inner layer model with the three-dimensional point cloud data model, obtaining an error between the inner layer model after parameter adjustment and the three-dimensional point cloud data model, and adjusting the parameters of the inner layer model based on the error, includes:

[0016] The parameters of the inner layer model are adjusted with the minimization of the value of the energy function as the optimization goal; the energy function includes: a data term, a regularization term and a priori term; the data term is used to measure the difference between the inner layer model and the three-dimensional point cloud data model; the regularization term is used to constrain the deformation degree of the inner layer model; the priori term is used to guide the convergence of the inner layer model by using the prior knowledge of human body shape and posture.

[0017] In an optional implementation of the first aspect, dynamically adjusting the double-layer node graph based on the motion features of the three-dimensional point cloud data model to perform joint motion tracking on the double-layer surface model includes:

[0018] Controlling the rotation of the rotating device and driving the human body to be measured located on the rotating device to rotate synchronously; the depth map data specifically includes: a continuous data stream of the depth map generated according to the light reflected by the surface of the human body to be measured during the period of synchronous rotation with the rotating device;

[0019] The double-layer surface model is controlled to rotate synchronously with the rotating device so that the double-layer surface model is matched with the corresponding target point cloud data in the three-dimensional point cloud data model at a target angle, and the parameters of the inner layer model and the outer layer model are optimized; the target point cloud data is point cloud data converted based on the target depth map in the depth map continuous data stream; the target depth map is a depth map generated based on the light reflected from the surface when the human body to be measured is rotated to the target angle.

[0020] In an optional implementation of the first aspect, the determining whether the current double-layer surface model satisfies a model optimization cutoff condition includes:

[0021] Determine whether the matching quality of the current double-layer surface model is qualified, and determine whether the joint motion tracking effect of the current double-layer surface model is qualified;

[0022] If the matching quality is qualified and the effect of the joint motion tracking is qualified, it is determined that the current double-layer surface model meets the model optimization cutoff condition;

[0023] If the matching quality is unsatisfactory, or the effect of the joint motion tracking is unsatisfactory, it is determined that the current double-layer surface model does not meet the model optimization cutoff condition.

[0024] In an optional implementation of the first aspect, the determining whether the matching quality of the current double-layer surface model is qualified includes:

[0025] If the matching error of the current double-layer surface model is less than or equal to the preset error threshold, visual inspection shows that the surface of the current double-layer surface model is smooth and consistent with the point cloud, then the matching quality of the current double-layer surface model is judged to be qualified;

[0026] If the matching error of the current double-layer surface model is greater than the preset error threshold, and a visual inspection shows that a key area of ​​the current double-layer surface model is deformed or broken, then it is determined that the matching quality of the current double-layer surface model is unqualified.

[0027] In an optional implementation of the first aspect, the determining whether the effect of the joint motion tracking of the current double-layer surface model is qualified includes:

[0028] If the current double-layer surface model is stable in updating and always consistent with the point cloud when it moves at a preset speed or performs a gesture of preset complexity, the model matches the point cloud consistently across multiple frames, and the motion tracking accuracy is greater than a preset accuracy threshold, then the joint motion tracking effect of the current double-layer surface model is judged to be qualified;

[0029] If the current double-layer surface model slides or misaligns when moving at a preset speed or performing a posture of preset complexity, causing the model to not match the actual posture, and if there is a freeze or distortion during the motion tracking process, then the joint motion tracking effect of the current double-layer surface model is judged to be unsatisfactory.

[0030] In an optional implementation of the first aspect, establishing a three-dimensional point cloud data model of the human body to be measured according to the depth map data includes:

[0031] Performing denoising and enhancement processing on the depth map data;

[0032] The denoised and enhanced depth map data is converted into a 3D point cloud data model of the human body to be measured using a 3D reconstruction algorithm.

[0033] In an optional implementation of the first aspect, while the 3D structured light camera module collects light reflected from the surface of the human body to be measured, the method further includes:

[0034] Collecting the position information and / or posture information of the human body to be measured by an inertial measurement unit;

[0035] The method of using a three-dimensional reconstruction algorithm to convert the denoised and enhanced depth map data into a three-dimensional point cloud data model of the human body to be measured includes:

[0036] Using a 3D reconstruction algorithm, the denoised and enhanced depth map data is converted into preliminary point cloud data;

[0037] The position information and / or the posture information are used to provide constraints for point cloud registration, and preliminary point cloud data from multiple perspectives are registered and fused to form a three-dimensional point cloud data model of the human body to be measured.

[0038] In an optional implementation of the first aspect, while the 3D structured light camera module collects light reflected from the surface of the human body to be measured, the method further includes:

[0039] Collecting the position information and / or posture information of the human body to be measured by an inertial measurement unit;

[0040] When matching the three-dimensional point cloud data model with the double-layer surface model, the dynamic parameters of the double-layer surface model are optimized using the position information and / or the posture information.

[0041] In a second aspect, the present application provides a three-dimensional human body measurement device based on 3D structured light and a double-layer surface model, wherein the double-layer surface model includes an inner layer model and an outer layer model, wherein the inner layer model is used to capture the basic form and posture of the human body, and the outer layer model is used to capture the detailed features of non-rigid objects on the surface of the human body; the device includes:

[0042] A depth map generation module is used to emit light with structural features to the surface of the human body to be tested through a 3D structured light camera module, and capture the light reflected by the surface of the human body to be tested, and generate depth map data of the human body to be tested according to the light reflected by the surface of the human body to be tested;

[0043] A point cloud conversion module, used to establish a three-dimensional point cloud data model of the human body to be measured according to the depth map data;

[0044] A model matching and adjustment module, used for matching the three-dimensional point cloud data model with the double-layer surface model, adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process, and obtaining the adjusted double-layer surface model after the matching process is completed; the double-layer node graph includes: in-body nodes and far-body nodes, wherein the in-body nodes correspond to the inner layer model, and the far-body nodes correspond to the outer layer model;

[0045] The measurement data acquisition module is used to obtain the three-dimensional measurement data of the human body to be measured according to the adjusted double-layer surface model.

[0046] Compared with the prior art, this application has the following beneficial effects:

[0047] The present application discloses a three-dimensional human body measurement method and device based on 3D structured light and double-layer surface model. A 3D structured light camera module emits light with structural features to the surface of the human body to be measured, and captures the reflected light from the human body surface to generate the depth map data of the human body to be measured, which is further converted into a three-dimensional point cloud data model. In the process of matching the three-dimensional point cloud data model with the double-layer surface model, the double-layer surface model is adjusted based on the data difference and the double-layer node graph of the double-layer surface model. Since the inner model in the double-layer surface model can capture the basic form and posture of the human body, and the outer model can capture the detailed features of non-rigid objects on the surface of the human body, the double-layer surface model is fitted and matched with the three-dimensional point cloud data model, and the model parameters can be adjusted from both the inside and the outside to approximate the human body features reflected by the three-dimensional point cloud data model. Finally, the three-dimensional measurement data of the human body to be measured is obtained according to the adjusted double-layer surface model. The present application uses 3D structured light and a double-layer surface model to accurately reconstruct the three-dimensional form and additional objects of the human body, which not only realizes three-dimensional precise measurement, but also significantly reduces the measurement cost compared with 3D scanning equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0049] Figure 1 A flowchart of a three-dimensional human body measurement method based on 3D structured light and a double-layer surface model provided in an embodiment of the present application;

[0050] Figure 2 A flowchart of an implementation method for matching and adjusting a double-layer surface model based on a three-dimensional point cloud data model provided in an embodiment of the present application;

[0051] Figure 3 A schematic structural diagram of a three-dimensional human body measurement device based on 3D structured light and a double-layer surface model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] As described above, the accuracy and reliability of traditional human body measurement methods are poor. Although the use of 3D scanning equipment has high accuracy, the measurement cost is high and it is not easy to promote. In view of the above problems, the inventors have conducted research and provided a three-dimensional human body measurement method and device based on 3D structured light and double-layer surface model in this application. With the help of a 3D structured light camera module, the light reflected from the human body surface is captured, and depth map data with depth information is formed. Based on this, it is converted into a three-dimensional point cloud data model for fitting and matching with a double-layer surface model. In the technical solution of the present application, a double-layer surface model is used to fit the parameters of human body features, wherein the inner model captures the basic form and posture of the human body, and the outer model captures the detailed features of non-rigid objects on the surface of the human body. As a result, the fitted double-layer surface model can reflect both the essential characteristics of the human body to be measured and the detailed characteristics, thereby ensuring the accuracy and reliability of the three-dimensional measurement data finally obtained.

[0053] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0054] See also Figure 1 , which is a flow chart of a three-dimensional human body measurement method based on 3D structured light and a double-layer surface model provided in an embodiment of the present application. Figure 1 As shown, the 3D human body measurement method based on 3D structured light and double-layer surface model includes:

[0055] S101, emitting light with structural features to the surface of a human body to be measured through a 3D structured light camera module, capturing the light reflected from the surface of the human body to be measured, and generating depth map data of the human body to be measured according to the light reflected from the surface of the human body to be measured.

[0056] The light with structural features emitted by the 3D structured light camera module can be, for example, a grid-shaped infrared light or a stripe-shaped infrared light. The structural features can be realized through mode settings. For example, different light emission modes are pre-configured, and when the user selects a specific light emission mode, infrared light with specific structural features of the mode will be emitted to the surface of the human body to be tested.

[0057] In an embodiment of the present application, a 3D structured light camera module includes a transmitter and a receiver. The receiver is configured with a lens assembly and an image sensor. The transmitter emits infrared light to the human body to be measured, and the image sensor is used to collect the reflected light transmitted through the lens assembly and perform photoelectric conversion on it. The converted electrical signal is used to generate an infrared image. A processing chip can also be configured in the 3D structured light camera module, which can obtain depth map data based on the infrared image conversion. In addition, multiple 3D structured light camera modules can also be configured in the scene. Combining multiple modules can cover a larger field of view, so that it is suitable for more three-dimensional human body measurement needs of different body shapes and postures.

[0058] S102: Establish a three-dimensional point cloud data model of the human body to be measured according to the depth map data.

[0059] In practical applications, computer vision technology can be used to generate a 3D point cloud data model based on the collected depth map data. Since the conversion of depth map data into point cloud data based on computer vision technology is a relatively mature technology, it will not be described in detail here.

[0060] S103, matching the three-dimensional point cloud data model with the double-layer surface model, adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process, and obtaining the adjusted double-layer surface model after the matching process is completed.

[0061] Since the three-dimensional point cloud data model is based on the collected reflected light and is obtained by combining the depth information transformation, it can truly reflect the characteristics of the human body to be tested from a spatial perspective. After the construction process of the three-dimensional point cloud data model of the human body to be tested has been completed through step S102, next, in this step S103, the three-dimensional point cloud data model is used as a template, and the parameters of an initialized double-layer surface model are fitted with reference to the three-dimensional point cloud data model, so that the double-layer surface model can also accurately reflect the characteristics of the human body to be tested through the optimization of parameters in the fitting and matching process.

[0062] The double-layer surface model includes an inner layer model and an outer layer model. The inner layer model is used to capture the basic form and posture of the human body, and the outer layer model is used to capture the detailed features of non-rigid objects on the surface of the human body. The double-layer node graph of the double-layer surface model expresses the structural connection between the nodes on the inner layer model and the nodes on the outer layer model. Specifically, the double-layer node graph includes: in-body nodes and far-body nodes, where the in-body nodes correspond to the inner layer model and the far-body nodes correspond to the outer layer model.

[0063] In-body nodes are responsible for representing the human skeleton morphology. Remote body nodes are responsible for describing the external details of the human body, such as the shape of clothing. In addition to its basic components (in-body nodes and remote body nodes), the two-layer node graph also includes additional information. The additional information includes the connection relationship between nodes, such as the connection relationship between in-body nodes and in-body nodes, the connection relationship between remote body nodes and remote body nodes, and the connection relationship between in-body nodes and remote body nodes. In addition, the additional information can also include constraint rules, which are used to ensure the consistency of the inner model and the outer model.

[0064] Regarding the specific implementation of matching the three-dimensional point cloud data model with the double-layer surface model mentioned in step S103, the inner model can be matched first, and then the outer model can be matched after the inner model is matched. This is because the inner model is used to capture the basic form of the human body. Matching the inner model first can provide a stable and relatively accurate foundation for the outer model. The outer model focuses on the details of the human body surface. Based on the matching results of the inner model, the outer model is further matched, and the captured details of the human body surface are not easily distorted. Figure 2 , which introduces this example implementation.

[0065] Figure 2 A flowchart of a method for implementing a three-dimensional point cloud data model matching and adjusting a double-layer surface model is provided in an embodiment of the present application. Figure 2 As shown, adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process may include:

[0066] S1031. Match the inner layer model with the three-dimensional point cloud data model to obtain an error between the inner layer model after parameter adjustment and the three-dimensional point cloud data model, and adjust the parameters of the inner layer model based on the error.

[0067] In a possible implementation of step S1031, the data term of the energy function is used to quantify the error between the inner layer model and the three-dimensional point cloud data model. In addition, the regularization term of the energy function is combined to constrain the deformation of the model, and the prior term of the energy function is used to guide the model convergence. Specifically, the value of minimizing the energy function can be used as the optimization goal to adjust the parameters of the inner layer model.

[0068] As mentioned above, the energy function includes: data term, regularization term and prior term. The data term is used to measure the difference between the inner layer model and the three-dimensional point cloud data model. The regularization term is used to constrain the deformation degree of the inner layer model. The prior term is used to guide the convergence of the inner layer model by using the prior knowledge of human body shape and posture.

[0069] Taking the minimization of the value of the energy function containing the above data items, regularization items and prior items as the optimization target of the inner model helps to quickly achieve the fitting convergence of the model, narrow the gap between the inner model after parameter adjustment and the three-dimensional point cloud data model, obtain a more ideal matching effect, and truly reflect the basic form of the human body to be tested through the inner model. Then, the outer model can be matched through S1032.

[0070] S1032: Based on the matched inner layer model, the point cloud data in the three-dimensional point cloud data model is integrated to match the outer layer model, and the parameters of the outer layer model are adjusted during the matching process.

[0071] In this application, the outer layer model is matched based on the matched inner layer model. The point cloud data representing the surface details of the human body in the 3D point cloud data model is integrated to match the outer layer model, so that the outer layer model can accurately capture the surface details of the human body when optimizing and adjusting the parameters.

[0072] S1033. Synchronously bind and fuse the matched inner layer model and the matched outer layer model, and dynamically adjust the double-layer node graph based on the motion characteristics of the three-dimensional point cloud data model to perform joint motion tracking on the double-layer surface model, and optimize the parameters of the double-layer surface model based on the effect of the joint motion tracking.

[0073] In the embodiment of the present application, it is necessary to synchronize and bind the matched inner model and outer model. Synchronous binding is a pre-step for joint motion tracking optimization. Through synchronous binding, it can ensure that the inner model and the outer model are aligned with each other and have a binding relationship. In this way, it provides a guarantee for parameter optimization in the joint motion tracking stage.

[0074] In the embodiment of the present application, the joint motion tracking of the double-layer surface model is mainly to evaluate whether the double-layer surface model after parameter adjustment and optimization has the ability to accurately reflect the motion characteristics of the human body to be tested. For example, to evaluate whether the double-layer surface model maintains good performance in the face of rapid human body movement or complex posture changes. Below, an example implementation of the joint motion tracking of the double-layer surface model by dynamically adjusting the double-layer node graph based on the motion characteristics of the three-dimensional point cloud data model is introduced.

[0075] In an embodiment of the present application, the human body to be measured is on a rotating device during the measurement phase. The rotating device can rotate under the drive of a motor, and then the human body to be measured also rotates. For example, the rotating device rotates 180°, and the human body to be measured also rotates 180°. During the rotation of the rotating device, for measurement needs, the human body to be measured can be required to remain relatively still with the rotating device according to instructions, or the human body to be measured can be required to make some specific postures (such as opening arms, waving arms, kicking legs, etc.) during the rotation of the rotating device.

[0076] In an embodiment of the present application, in order to achieve joint motion tracking of the double-layer surface model, the rotating device is controlled to rotate and drive the human body to be measured located on the rotating device to rotate synchronously; the double-layer surface model is controlled to rotate synchronously with the rotating device, so that the double-layer surface model is matched with the corresponding target point cloud data in the three-dimensional point cloud data model at a target angle, and the parameters of the inner layer model and the parameters of the outer layer model are optimized.

[0077] When the human body to be tested on the rotating device is driven to rotate synchronously, the 3D structured light camera module continuously transmits infrared light to the human body to be tested through the transmitter, and collects the infrared light reflected by the surface of the rotating human body to be tested through the receiver, thereby generating a continuous infrared image. Subsequently, a continuous data stream of a depth map can be generated based on the continuous infrared image. For example, the continuous data stream of the depth map contains the depth map data generated corresponding to each frame of the infrared image collected during the rotation of the human body to be tested. Since the depth map data is presented in the form of a continuous data stream, the constructed three-dimensional point cloud data model also reflects the dynamic characteristics of the human body to be tested during the rotation process. When performing joint motion tracking, it is necessary to make the double-layer surface model rotate synchronously with the rotation of the human body to be tested, so that the established three-dimensional point cloud data model can be matched with the double-layer surface model angle by angle (frame by frame) to achieve joint motion tracking. For example, the target depth map is a depth map generated based on the light reflected by the surface when the human body to be tested rotates to the target angle. The target point cloud data is point cloud data converted based on the target depth map in the continuous data stream of the depth map. In other words, the target depth map and the target point cloud data both correspond to the target angle. The double-layer surface model is rotated synchronously, so that the double-layer surface model can be matched with the target point cloud data when it is at a target rotation angle.

[0078] The purpose of performing joint motion tracking on the double-layer surface model is also to optimize the parameters of the double-layer surface model based on the effect of joint motion tracking. Thus, the optimized double-layer surface model has high-accuracy dynamic performance. When the model parameter optimization is terminated, it is also necessary to judge with the help of the model optimization cutoff condition. See step S1034 for details.

[0079] S1034, judging whether the current double-layer surface model meets the model optimization cutoff condition, if so, proceeding to step S1035.

[0080] Exemplarily, the model optimization cutoff condition of the double-layer surface model may include two specific conditions, one condition is about the matching quality of the model, and the other condition is related to the dynamic performance of the model. Specifically, it is determined whether the matching quality of the current double-layer surface model is qualified, and it is determined whether the effect of the joint motion tracking of the current double-layer surface model is qualified. If the matching quality is qualified, and the effect of the joint motion tracking is qualified, it is determined that the current double-layer surface model meets the model optimization cutoff condition. If the matching quality is unqualified, or the effect of the joint motion tracking is unqualified, it is determined that the current double-layer surface model does not meet the model optimization cutoff condition.

[0081] In other possible implementations, the model optimization cutoff condition may also be a unilateral condition, for example, the matching quality of the model is qualified, that is, the model optimization cutoff condition is satisfied. Alternatively, the joint motion tracking effect of the model is qualified, that is, the model optimization cutoff condition is satisfied. In addition, the model optimization cutoff condition may also involve the number of times the model parameters are optimized. The model parameter optimization is stopped by limiting the number of times.

[0082] Below, by way of example, an implementation method for determining whether the matching quality of the current double-layer surface model is qualified is introduced.

[0083] If the matching error of the current double-layer surface model is less than or equal to the preset error threshold, the surface of the current double-layer surface model is visually checked to be smooth and consistent with the point cloud, and the matching quality of the current double-layer surface model is determined to be qualified. For example, if the average error from the point cloud to the surface of the double-layer surface model is lower than the preset error threshold, the matching quality is determined to be qualified.

[0084] If the matching error of the current double-layer surface model is greater than the preset error threshold, a visual inspection shows that a key area of ​​the current double-layer surface model is deformed or broken, and the matching quality of the current double-layer surface model is judged to be unqualified. For example, if the average error from the point cloud to the surface of the double-layer surface model is greater than the preset error threshold, there is a large deviation on the surface, and the matching quality of the model is unqualified.

[0085] Below, by way of example, an implementation method for determining whether the effect of the joint motion tracking of the current double-layer surface model is qualified is introduced.

[0086] If the current double-layer surface model is moving at a preset speed or performing a posture of preset complexity, the model update is stable and always consistent with the point cloud, the model matches the point cloud coherently between multiple frames, and the motion tracking accuracy is greater than the preset accuracy threshold, then the joint motion tracking effect of the current double-layer surface model is judged to be qualified. Among them, the preset speed is used to distinguish the speed of movement, which can be customized; the preset complexity is used to distinguish the high and low complexity of the posture, which can also be customized. As an example, in fast motion or complex postures (such as dance performances or sports activities), the model update is stable and always consistent with the point cloud, the model and the point cloud match coherently for multiple frames, and the dynamic tracking accuracy is high, then the joint motion tracking effect of the model is determined to be qualified.

[0087] If the current double-layer surface model slides or is misaligned when it moves at a preset speed or performs a gesture of preset complexity, causing the model to not match the actual gesture, and causing freezes or distortions during motion tracking, then the joint motion tracking effect of the current double-layer surface model is judged to be unqualified. For example, if the model slides or is misaligned, causing the model to not match the actual gesture, and causing freezes or distortions during motion tracking, it indicates that the model cannot meet the dynamic real-time and accuracy requirements, and therefore the model parameters need to be adjusted and optimized. The joint tracking effect of the current double-layer surface model is unstable or does not achieve the expected accuracy.

[0088] S1035. Stop adjusting the model parameters to obtain an adjusted double-layer surface model.

[0089] Through the above steps S1031 to S1035, the optimization adjustment of the model parameters is completed, and the adjusted double-layer surface model that meets expectations is obtained. Next, the process proceeds to S104.

[0090] S104: Obtain three-dimensional measurement data of the human body to be measured according to the adjusted double-layer surface model.

[0091] As a parametric human body model, the double-layer surface model represents the shape and posture of the human body through parameters. Therefore, after obtaining the adjusted double-layer surface model, based on the accurate human body information contained in the model, specific measurement data representing the three-dimensional characteristics of the human body can be calculated. For example, specific dimensions such as height, weight, and body proportions. This can be used for specific applications in subsequent clothing design, motion analysis, and other fields.

[0092] The three-dimensional human body measurement method based on 3D structured light and double-layer surface model introduced above emits light with structural features to the surface of the human body to be measured through a 3D structured light camera module, and captures the reflected light on the human body surface to generate the depth map data of the human body to be measured, which is further converted into a three-dimensional point cloud data model. In the process of matching the three-dimensional point cloud data model with the double-layer surface model, the double-layer surface model is adjusted based on the data difference and the double-layer node graph of the double-layer surface model. Since the inner model in the double-layer surface model can capture the basic form and posture of the human body, and the outer model can capture the detailed features of non-rigid objects on the surface of the human body, the double-layer surface model is fitted and matched with the three-dimensional point cloud data model, and the model parameters can be adjusted from both the inside and the outside to approximate the human body features reflected by the three-dimensional point cloud data model. Finally, the three-dimensional measurement data of the human body to be measured is obtained according to the adjusted double-layer surface model. This application uses 3D structured light and double-layer surface models to accurately reconstruct the three-dimensional form and additional objects of the human body, which not only realizes three-dimensional precise measurement, but also significantly reduces the measurement cost compared to 3D scanning equipment.

[0093] Furthermore, the embodiment of the present application also proposes a specific implementation process for establishing a three-dimensional point cloud data model. First, the depth map data is denoised and enhanced. In this way, the data quality of the depth map data is improved. Next, a three-dimensional reconstruction algorithm, such as an iterative closest point algorithm, can be used to convert the processed data into a three-dimensional point cloud data model of the human body to be measured. Preliminary optimization measures such as denoising and enhancing the depth map data can be beneficial to achieve accurate conversion of the three-dimensional point cloud data model. This process ensures the accuracy and integrity of the data during the conversion from a two-dimensional image to a three-dimensional model.

[0094] The embodiment of the present application also proposes that multi-modal information can be comprehensively used to achieve accurate measurement of three-dimensional human body data. Two possible implementation methods of the comprehensive use of multi-modal information are introduced below.

[0095] In a first possible implementation manner, while the 3D structured light camera module collects light reflected from the surface of the human body to be measured, the method further includes:

[0096] Collecting the position information and / or posture information of the human body to be measured by an inertial measurement unit;

[0097] Furthermore, the 3D reconstruction algorithm is used to convert the denoised and enhanced depth map data into a 3D point cloud data model of the human body to be measured, including:

[0098] The denoised and enhanced depth map data is converted into preliminary point cloud data using a three-dimensional reconstruction algorithm. The position information and / or the posture information is used to provide constraints for point cloud registration, and the preliminary point cloud data from multiple perspectives are registered and fused to form a three-dimensional point cloud data model of the human body to be measured.

[0099] In the first possible implementation, the position information and / or attitude information measured by the inertial measurement unit is used as an example of fused modal information. Combining this modal information provides auxiliary constraints for point cloud registration, which is conducive to more accurate model matching. The first possible implementation reflects the fusion of multi-modal information in the point cloud processing stage.

[0100] In a second possible implementation manner, while the 3D structured light camera module collects light reflected from the surface of the human body to be measured, the method further includes:

[0101] Collecting the position information and / or posture information of the human body to be measured by an inertial measurement unit;

[0102] When matching the three-dimensional point cloud data model with the double-layer surface model, the dynamic parameters of the double-layer surface model are optimized using the position information and / or the posture information.

[0103] In the second possible implementation, the position information and / or attitude information measured by the inertial measurement unit is used as an example of fused modal information. Combined with this modal information, it provides guidance for dynamic parameter adjustment in model matching. The second possible implementation reflects the fusion of multi-modal information in the model matching stage.

[0104] The additional position and / or attitude information provided by sensors such as inertial measurement units enhances the robustness of the system through multimodal fusion and allows high-precision body measurement even in complex lighting conditions or occlusions.

[0105] It should be noted that the above only uses the measurement of position or posture by the inertial measurement unit as an example of the application of multimodal information fusion. The inertial measurement unit can be installed at the collar, corner of the clothes, etc. of the human body to be measured, or at other positions of the human body to be measured, which is not limited here. In addition, the sensor used to collect multimodal information is not limited to the inertial measurement unit, but can also be a sensor of other types or functions.

[0106] Parameterized Common Model refers to a parameter-based human body model in the field of human body modeling, which can describe and simulate the human body shape and posture of different individuals through a set of parameters. This model is usually built based on a large amount of human body scanning data, which can capture the main variation characteristics of the human body and express these characteristics in a parameterized form. The main advantage of the parameterized common model is its flexibility and universality, which can adapt to people of different genders, ages and body shapes. In the embodiment of the present application, the double-layer surface model adopts a parameterized model with a double-layer structure (inner layer and outer layer). In the embodiment of the present application, the double-layer surface model undertakes the task of accurately capturing the basic form of the human body. The model is adjusted to match the human body data of a specific individual through a series of predefined parameters, such as body shape parameters, posture parameters, etc. These parameters can describe the main geometric characteristics and motion characteristics of the human body, so that the model can not only adapt to the static human body shape, but also simulate dynamic posture changes.

[0107] The inner model, such as the SMPL model, is the core of the double-layer surface model. It accurately captures the basic form and dynamic posture changes of the human body through predefined parameters (such as body shape parameters and posture parameters). The outer model is responsible for capturing the dynamic non-rigid objects of the human body, such as clothing, and realizes the overall reconstruction of the human body and its attached objects through gradually fused surface data. This design enables the model to not only adapt to the static human body shape, but also simulate dynamic posture changes and external expressions. The design of the double-layer surface model reflects the advantages of the parametric public model. Through the collaborative work of the inner and outer layers, the comprehensive capture and accurate reconstruction of the human body and its attached objects are achieved.

[0108] This technology makes the human body measurement and 3D reconstruction process faster and more accurate, realizes automated operation, and significantly improves the credibility of data and the efficiency of measurement work. It has a wide range of applications, including medical health, sports science, clothing design, virtual reality and other fields. In particular, its value is particularly prominent in scenes that require high precision and real-time dynamic reconstruction.

[0109] In general, this application combines modern optical scanning technology and computer modeling technology to propose an innovative and efficient three-dimensional human body reconstruction solution. Its innovation lies in the ability to capture the three-dimensional form of the human body in real time, and to use a double-layer surface model to jointly model the human body and attached objects. This process not only improves the accuracy of the measurement, but also greatly reduces the required measurement time. In addition, this technology is extremely user-friendly, making it easy for even non-professionals to operate, further broadening its scope of application. Specifically, an intuitive and easy-to-use software interface has been developed that allows operators to easily set scanning parameters, start and stop the scanning process, and view measurement results in real time. In addition, the association with business needs simplifies the operating process and lowers the technical threshold, so that non-professionals can also use it effectively, thereby expanding the scope of application and market acceptance of the technology.

[0110] The technical solution adopted in this application is not limited to static human body measurement, but can also be applied to the capture of dynamic scenes, such as human body interaction in virtual reality (VR) and augmented reality (AR), and dynamic character creation in the film and game industries. Through these innovative applications, not only the accuracy and efficiency of human body measurement are improved, but also its application potential in multiple fields is expanded. The development of this technology indicates that there will be more breakthroughs and innovations in human body dynamic capture and virtual character creation in the future.

[0111] The entire system adopts a modular design, which is easy to maintain and upgrade. Each component such as the transmitter, receiver and processing software is designed as an independent module, which is easy to configure and adjust according to specific needs. The modular design not only improves the flexibility and adaptability of the system, but also reduces costs, making this advanced technology easier to promote and apply. Through this comprehensive technical solution, the new system not only solves the accuracy and efficiency problems in existing measurement technologies, but also meets the needs of modern high-speed development industries with its user-friendly design and powerful data processing capabilities. The technical solution can process the captured image data in real time and quickly generate accurate measurement results to meet the fast-paced needs in industrial and commercial applications. Through adjustable beam projection and signal-to-noise ratio adjustment, this technical solution can adapt to different measurement environments and complex human forms, ensuring high-quality data output under various conditions.

[0112] Based on the three-dimensional human body measurement method based on 3D structured light and double-layer surface model introduced in the above embodiments, correspondingly, the present application also provides a three-dimensional human body measurement device based on 3D structured light and double-layer surface model. Figure 3 Provide explanation.

[0113] Figure 3A schematic diagram of the structure of a three-dimensional human body measurement device based on 3D structured light and a double-layer surface model provided in an embodiment of the present application. The double-layer surface model includes an inner layer model and an outer layer model, wherein the inner layer model is used to capture the basic shape and posture of the human body, and the outer layer model is used to capture the detailed features of non-rigid objects on the surface of the human body. Figure 3 As shown, the device comprises:

[0114] A depth map generation module is used to emit light with structural features to the surface of the human body to be tested through a 3D structured light camera module, and capture the light reflected by the surface of the human body to be tested, and generate depth map data of the human body to be tested according to the light reflected by the surface of the human body to be tested;

[0115] A point cloud conversion module, used to establish a three-dimensional point cloud data model of the human body to be measured according to the depth map data;

[0116] A model matching and adjustment module, used for matching the three-dimensional point cloud data model with the double-layer surface model, adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process, and obtaining the adjusted double-layer surface model after the matching process is completed; the double-layer node graph includes: in-body nodes and far-body nodes, wherein the in-body nodes correspond to the inner layer model, and the far-body nodes correspond to the outer layer model;

[0117] The measurement data acquisition module is used to obtain the three-dimensional measurement data of the human body to be measured according to the adjusted double-layer surface model.

[0118] In an optional implementation, the model matching and adjustment module is specifically used to:

[0119] Matching the inner layer model with the three-dimensional point cloud data model to obtain an error between the inner layer model after parameter adjustment and the three-dimensional point cloud data model, and adjusting the parameters of the inner layer model based on the error;

[0120] Based on the matched inner layer model, the point cloud data in the three-dimensional point cloud data model is integrated to match the outer layer model, and the parameters of the outer layer model are adjusted during the matching process;

[0121] Synchronously binding and data fusing the matched inner layer model and the matched outer layer model, dynamically adjusting the double-layer node graph based on the motion characteristics of the three-dimensional point cloud data model, so as to perform joint motion tracking on the double-layer surface model, and optimizing the parameters of the double-layer surface model based on the effect of the joint motion tracking;

[0122] It is determined whether the current double-layer surface model meets the model optimization cutoff condition. If so, the adjustment of the model parameters is stopped to obtain the adjusted double-layer surface model.

[0123] In an optional implementation, the model matching and adjustment module is specifically used to:

[0124] The parameters of the inner layer model are adjusted with the minimization of the value of the energy function as the optimization goal; the energy function includes: a data term, a regularization term and a priori term; the data term is used to measure the difference between the inner layer model and the three-dimensional point cloud data model; the regularization term is used to constrain the deformation degree of the inner layer model; the priori term is used to guide the convergence of the inner layer model by using the prior knowledge of human body shape and posture.

[0125] In an optional implementation, the model matching and adjustment module is specifically used to:

[0126] Controlling the rotation of the rotating device and driving the human body to be measured located on the rotating device to rotate synchronously; the depth map data specifically includes: a continuous data stream of the depth map generated according to the light reflected by the surface of the human body to be measured during the period of synchronous rotation with the rotating device;

[0127] The double-layer surface model is controlled to rotate synchronously with the rotating device so that the double-layer surface model is matched with the corresponding target point cloud data in the three-dimensional point cloud data model at a target angle, and the parameters of the inner layer model and the outer layer model are optimized; the target point cloud data is point cloud data converted based on the target depth map in the depth map continuous data stream; the target depth map is a depth map generated based on the light reflected from the surface when the human body to be measured is rotated to the target angle.

[0128] In an optional implementation, the model matching and adjustment module is specifically used to:

[0129] Determine whether the matching quality of the current double-layer surface model is qualified, and determine whether the joint motion tracking effect of the current double-layer surface model is qualified;

[0130] If the matching quality is qualified and the effect of the joint motion tracking is qualified, it is determined that the current double-layer surface model meets the model optimization cutoff condition;

[0131] If the matching quality is unsatisfactory, or the effect of the joint motion tracking is unsatisfactory, it is determined that the current double-layer surface model does not meet the model optimization cutoff condition.

[0132] In an optional implementation, the model matching and adjustment module is specifically used to determine whether the matching quality of the current double-layer surface model is qualified by:

[0133] If the matching error of the current double-layer surface model is less than or equal to the preset error threshold, visual inspection shows that the surface of the current double-layer surface model is smooth and consistent with the point cloud, then the matching quality of the current double-layer surface model is judged to be qualified;

[0134] If the matching error of the current double-layer surface model is greater than the preset error threshold, and a visual inspection shows that a key area of ​​the current double-layer surface model is deformed or broken, then it is determined that the matching quality of the current double-layer surface model is unqualified.

[0135] In an optional implementation, the model matching and adjustment module is specifically used to determine whether the joint motion tracking effect of the current double-layer surface model is qualified by:

[0136] If the current double-layer surface model is stable in updating and always consistent with the point cloud when it moves at a preset speed or performs a gesture of preset complexity, the model matches the point cloud consistently across multiple frames, and the motion tracking accuracy is greater than a preset accuracy threshold, then the joint motion tracking effect of the current double-layer surface model is judged to be qualified;

[0137] If the current double-layer surface model slides or misaligns when moving at a preset speed or performing a posture of preset complexity, causing the model to not match the actual posture, and if there is a freeze or distortion during the motion tracking process, then the joint motion tracking effect of the current double-layer surface model is judged to be unsatisfactory.

[0138] In an optional implementation, the point cloud conversion module is specifically used to:

[0139] Performing denoising and enhancement processing on the depth map data;

[0140] The denoised and enhanced depth map data is converted into a 3D point cloud data model of the human body to be measured using a 3D reconstruction algorithm.

[0141] In an optional implementation, the device further includes: a multimodal information collection module; the multimodal information collection module is used to collect the position information and / or posture information of the human body to be measured through an inertial measurement unit while the 3D structured light camera module collects the light reflected from the surface of the human body to be measured;

[0142] The point cloud conversion module is specifically used for:

[0143] Using a 3D reconstruction algorithm, the denoised and enhanced depth map data is converted into preliminary point cloud data;

[0144] The position information and / or the posture information are used to provide constraints for point cloud registration, and preliminary point cloud data from multiple perspectives are registered and fused to form a three-dimensional point cloud data model of the human body to be measured.

[0145] In an optional implementation, the device also includes: a multimodal information collection module; the multimodal information collection module is used to collect the position information and / or posture information of the human body to be measured through an inertial measurement unit while the 3D structured light camera module collects the light reflected from the surface of the human body to be measured; when matching the three-dimensional point cloud data model with the double-layer surface model, the position information and / or the posture information is used to optimize the dynamic parameters of the double-layer surface model.

[0146] It should be noted that each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is merely schematic, in which the unit described as a separate component may or may not be physically separated, and the component prompted as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0147] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A three-dimensional human body measurement method based on 3D structured light and double-layer surface model, characterized in that: The double-layer surface model includes an inner layer model and an outer layer model, wherein the inner layer model is used to capture the basic form and posture of the human body, and the outer layer model is used to capture the detailed features of non-rigid objects on the surface of the human body; The method comprises: The 3D structured light camera module is used to emit light with structural features to the surface of the human body to be measured, and the light reflected by the surface of the human body to be measured is captured, and the depth map data of the human body to be measured is generated according to the light reflected by the surface of the human body to be measured; Establishing a three-dimensional point cloud data model of the human body to be measured according to the depth map data; Matching the three-dimensional point cloud data model with the double-layer surface model, adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process, and obtaining an adjusted double-layer surface model after the matching process is completed; the double-layer node graph includes: in-body nodes and far-body nodes, wherein the in-body nodes correspond to the inner-layer model, and the far-body nodes correspond to the outer-layer model; The three-dimensional measurement data of the human body to be measured is obtained according to the adjusted double-layer surface model.

2. The method according to claim 1, characterized in that The step of adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process includes: Matching the inner layer model with the three-dimensional point cloud data model to obtain an error between the inner layer model after parameter adjustment and the three-dimensional point cloud data model, and adjusting the parameters of the inner layer model based on the error; Based on the matched inner layer model, the point cloud data in the three-dimensional point cloud data model is integrated to match the outer layer model, and the parameters of the outer layer model are adjusted during the matching process; Synchronously binding and data fusing the matched inner layer model and the matched outer layer model, dynamically adjusting the double-layer node graph based on the motion characteristics of the three-dimensional point cloud data model, so as to perform joint motion tracking on the double-layer surface model, and optimizing the parameters of the double-layer surface model based on the effect of the joint motion tracking; It is determined whether the current double-layer surface model meets the model optimization cutoff condition. If so, the adjustment of the model parameters is stopped to obtain the adjusted double-layer surface model.

3. The method according to claim 2, characterized in that The matching of the inner layer model with the three-dimensional point cloud data model to obtain an error between the inner layer model after parameter adjustment and the three-dimensional point cloud data model, and adjusting the parameters of the inner layer model based on the error, comprises: The parameters of the inner layer model are adjusted with the minimization of the value of the energy function as the optimization goal; the energy function includes: a data term, a regularization term and a priori term; the data term is used to measure the difference between the inner layer model and the three-dimensional point cloud data model; the regularization term is used to constrain the deformation degree of the inner layer model; the priori term is used to guide the convergence of the inner layer model by using the prior knowledge of human body shape and posture.

4. The method according to claim 2, characterized in that: The dynamically adjusting the double-layer node graph based on the motion characteristics of the three-dimensional point cloud data model to perform joint motion tracking on the double-layer surface model includes: Controlling the rotation of the rotating device and driving the human body to be measured located on the rotating device to rotate synchronously; the depth map data specifically includes: a continuous data stream of the depth map generated according to the light reflected by the surface of the human body to be measured during the period of synchronous rotation with the rotating device; The double-layer surface model is controlled to rotate synchronously with the rotating device so that the double-layer surface model is matched with the corresponding target point cloud data in the three-dimensional point cloud data model at a target angle, and the parameters of the inner layer model and the outer layer model are optimized; the target point cloud data is point cloud data converted based on the target depth map in the depth map continuous data stream; the target depth map is a depth map generated based on the light reflected from the surface when the human body to be measured is rotated to the target angle.

5. The method according to claim 2, characterized in that: The determining whether the current double-layer surface model satisfies the model optimization cutoff condition includes: Determine whether the matching quality of the current double-layer surface model is qualified, and determine whether the joint motion tracking effect of the current double-layer surface model is qualified; If the matching quality is qualified and the effect of the joint motion tracking is qualified, it is determined that the current double-layer surface model meets the model optimization cutoff condition; If the matching quality is unsatisfactory, or the effect of the joint motion tracking is unsatisfactory, it is determined that the current double-layer surface model does not meet the model optimization cutoff condition.

6. The method according to claim 5, characterized in that The determining whether the matching quality of the current double-layer surface model is qualified includes: If the matching error of the current double-layer surface model is less than or equal to the preset error threshold, visual inspection shows that the surface of the current double-layer surface model is smooth and consistent with the point cloud, then the matching quality of the current double-layer surface model is judged to be qualified; If the matching error of the current double-layer surface model is greater than the preset error threshold, and a visual inspection shows that a key area of ​​the current double-layer surface model is deformed or broken, then it is determined that the matching quality of the current double-layer surface model is unqualified.

7. The method according to claim 5, characterized in that The determining whether the effect of the joint motion tracking of the current double-layer surface model is qualified includes: If the current double-layer surface model is stable in updating and always consistent with the point cloud when it moves at a preset speed or performs a gesture of preset complexity, the model matches the point cloud consistently across multiple frames, and the motion tracking accuracy is greater than a preset accuracy threshold, then the joint motion tracking effect of the current double-layer surface model is judged to be qualified; If the current double-layer surface model slides or misaligns when moving at a preset speed or performing a posture of preset complexity, causing the model to not match the actual posture, and if there is a freeze or distortion during the motion tracking process, then the joint motion tracking effect of the current double-layer surface model is judged to be unsatisfactory.

8. The method according to any one of claims 1 to 7, characterized in that The step of establishing a three-dimensional point cloud data model of the human body to be measured according to the depth map data comprises: Performing denoising and enhancement processing on the depth map data; The denoised and enhanced depth map data is converted into a 3D point cloud data model of the human body to be measured using a 3D reconstruction algorithm.

9. The method according to claim 8, characterized in that While the 3D structured light camera module collects light reflected from the surface of the human body to be measured, the method further includes: Collecting the position information and / or posture information of the human body to be measured by an inertial measurement unit; The method of using a three-dimensional reconstruction algorithm to convert the denoised and enhanced depth map data into a three-dimensional point cloud data model of the human body to be measured includes: Using a 3D reconstruction algorithm, the denoised and enhanced depth map data is converted into preliminary point cloud data; The position information and / or the posture information are used to provide constraints for point cloud registration, and preliminary point cloud data from multiple perspectives are registered and fused to form a three-dimensional point cloud data model of the human body to be measured.

10. The method according to claim 1, characterized in that While the 3D structured light camera module collects light reflected from the surface of the human body to be measured, the method further includes: Collecting the position information and / or posture information of the human body to be measured by an inertial measurement unit; When matching the three-dimensional point cloud data model with the double-layer surface model, the dynamic parameters of the double-layer surface model are optimized using the position information and / or the posture information.

11. A three-dimensional human body measurement device based on 3D structured light and double-layer surface model, characterized in that: The double-layer surface model includes an inner layer model and an outer layer model, wherein the inner layer model is used to capture the basic form and posture of the human body, and the outer layer model is used to capture the detailed features of non-rigid objects on the surface of the human body; The device comprises: A depth map generation module is used to emit light with structural features to the surface of the human body to be tested through a 3D structured light camera module, and capture the light reflected by the surface of the human body to be tested, and generate depth map data of the human body to be tested according to the light reflected by the surface of the human body to be tested; A point cloud conversion module, used to establish a three-dimensional point cloud data model of the human body to be measured according to the depth map data; A model matching and adjustment module, used for matching the three-dimensional point cloud data model with the double-layer surface model, adjusting the double-layer surface model based on the difference between the double-layer surface model and the three-dimensional point cloud data model and the double-layer node graph of the double-layer surface model during the matching process, and obtaining the adjusted double-layer surface model after the matching process is completed; the double-layer node graph includes: in-body nodes and far-body nodes, wherein the in-body nodes correspond to the inner layer model, and the far-body nodes correspond to the outer layer model; The measurement data acquisition module is used to obtain the three-dimensional measurement data of the human body to be measured according to the adjusted double-layer surface model.

Images