A system for analyzing and customizing mattresses based on 3D human body reconstruction technology

Abstract

The present invention discloses a system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology, including a human body posture monitoring module, a mattress support matrix, a signal acquisition module, a storage and control module, a control panel, a display module and a power supply module; through human body engineering mechanics modeling, and on the basis of Kinect Fusion, the deformation estimation of non-rigid bodies is added, and a real-time scanning splicing algorithm is implemented to perform three-dimensional human body reconstruction on the images continuously captured by the camera to obtain a model, and then the position where the human body contacts the mattress is obtained as the basis for judging the relative position relationship between the user's posture and the mattress support matrix; the pressure distribution data is combined with this model to construct a user's exclusive sleep model; when the user's posture changes, the mattress control system can monitor the user's position and sleeping posture in real time, and automatically follow the user's posture to adjust the support degree of each part of the mattress to maintain the optimal comfort level. Its advantage is: to achieve the technical effect of "the bed moves with the person".

Description

Technical Field

[0001] The present invention relates to the field of smart home, and in particular to a control system of a smart mattress. Background Art

[0002] With the continuous improvement of living standards and health awareness, people have higher and higher requirements for sleep quality. And there is a direct relationship between sleep quality and physical health. Modern medical research shows that sleep support is closely related to the health of human bones, muscles, blood circulation and other aspects. Sleep quality is related to mattresses. More than 90% of sleep problems are caused by improper sleep support related to mattresses. Traditional mattresses use springs, sponges, latex or air bags as the main support materials. The disadvantage is that the fixed structure cannot adapt to people's changing sleeping positions, and the pressure distribution of the mattress is not humane enough. Therefore, many people will experience "back pain after sleeping", "shoulder pain after getting up", "neck pain after sleeping" and so on when they get up in the morning. Summary of the invention

[0003] In order to solve the above-mentioned problems of pressure distribution in existing mattresses, the present invention provides a system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology to achieve scientific modeling of dynamic pressure distribution.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] A system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology, comprising a human body posture monitoring module, a mattress support matrix, a signal acquisition module, a storage and control module, a display module and a power supply module; the mattress support matrix is ​​arranged at the bottom of the mattress, controlled by the storage and control module, and is used to adjust the support degree of each part of the mattress to adjust the hardness, and is composed of a plurality of support members distributed in a matrix and with adjustable elongation;

[0006] The human posture monitoring module is used to monitor the contact position between the human body and the mattress, as a means of determining the relative position relationship between the user's posture and the mattress support matrix; the human posture monitoring module includes a camera that can monitor the entire mattress and a pressure sensor matrix that detects the pressure of the human body on the mattress; the camera includes an auxiliary fixing device that can drive the camera body to rotate around the mattress;

[0007] The signal acquisition module, whose input end is connected to the output end of the human posture monitoring module, is used to collect human position and posture data and pressure distribution data generated by the human body on the mattress from the human posture monitoring module;

[0008] The storage and control module, whose input end is connected to the output end of the signal acquisition module and whose output end is connected to the input end of the mattress support matrix, is used to store the sleep mode and output control instructions to the mattress support matrix and display data to the display module;

[0009] The power supply module is used to provide low-voltage DC power for the operation of each module in the system;

[0010] Based on the above modules, through human engineering mechanics modeling and on the basis of Kinect Fusion, the deformation estimation of non-rigid bodies is added, and the real-time scanning stitching algorithm is implemented to reconstruct the three-dimensional human body model from the images continuously captured by the camera, and then the contact position between the human body and the mattress is obtained, which serves as the basis for judging the relative position relationship between the user's posture and the mattress support matrix; the pressure distribution data is combined with this model to build a user's exclusive sleep model; when the user's posture changes, the mattress control system can monitor the user's position and sleeping posture in real time, and automatically adjust the support level of each part of the mattress according to the user's posture to maintain the optimal comfort level;

[0011] The method is characterized in that it comprises the following steps:

[0012] Step 1: Depth data preprocessing, using the camera intrinsic parameters and the depth frame mapping to camera space function to convert the depth map into a 3D point cloud, and then calculate the normal vector of each point; the camera intrinsic parameters include fx, fy, cx, cy and three radial distortion parameters, and the camera space function is provided by the Kinect2.0 SDK;

[0013] Step 2: Camera tracking: ICP matching is performed between the 3D point cloud converted from the current frame and the predicted 3D point cloud generated by the existing model to obtain the current camera pose.

[0014] Step 3: Deep data fusion: According to the calculated current camera pose, the TSDF point cloud fusion algorithm is used to fuse the 3D point cloud of the current frame into the existing model to achieve TSDF character surface model reconstruction;

[0015] Step 4: Scene rendering, using ray tracing computer graphics methods, combined with the existing model and the current camera pose to predict the environment point cloud observed by the current camera; the obtained environment point cloud is used for display and also provided to step 2 for ICP matching. ICP is the abbreviation of Iterative Closest Point, which refers to the iterative closest point cloud matching algorithm.

[0016] Preferably, in step 1, when the system starts to start, the human body posture monitoring module starts to monitor the user's body position and posture, and the camera of Kinect, which is a single depth sensor, circles around the user lying on the mattress as the center, and the human body position and posture data is collected through the signal acquisition module. The system uses the Dynamic Fusion three-dimensional human body model reconstruction technology in computer graphics to perform real-time reconstruction of the character surface model.

[0017] Preferably, the system adjusts the human body model in combination with the pressure distribution data collected by the signal acquisition module, so that the parts of the human body that were originally suspended in the air can obtain sufficient support from the mattress, that is, the pressure exerted on the mattress by the protruding parts of the human body is evenly dispersed, so that the user's body is in the most comfortable posture on the mattress; the storage and control module automatically stores the height change data of the support under the current sleeping position, and the height change data and the human body position and posture data together constitute the sleeping mode under this posture.

[0018] Preferably, the support member is a hydraulic spring, which can change its elongation under the drive of an electrical signal, and change the hardness of the corresponding position of the mattress through the supporting force of the hydraulic spring; the hydraulic spring can tolerate the reaction force of the mattress to fluctuate within a set interval value, and when it exceeds the set interval value, the hydraulic spring can adjust the elongation in accordance with the reaction force, and restore to the original elongation when it returns to the set interval value. A locker is provided in the hydraulic spring to prevent the hydraulic spring from being damaged by excessive external force.

[0019] Preferably, the storage and control module can record several sleep modes, including deep sleep mode, light sleep mode, deep sleep mode at night, nap mode, and massage sleep mode.

[0020] Preferably, the storage and control module can store sleep mode data corresponding to several different sleeping positions; when the user changes the sleeping position, the storage and control module automatically matches the corresponding comfortable sleeping position. When the user's habitual sleeping position cannot be matched, the user is marked as not in normal state, and the storage and control module sends a reminder signal to the display module.

[0021] Preferably, the mattress is made of coconut palm, memory foam or latex.

[0022] Preferably, the ICP matching comprises the following steps: (1) matching each point in the input source point cloud with the nearest point in the reference point cloud; (2) estimating a combination of rotation and translation by minimizing the root mean square point-to-point distance metric to optimally align each source point with the matching point obtained in step (1); (3) transforming the source point using the obtained transformation; and (4) reselecting associated points and iterating the above process.

[0023] Preferably, the TSDF character surface model reconstruction includes the following steps: (1) the point cloud model after ICP matching is transformed from the world coordinate system to the camera coordinate system through the rotation and translation matrix; (2) the point cloud coordinates in the camera coordinate system are transformed into the screen image coordinate system through the intrinsic parameter matrix of the Kinect camera, and the depth H of each point is obtained; (3) by comparing the image depth H and the coordinate Z of the corresponding coordinate point, the distance between the point cloud of different frames and the camera screen can be obtained; (4) this distance is defined as D, and the weight W is assigned, and D and W in different frames are calculated and updated; (5) the surface vertices intersecting under multiple point clouds are obtained using a preset distance cutoff range; (6) the above process is iterated to obtain the final model surface point cloud, constitute the character surface model, and complete the TSDF character surface model reconstruction.

[0024] As a principle description of the technical solution of the present invention, Kinect Fusion uses the depth data of a low-cost single depth sensor (Kinect) to perform real-time 3D reconstruction, and proposes a method of combining GPU and TSDF model to reconstruct the scene in real time. Kinect Fusion has achieved the first success in achieving dense real-time 3D reconstruction by utilizing the high parallelism of GPU, and the reconstruction effect is very stable and excellent. The relationship between its four modules is as follows: Figure 1 As shown. Dynamic Fusion is further implemented based on KinectFusion. Dynamic Fusion solves the problem of how to perform real-time surface reconstruction when the measured object is moving at the same time. Even when the surface of the object is deformed, the surface shape of the object can be accurately restored in real time. Dynamic Fusion converts the dynamically changing scene (object to be reconstructed) obtained in each frame into a canonical space (world space) through some kind of transformation, that is, creates or updates a static object surface model in this space. At the same time, each frame has a corresponding volumetric warp field, which can restore the model in the canonical space to the live frame (camera observation space).

[0025] Through the reasonable arrangement of each module of the system, under the pressure of the user, the pressure change of the mattress is fed back in a closed loop, and the force between the mattress and the corresponding position of the human body is adjusted in real time, forming an intelligent automatic pressure adjustment of the mattress. The user's posture and pressure distribution on the mattress can be quickly monitored in real time, and the pressure at various locations on the mattress can be fed back and adjusted, so as to achieve a pressure distribution state that satisfies and is comfortable for the user.

[0026] As can be seen from the above, compared with the prior art, the present invention has the following advantages: the present invention innovatively proposes to use human engineering mechanics modeling based on real-time pressure distribution data during sleep; and on the basis of Kinect Fusion, it adds deformation estimation of non-rigid bodies, and implements real-time scanning and stitching algorithm to perform three-dimensional human body reconstruction on images continuously captured by the camera to obtain a model, and then obtain the position where the human body contacts the mattress, which serves as the basis for judging the relative position relationship between the user's posture and the mattress support matrix; the pressure distribution data is combined with this model to construct a user's exclusive sleep model; and when the user's posture changes, the mattress control system can monitor the user's position and sleeping posture in real time, and automatically follow the user's posture to adjust the support level of various parts of the mattress to maintain the optimal comfort level, and achieve the technical effect of "the bed moves with the person". BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a schematic diagram of the Kinect Fusion module structure.

[0028] Figure 2 This is the ICP algorithm flow of the present invention.

[0029] Figure 3 It is a structural block diagram of the system of the present invention.

[0030] Figure 4 It is the working flow chart of the system of the present invention.

[0031] Description of the accompanying figures: 1-human posture monitoring module, 2-pressure support matrix, 3-signal acquisition module, 4-storage and control module, 5-display module, 6-power supply module. DETAILED DESCRIPTION

[0032] The present invention and its beneficial technical effects are further described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0033] See also Figure 1 to Figure 4 The present invention preferably implements a system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology, comprising a human body posture monitoring module 1, a mattress support matrix 2, a signal acquisition module 3, a storage and control module 4, a display module 5 and a power supply module 6; the mattress support matrix 2 is arranged at the bottom of the mattress, controlled by the storage and control module 4, and is used to adjust the support degree of each part of the mattress to adjust the hardness, and is composed of a plurality of support members distributed in a matrix and with adjustable elongation;

[0034] The human posture monitoring module 1 is used to monitor the contact position between the human body and the mattress, as a means of determining the relative position relationship between the user's posture and the mattress support matrix 2; the human posture monitoring module 1 includes a camera that can monitor the entire mattress and a pressure sensor matrix that detects the pressure of the human body on the mattress; the camera includes an auxiliary fixing device that can drive the camera body to rotate around the mattress;

[0035] The signal acquisition module 3, whose input end is connected to the output end of the human posture monitoring module 1, is used to collect human position and posture data and pressure distribution data generated by the human body on the mattress from the human posture monitoring module 1;

[0036] The storage and control module 4, whose input end is connected to the output end of the signal acquisition module 3, and whose output end is connected to the input end of the mattress support matrix 2, is used to store the sleep mode and output control instructions to the mattress support matrix 2 and display data to the display module 5;

[0037] The power supply module 6 is used to provide low-voltage DC power for the operation of each module in the system;

[0038] On the basis of the above modules, through human engineering mechanics modeling, and on the basis of Kinect Fusion, the deformation estimation of non-rigid bodies is added, and the real-time scanning stitching algorithm is implemented to reconstruct the three-dimensional human body model from the images continuously captured by the camera, and then the contact position between the human body and the mattress is obtained, which is used as the basis for judging the relative position relationship between the user's posture and the mattress support matrix 2; the pressure distribution data is combined with this model to build a user's exclusive sleep model; when the user's posture changes, the mattress control system can monitor the user's position and sleeping posture in real time, and automatically adjust the support level of each part of the mattress according to the user's posture to maintain the optimal comfort level;

[0039] The method is characterized in that it comprises the following steps:

[0040] Step 1: Depth data preprocessing, using the camera intrinsic parameters and the depth frame mapping to camera space function to convert the depth map into a 3D point cloud, and then calculate the normal vector of each point; the camera intrinsic parameters include fx, fy, cx, cy and three radial distortion parameters, and the camera space function is provided by the Kinect2.0 SDK;

[0041] Step 2: Camera tracking, ICP matching is performed on the 3D point cloud converted from the current frame and the predicted 3D point cloud generated by the existing model to obtain the current camera pose; wherein ICP is the abbreviation of Iterative Closest Point, which refers to the iterative closest point point cloud matching algorithm. The ICP algorithm process is shown in Figure 2; preferably, the ICP matching includes the following steps: (1) for each point in the input source point cloud (usually from a dense entire vertex set or a selection of vertex pairs for each model), match the closest point in the reference point cloud; (2) by minimizing the root mean square point-to-point distance metric to estimate the combination of rotation and translation, each source point is optimally aligned with the matching point obtained in step (1); (3) the source point is transformed using the obtained transformation; (4) the associated point is reselected and the above process is iterated.

[0042] Step 3: Depth data fusion. According to the calculated current camera pose, the TSDF point cloud fusion algorithm is used to fuse the 3D point cloud of the current frame into the existing model to achieve TSDF character surface model reconstruction. TSDF is the abbreviation of Truncated Signed Distance Functions, which refers to the truncated signed distance function, which allows the integration of multiple depth images obtained from different viewpoints. TSDF is different from Kinect Fusion in that DynamicFusion introduces projective TSDF, abbreviated as PSDF, is actually the calculation of TSDF value in camera space; preferably, the TSDF character surface model reconstruction includes the following steps: (1) After ICP matching, the point cloud model is transformed from the world coordinate system to the camera coordinate system through rotation and translation matrices; (2) The point cloud coordinates in the camera coordinate system are transformed to the screen image coordinate system through the intrinsic parameter matrix of the Kinect camera, and the depth H of each point is obtained; (3) By comparing the image depth H and the coordinate Z of the corresponding coordinate point, the distance between the point cloud of different frames and the camera screen can be obtained; (4) This distance is defined as D, and a weight W is assigned, and D and W in different frames are calculated and updated; (5) A preset distance cutoff range is used to obtain the surface vertices intersecting under multiple point clouds; (6) The above process is iterated to obtain the final model surface point cloud, constitute the character surface model, and complete the TSDF character surface model reconstruction.

[0043] Step 4: Scene rendering, using ray tracing computer graphics methods, combined with the existing model and the current camera pose to predict the environment point cloud observed by the current camera; the obtained environment point cloud is used for display and also provided to step 2 for ICP matching.

[0044] Preferably, the support member is a hydraulic spring, which can change its elongation under the drive of an electrical signal, and change the hardness of the corresponding position of the mattress through the supporting force of the hydraulic spring; the hydraulic spring can tolerate the reaction force of the mattress to fluctuate within a set interval value, and when it exceeds the set interval value, the hydraulic spring can adjust the elongation in accordance with the reaction force, and restore to the original elongation when it returns to the set interval value. A locker is provided in the hydraulic spring to prevent the hydraulic spring from being damaged by excessive external force.

[0045] Preferably, the storage and control module 4 can record several sleep modes, including a deep sleep mode, a light sleep mode, a deep sleep mode at night, a nap mode, and a massage sleep mode.

[0046] Preferably, the storage and control module 4 can store sleep mode data corresponding to several different sleeping positions; when the user changes the sleeping position, the storage and control module 4 automatically matches the corresponding comfortable sleeping position. When the user's habitual sleeping position cannot be matched, the user is marked as not in normal state, and the storage and control module 4 sends a reminder signal to the display module 5.

[0047] Preferably, the mattress is made of coconut palm, memory foam or latex.

[0048] See also Figure 4 , the system workflow is as follows:

[0049] Step 1: When the system starts, the human posture monitoring module 1 starts to monitor the user's body position and posture. The camera of Kinect, which is a single depth sensor, circles around the user lying on the mattress. The signal acquisition module collects the body position and posture data. The system uses the Dynamic Fusion three-dimensional human model reconstruction technology in computer graphics to reconstruct the character surface model in real time.

[0050] Step 2: The system adjusts the human body model in combination with the pressure distribution data collected by the signal acquisition module 3, so that the originally suspended parts of the human body can obtain sufficient support from the mattress, that is, the pressure exerted on the mattress by the protruding parts of the human body is evenly dispersed, that is, the contact area between the human body and the mattress is increased, thereby reducing the reaction force of the mattress on the protruding parts of the human body, so that the user's body is in the most comfortable posture on the mattress; the storage and control module 4 automatically stores the height change data of the support under the current sleeping posture, and the height change data and the human body position and posture data together constitute the sleeping mode under this posture.

[0051] Step 3: When the user changes his sleeping position, repeat steps 1 and 2 to correct the sleeping position to maintain the optimal comfort level.

[0052] Step 4: When the user leaves the bed to do other activities (for example, going to the bathroom), the system switches to standby mode; when the user lies back in bed, the body position and posture information is collected again, and the relevant sleep mode data is retrieved and re-executed. The system can quickly adapt to the same sleeping posture of the same user without repeated adjustments, and the execution efficiency is high.

[0053] When in use, the signal acquisition module 3, the storage and control module 4, the display module 5 and the power supply module 6 are integrated into a controller, which is set on or near the mattress, and is electrically connected to the relevant interfaces of the human posture monitoring module 1 and the mattress support matrix 2 according to specific wiring requirements. The above-mentioned various algorithms or control processes are stored in the storage and control module 4 in the form of a control execution program, such as an MCU or a program memory outside the MCU.

[0054] In the above description, the contents commonly used in the prior art, such as conventional algorithms, physical structures and processes, are not described in detail in order to save space. The processing techniques and parts not disclosed can be processed according to the conventional techniques in the prior art.

[0055] The above description and specific embodiments do not constitute any limitation to the present invention. The present invention is not limited to the specific implementation modes disclosed and described above. Some modifications and variations of the present invention should also fall within the protection scope of the claims of the present invention.

Claims

1. A system for analyzing and customizing a mattress based on three-dimensional human body reconstruction technology, comprising a human body posture monitoring module (1), a mattress support matrix (2), a signal acquisition module (3), a storage and control module (4), a display module (5) and a power supply module (6); the mattress support matrix (2) is arranged at the bottom of the mattress, is controlled by the storage and control module (4), is used to adjust the support degree of each part of the mattress to achieve hardness adjustment, and is composed of a plurality of support members distributed in a matrix and with adjustable elongation; The human posture monitoring module (1) is used to monitor the contact position between the human body and the mattress, as a means of determining the relative positional relationship between the user's posture and the mattress support matrix (2); the human posture monitoring module (1) comprises a camera capable of monitoring the entire mattress and a pressure sensor matrix capable of detecting the pressure exerted by the human body on the mattress; the camera comprises an auxiliary fixing device capable of driving the camera body to rotate around the mattress; The signal acquisition module (3) has an input end connected to the output end of the human posture monitoring module (1) and is used to collect human position and posture data and pressure distribution data generated by the human body on the mattress from the human posture monitoring module (1); The storage and control module (4) has an input end connected to the output end of the signal acquisition module (3), and an output end connected to the input end of the mattress support matrix (2), and is used to store the sleep mode and output control instructions to the mattress support matrix (2) and display data to the display module (5); The power supply module (6) is used to provide low-voltage DC power for the operation of each module in the system; On the basis of the above modules, through human engineering mechanics modeling and based on Kinect Fusion, non-rigid body deformation estimation is added to realize real-time scanning and stitching algorithm to reconstruct the three-dimensional human body model from the images continuously captured by the camera, and then obtain the contact position between the human body and the mattress, which is used as the basis for judging the relative position relationship between the user's posture and the mattress support matrix (2); the pressure distribution data is combined with this model to construct a user's exclusive sleep model; when the user's posture changes, the mattress control system can monitor the user's position and sleeping posture in real time, and automatically adjust the support level of each part of the mattress according to the user's posture to maintain the optimal comfort level; The method is characterized in that it comprises the following steps: Step 1: Depth data preprocessing: convert the depth map into a 3D point cloud using the camera intrinsic parameters and the depth frame mapping to camera space function, and then calculate the normal vector of each point; the camera intrinsic parameters include fx, fy, cx, cy and three radial distortion parameters, and the camera space function is provided by the Kinect2.0 SDK; Step 2: Camera tracking, ICP matching the 3D point cloud converted from the current frame and the predicted 3D point cloud generated by the existing model to obtain the current camera pose; Step 3: Deep data fusion: According to the calculated current camera pose, the TSDF point cloud fusion algorithm is used to fuse the 3D point cloud of the current frame into the existing model to achieve TSDF character surface model reconstruction; Step 4: Scene rendering, using ray tracing computer graphics methods, combined with the existing model and the current camera pose to predict the environment point cloud observed by the current camera; the obtained environment point cloud is used for display and also provided to step 2 for ICP matching.

2. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 1, characterized in that: In step 1, when the system starts to start, the human posture monitoring module (1) starts to monitor the user's human body position and posture, and uses the camera of Kinect, which is a single depth sensor, to circle around the user lying on the mattress, and collects human body position and posture data through the signal acquisition module (3). The system uses the Dynamic Fusion three-dimensional human body model reconstruction technology in computer graphics to reconstruct the human body surface model in real time.

3. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 1 or 2, characterized in that: The system adjusts the human body model in combination with the pressure distribution data collected by the signal collection module (3), so that the parts of the human body that were originally suspended in the air can obtain sufficient support from the mattress, that is, the pressure exerted on the mattress by the protruding parts of the human body is evenly dispersed, so that the user's body is in the most comfortable posture on the mattress; The storage and control module (4) automatically stores the height change data of the support member in the current sleeping posture, and the height change data and the human body position and posture data together constitute the sleeping mode in the posture.

4. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 3, characterized in that: The support member is a hydraulic spring, which can change its elongation under the drive of an electrical signal, and change the hardness of the corresponding position of the mattress through the supporting force of the hydraulic spring; the hydraulic spring can tolerate the reaction force of the mattress to fluctuate within the range of a set interval value. When it exceeds the range of the set interval value, the hydraulic spring can adjust the elongation in accordance with the reaction force, and restore to the original elongation when it returns to the range of the set interval value. A locker is provided in the hydraulic spring to prevent the hydraulic spring from being damaged by excessive external force.

5. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 4, characterized in that: The storage and control module (4) can record a number of sleep modes, including a deep sleep mode, a light sleep mode, a deep sleep mode at night, a nap mode, and a massage sleep mode.

6. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 5, characterized in that: The storage and control module (4) can store sleep mode data corresponding to a plurality of different sleeping positions; when the user changes the sleeping position, the storage and control module (4) automatically matches the corresponding comfortable sleeping position; when the sleeping position that the user is accustomed to cannot be matched, the user is marked as not in the normal state, and the storage and control module (4) sends a reminder signal to the display module (5).

7. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 6, characterized in that: The mattress is made of coconut palm, memory foam or latex.

8. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 1 or 2, characterized in that: The ICP matching method comprises the following steps: (1) matching each point in the input source point cloud to the nearest point in the reference point cloud; (2) estimating a combination of rotation and translation by minimizing the root mean square point-to-point distance metric to optimally align each source point with the matching point obtained in step (1); (3) transforming the source point using the obtained transformation; and (4) reselecting the associated points and iterating the above process.

9. The system for analyzing and customizing mattresses based on three-dimensional human body reconstruction technology as claimed in claim 1 or 2, characterized in that: The TSDF character surface model reconstruction includes the following steps: (1) After ICP matching, the point cloud model is transformed from the world coordinate system to the camera coordinate system through the rotation and translation matrix; (2) The point cloud coordinates in the camera coordinate system are transformed into the screen image coordinate system through the intrinsic parameter matrix of the Kinect camera, and the depth H of each point is obtained; (3) By comparing the image depth H and the coordinate Z of the corresponding coordinate point, the distance between the point cloud in different frames and the camera screen can be obtained; (4) This distance is defined as D, and the weight W is assigned, and D and W in different frames are calculated and updated; (5) The surface vertices intersecting under multiple point clouds are obtained using a preset distance cutoff range; (6) The above process is iterated to obtain the final model surface point cloud, constitute the character surface model, and complete the TSDF character surface model reconstruction.

Images