TWI764393B - Manufacturing method of pressure garment - Google Patents

Abstract

The invention discloses a manufacturing method of a pressure garment, which includes a scanning step and a manufacturing step. In scanning step includes: using the processing device to first control a plurality of depth cameras to shoot the user to generate a plurality of first model files; controlling the plurality of depth cameras and the carrying device to rotate a predetermined angle relative to each other and then control the plurality of depth cameras to shoot the user again to generate a plurality of second model files; converting each first model file and each second model file into the same world coordinates through two times coordinate conversions, and then fusion all model files to generate a three-dimensional model file. The manufacturing step is to calculate the size of the injured area using the three-dimensional model file and the injured area data of the user's demand information, and then control the production equipment to produce the pressure garment based on the injured area size and the pressure data of the user's demand information.

Description

Manufacturing method of pressure fabric

The invention relates to a manufacturing method of a pressure fabric, in particular to a manufacturing method of a pressure fabric which does not need to manually measure the position of a user's affected part to manufacture the pressure fabric.

Since the pressure garment creates a negative pressure on the surface of the human body to form a tightening effect, if the pressure distribution on the surface of the human body is to achieve the desired medical effect, the pressure garment must be customized to a certain extent according to the body shape of the individual user. For this reason, the pressure garment manufacturers must measure the affected part of the patient many times, and the patient must try it on many times. When measuring the size of the injured part, the personnel must touch the injured part, which may cause discomfort to the injured.

The invention discloses a manufacturing method of a pressure fabric, which is mainly used to improve the traditional manufacturing method of pressure. The affected part of the injured must be contacted manually to measure the affected part, which will lead to the problem of discomfort of the injured.

One embodiment of the present invention discloses a method for manufacturing a pressure fabric, which includes the following steps: using a scanning system to perform a scanning step to perform a three-dimensional scanning operation on a user; the scanning system includes a processing device and a carrying device , a plurality of supporting structures, a plurality of depth cameras (Depth Camera) and a rotating mechanism, the processing device is electrically connected to the rotating mechanism, and the processing device can control the rotating mechanism, so that the plurality of supporting structures and the carrying device rotate relative to each other, and the carrying device has A bearing surface, the bearing surface is used for the user to stand, a plurality of supporting structures are arranged around the bearing device, each supporting structure is provided with a depth camera, and the horizontal distance between each depth camera and the bearing device is the same; the scanning system can be used in When the user stands on the bearing surface, the scanning step is performed, and the scanning step includes: a first model file generating step: controlling each depth camera to photograph the user standing on the bearing surface, so that each depth camera generates a first model performing a second model file generating step M times, where M is a positive integer greater than or equal to 1, and the second model file generating step includes: firstly controlling the rotation mechanism to make each depth camera and the carrying device rotate relative to each other by a predetermined amount angle, and then control each depth camera to photograph the user, so that each depth camera generates a second model file; a camera and a carrier device coordinate conversion step: use the processing device to convert the first model file of each depth camera , carry out a camera and carrier device coordinate conversion program corresponding to each depth camera, so as to convert the first model file and the second model file generated by each depth camera into a first conversion model file and a second model file respectively Converting model files; a world coordinate conversion step: using the processing device to perform a camera and camera coordinate conversion program on the first conversion model file and the second conversion model file of each depth camera to convert the first conversion model file of each depth camera. A conversion model file and a second conversion model file are respectively converted into a first world coordinate model file and a second world coordinate model file; a fusion step: convert all the first world coordinate model files and all the second world coordinate model files file, and fused into a three-dimensional model file; a manufacturing step: calculating an affected part size according to the three-dimensional three-dimensional model file and an affected part data of a user's demand information, and then applying pressure according to the affected part size and the user's demand information data, control a production equipment to produce a pressure fabric.

To sum up, the manufacturing method of the pressure fabric of the present invention establishes the user's three-dimensional three-dimensional model file in a non-contact manner through the depth camera, and then uses the three-dimensional three-dimensional model file to manufacture the pressure garment. During the process of the pressure garment, the relevant personnel basically do not need to touch the affected part of the injured patient, so as to avoid the problem of discomfort caused by the affected part of the patient being touched.

In order to further understand the features and technical content of the present invention, please refer to the following detailed description and accompanying drawings of the present invention, but these descriptions and drawings are only used to illustrate the present invention, rather than make any claims to the protection scope of the present invention. limit.

In the following description, if it is indicated to refer to a specific figure or as shown in a specific figure, it is only used for emphasis in the subsequent description, and most of the related content mentioned appears in the specific figure, However, it is not limited that only the specific drawings may be referred to in this subsequent description.

Please refer to FIG. 1 and FIG. 2 together. In the manufacturing method of the pressure fabric of the present invention, a scanning system A is used to perform a scanning step S1 to perform a three-dimensional three-dimensional scanning operation on a user to obtain a three-dimensional three-dimensional model file A11 , and then control a production equipment (not shown, such as various textile equipment) to manufacture a pressure fabric according to the three-dimensional three-dimensional model file A11 and a user's demand information.

As shown in FIG. 2 , the scanning system A may include a processing device A1 , a carrying device A2 , a rotating mechanism (not shown), four supporting structures A3 and four depth cameras A4 (Depth Camera). The processing device A1 is electrically connected to the rotating mechanism, and the processing device A1 can control the rotating mechanism to make the four supporting structures A3 and the carrying device A2 rotate relative to each other. For example, the processing device A1 may include various microprocessors, and in different embodiments, the processing device A1 may also be various computers, servers, and the like.

The carrying device A2 has a carrying surface A21, and the carrying surface A21 is used for the user to stand. Four supporting structures A3 are arranged around the carrier A2. The appearance of the carrying device A2 is not limited to what is shown in the figure. For example, the rotating mechanism may include a turntable (not shown) and a motor (not shown), the turntable is connected with the carrying device A2, and the turntable is connected with the motor, and the processing device A1 can control the motor to operate, so that the turntable rotates a predetermined angle. The specific components included in the rotating mechanism are not limited to the above, and any components that can make the carrying device A2 and the plurality of supporting structures A3 rotate relative to each other belong to the practical scope of the rotating mechanism. In the present embodiment, the carrying device A2 can be driven by the rotating mechanism to rotate relative to the four supporting structures A3 as an example, but it is not limited thereto. In different embodiments, the carrying device A2 may also be a fixed member, and the four supporting structures A3 can be driven by the rotating mechanism to rotate relative to the carrying device A2.

As shown in FIGS. 1 and 3 , FIG. 3 is a top view of the scanning system A. As shown in FIG. Each support structure A3 is provided with one depth camera A4, and the four depth cameras A4 are generally located on the same plane. Specifically, the optical axes of the four depth cameras A4 are located on the same plane. In practical applications, the four supporting structures A3 may be connected to a fixed structure A5 together, and in the top view of the scanning system A, the four depth cameras A4 may be approximately located at 12 o'clock, 3 o'clock, At the positions of 6 o'clock and 9 o'clock, that is, the included angle P1 between the optical axes OA of the two adjacent depth cameras A4 is 90 degrees.

In the embodiment in which the scanning system A has three supporting structures A3, and each supporting structure A3 is provided with one depth camera A4, the angle between the optical axes of the two adjacent depth cameras A4 may be 120 degrees. . That is to say, in the embodiment where each support structure A3 is provided with only one depth camera A4, the angle P1 between the optical axes of the two depth cameras A4 is equal to the number Q of the support structures A3 included in the scanning system A. The relationship can be: 360/Q=P1.

As shown in FIG. 1 and FIG. 3 , when the processing device A1 controls the operation of the rotating mechanism, the carrying device A2 will rotate relative to the four supporting structures A3 with a central axis C as the center, and each depth camera A4 is connected to the central axis C. The horizontal distances L are the same. Specifically, the four depth cameras A4 can be the same depth cameras, and the horizontal distances between the photosensitive elements of each depth camera A4 and the central axis C are the same as each other.

In the drawings of this embodiment, a single support structure A3 is provided with a single depth camera A4, but the number of the depth cameras A4 set on a single support structure A3 is not limited thereto. In different embodiments, a single support structure A3 may be provided with more than two depth cameras A4. In the embodiment in which a single support structure A3 is provided with two depth cameras A4, the two depth cameras A4 may be arranged on the support structure A3 up and down, and the two depth cameras A4 may be respectively used to scan the upper body of the user and the lower body; in contrast, in the embodiment in which a single support structure A3 is provided with three depth cameras A4, the three depth cameras A4 may scan different parts of the human body respectively.

As shown in FIG. 1 and FIG. 2 , the scanning system A can execute the scanning step S1 when the user stands on the bearing surface A21 . In practical applications, the carrying device A2 may be provided with a detector, for example, the detector is used to detect whether a user is carried on the carrying surface A21 , and the processing device A1 can execute sequentially according to the detection result of the detector. The scanning step S1 and a manufacturing step S2. The scanning step S1 includes:

A first model file generating step S11: controlling each depth camera A4 to photograph a user standing on the bearing surface A21, so that each depth camera A4 corresponds to a first model file A41;

Execute M times a second model file generation step S12, M is a positive integer greater than or equal to 1, the second model generation step S12: firstly control the rotation mechanism to make each depth camera A4 and the carrying device A2 rotate relative to each other a predetermined angle, and then control each depth camera A4 to photograph the user, so that each depth camera A4 correspondingly generates a second model file A42;

A camera and carrier device coordinate conversion step S13: Using the processing device A1 to perform a camera and carrier device coordinate conversion program corresponding to each depth camera A4 on the first model file A41 of each depth camera A4, to convert each depth camera A4 The first model file A41 and the second model file A42 of the depth camera A4 are respectively converted into a first conversion model file A12 and a second conversion model file A13;

A world coordinate conversion step S14: using the processing device A12 to perform a camera and camera coordinate conversion program on the first conversion model file A12 and the second conversion model file A13 of each depth camera A4, to convert the depth camera A4 The first conversion model file A12 and the second conversion model file A13 are respectively converted into a first world coordinate model file A14 and a second world coordinate model file A15;

A fusion step S15: fuse all the first world coordinate model files A14 and all the second world coordinate model files A15 into a three-dimensional three-dimensional model file A11;

A manufacturing step S2: Calculate an affected part size according to the three-dimensional model file A11 and an affected part data of a user's demand information, and then control the production equipment to produce a pressure fabric according to the affected part size and a pressure data of the user's demand information.

In practical applications, after the processing device A1 controls each depth camera A4 to generate the first model file A41, the processing device A1 may then perform a noise removal on the point cloud included in the first model file operation and an outlier removal operation; similarly, after the processing device A1 controls each depth camera A4 to generate the second model file A42, the processing device A1 may then proceed to analyze the point cloud ( point cloud) to remove noise and outliers; in this way, it will help to fuse a better three-dimensional model file in the fusion step S15. In addition, a surface reconstruction step may be included after the fusion step S15 : performing a 3D Surface Reconstruction procedure on the three-dimensional model, so that the result of the three-dimensional model file can be better.

The number of times that the second model file generating step S12 is performed may be designed according to the support structure A3, the depth camera A4, the predetermined angle, and the field of view (FOV) of the depth camera A4. It is assumed that each support structure A3 only has If a single depth camera A4 is set, the more the number of support structures A3, the smaller the predetermined angle can be. On the contrary, the smaller the number of support structures A3, the larger the predetermined angle; it is assumed that each support structure A3 is only provided with a single If the depth camera A4 is used, the greater the number of support structures A3, the depth camera A4 with a relatively small field of view can be selected. Otherwise, the depth camera A4 with a relatively large field of view needs to be selected.

In order to facilitate the fusion of a more accurate three-dimensional three-dimensional model file A11 in the fusion step S14, in a preferred embodiment, the point cloud and the first model file A41 included in the first model file A41 generated by each depth camera A4 The overlap ratio of the point clouds included in the two model files A42 is greater than 60%. In the embodiment in which the second model file generating step is performed more than twice, the overlap ratio of the point clouds included in the two second model files A42 generated successively is greater than 60%. Similarly, if the two adjacent depth cameras A4 are respectively defined as a first depth camera and a second depth camera, then the point cloud (point cloud) included in the first model file generated by the first depth camera , the overlap ratio with the point cloud (point cloud) included in the first model file A41 generated by the second depth camera is greater than 60%; the point cloud (point cloud) included in the second model file generated by the first depth camera cloud), the overlap ratio with the point cloud (point cloud) contained in the first model file generated by the second depth camera is greater than 60%, which will help to fuse a more accurate three-dimensional stereo in the fusion step S14. Model file A11.

The first model file A41 and the second model file A42 are the data sets directly output by the depth camera A4, respectively, and the first model file A41 and the second model file A42 may be X containing each pixel of the image captured by the depth camera A4. Coordinates, Y coordinates, Z coordinates, depth values, etc. In the manufacturing step S2, the processing device A1 or the production equipment may be controlled to calculate the size of the affected part according to the three-dimensional model file A11 and the data of the affected part, which is not limited here. The depth camera A4 can capture images by, for example, structured light technology or photogrammetry.

In the embodiment where the scanning system A has 4 depth cameras A4, the processing device A1 will store 4 cameras and a rigid transformation matrix of the carrying device, 4 cameras and 4 cameras and cameras Rigid transformation matrix (camera-camera rigid transformation matrix). In the camera and carrier device coordinate conversion step S13, when the processing device A1 performs the camera and carrier device coordinate conversion procedure on each depth camera A4, the processing device A1 uses the camera and carrier device rigid conversion corresponding to the respective depth cameras A4. matrix, and convert the first model file A41 and the second model file A42 of the depth camera A4 into the first conversion model file A12 and the second conversion model file A13, respectively. In the world coordinate conversion step S14, when the processing device A1 performs the camera and camera coordinate conversion program for each depth camera A4, the processing device A1 uses the camera and camera rigid conversion matrix corresponding to the depth camera A4 to convert the depth camera. The first conversion model file A12 and the second conversion model file A13 of A4 are converted into a first world coordinate model file A14 and a second world coordinate model file A15, respectively.

It is worth mentioning that the affected part data is used to describe the position and range of the affected part of the human body, and the processing device A1 can know the position and range of the affected part to be covered by the pressure fabric through the affected part data. The applied pressure data is used to describe the pressure value applied by the pressure fabric to the affected part when the pressure fabric is covered on the affected part, and the processing device A1 can know how much pressure the pressure fabric needs to provide when the pressure fabric is covered on the affected part through the applied pressure data. surface of the human body.

According to the above, the manufacturing method of the pressure fabric of the present invention can quickly and accurately manufacture the pressure fabric that meets the needs of the user through the design of the scanning step S1 and the manufacturing step S2. The existing traditional manufacturing method of the pressure fabric is to manually measure the relative size of the part where the user wants to wear the pressure fabric, which is not only time-consuming and labor-intensive, but also the measurement results of different people may be different due to various factors; On the other hand, the manufacturing method of the pressure fabric of the present invention can avoid various problems caused by the above manual measurement by scanning with multiple depth cameras, and the scanning step S1 in the manufacturing method of the pressure fabric of the present invention , which can be completed within 1 minute, while the traditional manual measurement method takes at least 15 to 30 minutes.

Please refer to FIG. 4 and FIG. 5 together. FIG. 4 is a schematic flowchart of one embodiment of the manufacturing method of the pressure fabric of the present invention, and FIG. 5 is a schematic diagram of a display device displaying a three-dimensional image corresponding to a three-dimensional model file. . The biggest difference between this embodiment and the previous embodiment is that after the fusion step S15 , it further includes: a feature calibration step S16 , a display step S17 and a judgment step S18 .

The feature calibration step S16 is: comparing the 3D model file with at least one standard 3D model file stored in a database, so as to compare at least a part of the point cloud included in the 3D model file It is divided into a plurality of predetermined parts, and a part of the plurality of point coordinates included in each predetermined part is marked as a part feature point (feature point).

In practical applications, the database may store, for example, standard three-dimensional three-dimensional model files corresponding to races in different regions such as Asians, Americans, Europeans, etc., and in the feature calibration step S16 , the processing device may be, for example, based on the ethnic data contained in the user demand information, etc., to determine which standard three-dimensional three-dimensional model file needs to be extracted from the database for comparison. Of course, the standard three-dimensional three-dimensional model files stored in the database can be further distinguished by region, height, weight, body fat, gender, etc. For example, a standard 3D model file in the database can be marked as an Asian male, with a weight of 70-75 kg, a body fat of 20-25%, and a height of 175-180 cm, and when the processing device According to the user demand information, when it is determined that the user is an Asian male, and the weight, body fat and height are all within the marked range of the standard three-dimensional model file, the processing device will extract the standard three-dimensional model from the database. Compare the three-dimensional model files. In different embodiments, in the feature calibration step S15, the processing device may also use a trained machine learning model to determine which standard three-dimensional model file is to be compared with the current three-dimensional three-dimensional model file. Compare the three-dimensional model files.

As shown in FIG. 5 , the displaying step S17 is: inputting the 3D stereoscopic model file into a display device A6, so as to display a 3D stereoscopic image A62 on a display screen A61 of the display device A6, and the 3D stereoscopic image A62 includes a plurality of Predetermined area, a plurality of predetermined areas correspond to a plurality of predetermined parts of the three-dimensional model file, each predetermined area contains a plurality of image points (black dots located in the outline A623 in FIG. 5), and a plurality of images in each predetermined area The points correspond to a plurality of point coordinates contained in one of the predetermined parts of the three-dimensional solid model. Specifically, each predetermined area corresponds to each part of the human body, such as head, chest, waist, buttocks, left thigh, right thigh, left calf, right calf, left upper arm, left forearm, right upper arm, right forearm, left Palm, right palm, left foot, right foot, etc.

A part of the image points in each predetermined area is presented in a first pattern, and the rest of the image points are presented in a second pattern; wherein, each image point presented in the second pattern is defined as a characteristic image point, and each image point is presented in a second pattern. The feature image point corresponds to one of the feature points in the three-dimensional model file. Specifically, the user can view the appearance of the 3D stereoscopic image through a plurality of image points presented in the first format, and these characteristic image points are used as reference coordinate points for subsequent size measurement.

The display device A6 is electrically connected to an input device (not shown in the figure), and the input device can modify at least one image point presented in the first style to be presented in the second style according to the user's operation, or, at least An image point rendered in the second style is modified to be rendered in the first style. When the user modifies the style of any one of the image points through the input device, the input device will generate a corresponding modification file, and the input device will transmit the modification file to the processing device, and the processing device can correspond to the modification file according to the modification file. Modify the point coordinates and part feature points in the 3D solid model file. The input device is, for example, a keyboard, a mouse, a touch screen, a handwriting pad, etc., which are not limited herein. Wherein, regarding the specific presentation manner of the first style and the second style, as long as the relevant personnel can clearly distinguish the difference between the two on the display screen A61, no limitation is imposed here.

In practical applications, in the displaying step S17, the relevant personnel can view the 3D stereoscopic image A62 on a display screen A61 of the display device A6. In addition to including a plurality of image points, the 3D stereoscopic image A62 may also include At least one contour line A623 and a plurality of regional contour lines A624, the contour line A623 mainly presents the appearance of the three-dimensional stereoscopic image A62, and the area surrounded by each regional contour line A624 corresponds to one of the predetermined parts of the aforementioned three-dimensional three-dimensional model file, In this way, the relevant personnel can judge whether the predetermined parts calibrated by the processing device in the three-dimensional image model correctly correspond to the human body by viewing the contour lines A624 of the regions of the three-dimensional three-dimensional image A62 on the display screen A61. Various parts (such as shoulders, chest, waist, buttocks, thighs, calves, arms, elbows, etc.).

The judging step S18 is: judging whether any of the image points included in the 3D stereoscopic image has been modified; if no pattern of the image points has been modified, execute the aforementioned manufacturing step S2; When the pattern of the image point is modified, the manufacturing step S2A is executed: firstly, the processing device is made to make at least one of the point coordinates in the three-dimensional three-dimensional model file no longer be marked as a part feature point according to the modification file generated by the input device, and Mark one of the point coordinates as the part feature point; then make the processing device search for a plurality of corresponding part feature points in the three-dimensional model file according to the affected part data, and use the found multiple part feature points to calculate The three-dimensional model file corresponds to the size of the affected part at the location of the affected part data.

Specifically, the modification file will record relevant data (such as coordinates, original style, etc.) of the image points of the modified style, and the processing device can find the corresponding point coordinates in the three-dimensional model file according to the modification file , and mark it as a part feature point, or the processing device can find the corresponding part feature point in the three-dimensional model file according to the modification file, and make the part feature point no longer marked as a part feature point.

As shown in FIG. 5 and FIG. 6 , FIG. 6 shows a partial cross-sectional perspective view of the three-dimensional image A62 at the waist. For example, if the relevant person wants to make a pressure garment for the user to wear on the waist, in the feature calibration step S16, the processing device will use a comparison method to determine which part of the three-dimensional model file corresponds to The waist of the human body is marked, and the part point coordinates corresponding to this part are marked as part feature points. Then, when the relevant person clicks on the waist of the 3D stereoscopic image, the display screen A61 of the display device A6 may be, for example, converted from FIG. 5 to FIG. 6 , and the relevant person will be able to see the 3D stereoscopic image A62 in the display screen A61 , the position corresponding to the waist of the human body is framed by the area outline A624, and the area outline A624 includes a plurality of solid dots (the first pattern) image points A621 and a plurality of hollow dots (the first pattern) around the area outline A624. Two patterns) of the feature image point A622. Wherein, each solid circle point and each space circle point respectively correspond to one of the point coordinates in the three-dimensional three-dimensional model file, and the area contour line A624 may be the waist contour line simulated by the processing device, and this area The contour line A624 may only be used to allow the user to see the contour of the waist more clearly.

In the manufacturing step S2, when the processing device calculates the waist circumference of the user, the processing device will make a plurality of feature image points A622 located at the waist position in the three-dimensional three-dimensional image (that is, corresponding to the predetermined part of the waist in the three-dimensional three-dimensional model file). A plurality of part feature points) are connected into a line segment, and then the length of the line segment is calculated to calculate the waist circumference of the user. It should be noted that, the processing device may use the three-dimensional model file and the feature points of the part to calculate the size of the part in various mathematical ways, which is not limited here.

Following the above example, in the display step S16, the relevant personnel can see the image points A622 marked as part feature points by the processing device and presented in the second style in the display screen A61. At this time, the relevant personnel can At least one image point A622 is selectively modified through the input device to change the reference when the processing device subsequently calculates the size of the affected part, so that the processing device can calculate the waist circumference more accurately.

In a different embodiment, in the feature calibration step S16, the processing device may also load the three-dimensional three-dimensional model file A11 into a machine learning model, so as to use the machine learning model to convert the points included in the three-dimensional three-dimensional model file At least a part of a cloud (point cloud) is divided into a plurality of predetermined parts, and a part of a plurality of point coordinates included in each predetermined part is marked as a part feature point. In practical applications, in order to enable the machine learning model to more accurately demarcate various parts of the human body corresponding to the three-dimensional model file A11, the processing device may store multiple machine learning models, and the processing device may be based on user requirements. Race, height, weight, gender, body fat, etc. in the information, select the corresponding machine learning model to calibrate the current three-dimensional model file. That is to say, the multiple machine learning models stored in the processing device are trained using different training files. For example, one of the machine learning models may be trained using a three-dimensional model file of an Asian male, and one of the The machine learning model can be trained by using the three-dimensional model files of American women, and so on. In this way, the machine learning model can more accurately calibrate the current three-dimensional model files corresponding to various parts of the human body.

According to the above, through the design of the feature calibration step S16 and the display step S17 , the production equipment can more accurately manufacture a pressure fabric that conforms to the size of the part where the user wants to wear the pressure fabric.

Please refer to FIG. 7 , which is a schematic flow chart of one embodiment of the manufacturing method of the pressure fabric of the present invention. The biggest difference between this embodiment and the previous embodiment is that between the fusion step S15 and the manufacturing step S2, the processing device further divides at least a part of the point cloud included in the three-dimensional model file into a plurality of predetermined parts , and the processing device also performs the following steps:

A display step S16X: Input the three-dimensional model file into a display device, so as to display a corresponding three-dimensional three-dimensional image in the display device. The three-dimensional three-dimensional image includes a plurality of predetermined areas, and the plurality of predetermined areas correspond to the A plurality of predetermined parts, each predetermined area contains a plurality of image points, and the plurality of image points in each predetermined area correspond to a plurality of point coordinates contained in one of the predetermined parts of the three-dimensional model; each image point is a first image point The display device is electrically connected to an input device;

A calibration step S17X: using the input device to modify at least one image point presented in the first style to be presented in the second style according to the user's operation, and using the input device to generate a corresponding modification file;

A modification step S18X: causing the processing device to demarcate the coordinates of each point corresponding to the three-dimensional model file as a feature point of a part according to the modification file;

Another difference between this embodiment and the previous embodiment is that after step S18X is modified, the executed manufacturing step S2B is not exactly the same as the aforementioned manufacturing step S2. One of the corresponding predetermined parts in the three-dimensional model file is used to calculate the size of the affected part by using the part feature points included in the predetermined part.

In the foregoing embodiment, the processing device automatically calibrates the feature points of the parts in the three-dimensional three-dimensional model by comparing the standard three-dimensional three-dimensional model in the database or using the machine learning model, but in this embodiment, the processing device is The part feature points in the 3D model will not be automatically calibrated, and the relevant personnel will only see multiple image points presented in the first style on the display screen. The relevant personnel must use the input device to select multiple image points by themselves. image points, so that the image points are presented in a second pattern as the feature points of the part.

According to the above, through the design of the calibration step S16 and the modification step S17, the relevant personnel can manufacture a pressure fabric that better meets the needs of the user. That is, the calibration step S16 and the modification step S17 in this embodiment can make Relevant personnel manufacture customized pressure fabrics according to the actual location of the affected part of the user. For example, the part where the patient wants to wear the pressure fabric may not be a common part. In this case, the relevant personnel can use the calibration step S16 and the modification step S17 in this embodiment to select the part according to the part required by the patient. Feature points are selected to create pressure fabrics that fully meet the patient's needs.

Please refer to FIG. 9 to FIG. 11 together. FIG. 9 is a schematic flowchart of one embodiment of the manufacturing method of the pressure fabric according to the present invention. Shown as a top view of the carrier provided with the reference body. The biggest difference between this embodiment and the previous embodiment is that before using the scanning system to perform the scanning step S1, it also includes performing a camera and carrier device calibration step for each depth camera, which includes:

A reference body placement step S011: a reference body D is arranged on the bearing surface A21, and a reference surface D1 of the reference body is perpendicular to the bearing surface A21, and the reference surface is parallel to a central axis C passing through the bearing surface A21; Wherein, when the processing device controls the rotation mechanism, the plurality of depth cameras and the carrying device rotate relative to each other with the center axis C as the center, and the intersection of the center axis C and the carrying surface A21 is defined as a world coordinate origin OP ([O _X , O _y ,O _z ] ^T );

； A first calibration plane equation calculation step S012 : control the depth camera A4 to photograph the reference surface D1, so that the depth camera A4 generates a first calibration cloud point information, and uses the first calibration cloud point information to calculate a first calibration plane equation ( A _i X+B _i Y+C _i Z=0) and a first plane normal vector ([A _i ,B _i ,C _i ] ^T ); in this step, the reference plane is defined as a first reference plane, the first plane A calibration plane equation is used to describe the first reference plane in space, and the first plane normal vector is the normal vector of the first reference plane in space

;

； Execute n times a second correction plane equation calculation step S013, n is a positive integer greater than or equal to 3, and n is greater than M, the second correction plane equation calculation step S03 includes: firstly controlling the rotation mechanism to make the depth camera A4 and the The carrying devices A2 are rotated relative to each other by a correction angle; the depth camera A4 is then controlled to photograph the reference surface D1 of the reference body D to generate a second correction cloud point information, and then a second correction plane equation is calculated using the second correction cloud point information (A _j X+B _j Y+C _j Z=0) and a second plane normal vector ([A _j , B _j , C _j ] ^T ); in this step, the reference plane is defined as a second reference plane, The second calibration plane equation is used to describe the second reference plane in space, and the second plane normal vector is the normal vector of the second reference plane in space

;

； An average normal vector calculation step S014: using the first plane normal vector and the n second plane normal vectors to calculate an average normal vector ([n _X , _ny , n _z ] ^T ) of the bearing surface A21; the average method The vector is the direction vector of the central axis C

;

A bearing surface plane equation calculation step S015: define the origin of the world coordinate ([O _X , O _y , O _z ] ^T ) as the origin of the world coordinate system, and use the average normal vector to define a bearing surface plane equation (n _x X+n _y Y+n _z Z=0);

A first intersection line equation calculation step S016: using the first calibration plane equation (A _i X+B _i Y+C _i Z=0) and the bearing surface plane equation (n _x X+n _y Y+n _z Z=0 ), calculate a first intersection line equation; the first intersection line equation is used to describe a first intersection line CL1 where the first reference surface and the bearing surface intersect;

A second intersection line equation calculation step S017: using the second plane equation (A _j X+B _j Y+C _j Z=0) and the bearing surface plane equation (n _x X+n _y Y+n _z Z=0) , calculate a second intersection line equation, the second intersection line equation is used to describe a second intersection line CL2 where the second reference surface and the bearing surface intersect;

+

]； A rigid transformation matrix calculation step S018: use the vertical distance d _i between the first intersecting line CL1 and the world coordinate origin ([O _X , O _y , O _z ] ^T ), and the second intersecting line CL2 and the world coordinate origin ( [O _X , O _y , O _z ] ^T ) with the same relationship as the vertical distance d _j ), a coordinate conversion relational expression including a predetermined angle θ is obtained by calculation; wherein, in the coordinate conversion program of the camera and the carrying device, the processing device is The first model file is converted into the first conversion model file by using the coordinate conversion relational expression, and the processing device uses the coordinate conversion relational expression to convert the second model file into the second conversion model file; the coordinate conversion relational expression is:

+

];

為

，

是第一模型檔或第二模型檔所包含的點座標矩陣；

是世界座標原點。 _Prot is the point coordinate matrix contained in the first conversion model file or the second conversion model file; θ is the predetermined angle;

for

,

is the point coordinate matrix contained in the first model file or the second model file;

is the origin of the world coordinates.

In the average normal vector calculation step S014, the average normal vector ([n _X , _ny ,n _z ] ^T ) can be calculated by using the following formula:

In the above formula, n represents the number of times the second correction plane equation calculation step S013 is performed. In a preferred embodiment, n may be more than 5 times. Of course, the selection of the times of n and the correction angle must allow the depth camera to capture the same reference surface D1 of the reference body D every time.

In the rigid transformation matrix calculation step S018, the vertical distance d _i between the first intersecting line CL1 and the world coordinate origin ([O _X , O _y , O _z ] ^T ), and the second intersecting line CL2 and the world coordinate origin The same relationship of the vertical distance d _j of the point ([O _X ,O _y ,O _z ] ^T ) can be expressed mathematically as:

The above mathematical formula can be represented in the following matrix form:

是與A _i、A _j、B _i、B _j、C _i、C _j有關，

則是與D _i、D _j有關，且上述數學式可以利用電腦相關應用程式(例如Matlab ^R)，以奇異值分解(singular value decomposition, SVD)方法解出世界座標原點[O _X,O _y,O _z]。 in the above formula

is related to A _i , A _j , B _i , B _j , C _i , C _j ,

It is related to D _i , D _j , and the above mathematical formula can use computer related applications (such as Matlab ^R ) to solve the world coordinate origin [O _X , O _y by singular value decomposition (SVD) method , _Oz ].

It should be emphasized that, the manufacturing method of the pressure fabric of the present invention uses the above-mentioned method to perform the registration of the depth camera and the carrying device, instead of using the common iterative closest algorithm. points, ICP) for coordinate calibration, and the point cloud included in the first model file and the point cloud included in the second model file generated by each depth camera in the pressure fabric manufacturing method of the present invention (point cloud) As long as the overlap rate of the two is greater than 60%, a good fusion effect can be achieved. On the other hand, if the traditional near point algorithm (ICP) is used for coordinate calibration, the point cloud included in the first model file and the point cloud included in the second model file generated by each depth camera The overlap rate of cloud) must be greater than 80% to achieve a good fusion effect. In addition, if the traditional near point algorithm (ICP) is used for coordinate calibration, because of its high requirements on the overlap rate of point clouds, the depth camera and the carrying device can only be rotated by a small angle each time. Relatively long scanner room.

In a preferred embodiment, each time the second model file generating step S12 is executed, when each depth camera and the carrying device are rotated relative to each other by a predetermined angle, an angle detector is also controlled to detect Measure a real rotation angle of the bearing device relative to the rotation of the plurality of depth cameras; if the difference between the real rotation angle and the predetermined angle is greater than a predetermined range, the processing device brings the real rotation angle into the above-mentioned coordinate conversion relationship to convert the first A model file and a second model file are respectively converted into a first conversion model file and a second model conversion file.

Continuing from the above, specifically, it is assumed that the processing device controls the rotation mechanism to rotate 45 degrees (that is, the predetermined angle), but the actual rotation angle actually detected by the angle detector is 35 degrees. At this time, if the processing device uses θ= 45∘ Bring in the aforesaid coordinate conversion relationship, and convert the first model file and the second model file into the first conversion model file and the second model conversion file respectively. In this way, the correct 3D model file may not be merged in the end. Therefore, in order to avoid such a situation, the processing device can take θ=35∘ into the above-mentioned coordinate conversion relationship, thereby avoiding the problem that the correct three-dimensional model file cannot be merged in the end.

Another difference between this embodiment and the previous embodiment is that the biggest difference between this embodiment and the previous embodiment is that: before using the scanning system to perform the scanning step S1, it also includes performing a camera and camera calibration on all the depth cameras Step S02 , which includes: using Zhang Youzheng calibration method to perform coordinate calibration on two depth cameras adjacent to each other, so as to obtain the camera and a rigid transformation matrix required in the camera coordinate transformation program.

According to the above, through the camera and the carrier device calibration step S01 and the camera and the camera calibration step S02, the coordinates of the plurality of depth cameras and the carrier device are calibrated, compared to the traditional near point algorithm (ICP). It has the advantages of fast scanning speed and relatively fast registration speed.

The above descriptions are only preferred feasible embodiments of the present invention, which do not limit the scope of the present invention. Therefore, any equivalent technical changes made by using the contents of the description and drawings of the present invention are included in the protection scope of the present invention. .

A: Scanning system A1: Processing device A11: Three-dimensional model file A12: First conversion model file A13: Second conversion model file A14: First world coordinate model file A15: Second world coordinate model file A2: Carrying device A21: Bearing surface A3: Support structure A4: Depth camera A41: First model file A42: Second model file A5: Fixed structure A6: Display device A61: Display screen A62: 3D stereo image A621: Image point A622: Image point A623: Outline Line A624: Area outline C: Central axis CL1: First intersection line CL2: Second intersection line D: Reference body D1: Reference plane d _i : Vertical distance d _j : Vertical distance OA: Optical axis OP: World coordinate origin P1: included angle L: horizontal distance

FIG. 1 is a schematic diagram of a scanning system suitable for applying the manufacturing method of the pressure fabric of the present invention.

FIG. 2 is a schematic flow chart of one embodiment of the manufacturing method of the pressure fabric of the present invention.

3 is a schematic top view of a scanning system suitable for applying the manufacturing method of the pressure fabric of the present invention.

FIG. 4 is a schematic flow chart of one embodiment of the manufacturing method of the pressure fabric of the present invention.

FIG. 5 is a schematic diagram of a display device displaying a 3D stereoscopic image corresponding to a 3D stereoscopic model file.

FIG. 6 is a schematic perspective view of a partial cross-section of a 3D stereoscopic image at the waist.

FIG. 7 is a schematic flow chart showing one embodiment of the manufacturing method of the pressure fabric of the present invention.

FIG. 8 is a schematic flow chart showing one embodiment of the manufacturing method of the pressure fabric of the present invention.

FIG. 9 is a schematic diagram illustrating the steps of performing camera and carrier calibration on the depth camera.

Figure 10 shows a top view of a carrier device provided with a reference body.

S1, S11~S15, S2: process steps

Claims (9)

A method for manufacturing a pressure fabric, comprising the following steps: using a scanning system to perform a scanning step to perform a three-dimensional scanning operation on a user; the scanning system includes a processing device, a carrying device, a plurality of supporting structures, A plurality of depth cameras and a rotating mechanism, the processing device is electrically connected to the rotating mechanism, and the processing device can control the rotating mechanism, so that the plurality of the supporting structures and the carrying device are connected to each other Relatively rotated, the carrying device has a carrying surface, the carrying surface is used for the user to stand, a plurality of the supporting structures are arranged around the carrying device, and each of the supporting structures is provided with one of the depth cameras , the horizontal distance between each of the depth cameras and the carrying device is the same; the scanning system can perform the scanning step when the user is standing on the carrying surface, and the scanning step includes: a The first model file generation step: controlling each of the depth cameras to photograph the user standing on the bearing surface, so that each of the depth cameras generates a first model file; execute a second model file The generating step is M times, where M is a positive integer greater than or equal to 1, and the second model file generating step includes: firstly controlling the rotating mechanism to make each of the depth cameras and the carrying device rotate relative to each other by a predetermined amount angle, and then control each of the depth cameras to photograph the user, so that each of the depth cameras generates a second model file; a camera and a carrier device coordinate conversion step: using the processing device to convert each The first model file of each of the depth cameras, and a camera and carrying device coordinate conversion program corresponding to each of the depth cameras is performed to convert the first model file generated by each of the depth cameras. and the second model file are respectively converted into a first conversion model file and a second conversion model file; a world coordinate conversion step: using the processing device to convert the first conversion model of each of the depth cameras file and the second conversion model file, and perform a camera and camera coordinate conversion program to convert the first conversion model file and the second conversion model file of each depth camera into a first conversion model file respectively. A world coordinate model file and a second world coordinate model file; a fusion step: fuse all the first world coordinate model files and all the second world coordinate model files into a three-dimensional three-dimensional model file; A manufacturing step: calculating an affected part size according to the three-dimensional model file and an affected part data of a user demand information, and then controlling a production equipment according to the affected part size and a pressure data of the user demand information A pressure fabric is produced; wherein, when the processing device controls the rotating mechanism, the carrying device will be driven by the rotating mechanism to rotate relative to the plurality of depth cameras; when the scanning system is used to perform all Before the scanning step, it also includes performing a camera and carrying device calibration step for each of the depth cameras, which includes: a reference body placing step: setting a reference body on the carrying surface, and making the reference body body A reference surface is perpendicular to the bearing surface, and the reference surface is parallel to a central axis passing through the bearing surface; wherein, when the processing device controls the rotation mechanism, a plurality of the cameras and the bearing surface The devices are rotated relative to each other centered on the central axis, and the intersection of the central axis and the bearing surface is defined as a world coordinate origin; a first correction plane equation calculation step: controlling the camera to photograph the reference surface , so that the camera generates a first calibration cloud point information, and uses the first calibration cloud point information to calculate a first calibration plane equation and a first plane normal vector; in this step, the reference plane is defined is a first reference plane, the first calibration plane equation is used to describe the first reference plane in space, and the first plane normal vector is the normal vector of the first reference plane in space; execute A second calibration plane equation calculation step n times, n is a positive integer greater than or equal to 3, and n is greater than M; controlling the rotation mechanism, so that the depth camera and the carrying device are rotated relative to each other by a correction angle; Controlling the depth camera to photograph the reference surface of the reference body to generate a second calibration cloud point information, and using the second calibration cloud point information to calculate a second calibration plane equation and a second plane method vector; in this step, the reference plane is defined as a second reference plane, the second calibration plane equation is used to describe the second reference plane in space, and the normal vector of the second plane is the the normal vector of the second reference surface; an average normal vector calculation step: using the first plane normal vector and the P second plane normal vectors to calculate an average normal vector of the bearing surface; the average The normal vector is the direction vector of the central axis; a bearing surface plane equation calculation step: define the origin of the world coordinate as the origin of the world coordinate system, and use the average normal vector to define a bearing surface plane equation ; a first intersection line equation calculation step: using the first plane equation and the bearing surface plane equation to calculate a first intersection line equation; the first intersection line equation is used to describe the first reference surface and the a first intersection line where the bearing surface intersects; a second intersection line equation calculation step: using the second plane equation and the bearing surface plane equation to calculate a second intersection line equation, the second intersection line The equation is used to describe a second intersection line that intersects the second reference surface and the bearing surface; a rigid transformation matrix calculation step: using the vertical distance between the first intersection line and the origin of the world coordinate, and the The relationship between the second intersection line and the vertical distance of the origin of the world coordinate is the same, and a coordinate conversion relational expression including the predetermined angle is calculated; wherein, in the coordinate conversion program of the camera and the carrying device, the The processing device converts the first model file into the first conversion model file by using the coordinate conversion relationship, and the processing device converts the second model file using the coordinate conversion relationship is the second conversion model file; the coordinate conversion relationship is:

(1-cos θ )+[ O _x , O _y , O _z ]; the predetermined angle is θ,

for

-

,

is the point coordinate matrix contained in the first model file or the second model file, and P _rot is the point coordinate matrix contained in the first conversion model file or the second conversion model file; wherein, when using Before the scanning system performs the scanning step, it also includes performing a camera and camera calibration step on all the depth cameras, which includes: using the Zhang Youzheng calibration method to perform coordinate calibration on two depth cameras adjacent to each other to obtain The rigid transformation matrix required by each of the depth cameras in the camera and camera coordinate transformation procedures. The manufacturing method of a pressure fabric as claimed in claim 1, wherein each of the The overlap ratio between the point cloud (point cloud) included in the first model file and the point cloud (point cloud) included in the second model file generated by the depth camera is greater than 60%. The method for manufacturing a pressure fabric according to claim 1, wherein the two depth cameras adjacent to each other are defined as a first depth camera and a second depth camera, respectively; The overlap ratio between the point cloud (point cloud) included in the first model file and the point cloud (point cloud) included in the first model file generated by the second depth camera is greater than 60%; a point cloud (point cloud) included in the second model file generated by the first depth camera, and a point cloud (point cloud) included in the first model file generated by the second depth camera The overlap ratio of the two is greater than 60%; the point cloud included in the first model file and the point cloud included in the second model file generated by each of the depth cameras are two The overlap rate of the first model file is greater than 60%; the second model file generation step is performed more than 2 times, and the point clouds respectively included in the two second model files generated successively, the two The overlap rate is greater than 60%. The manufacturing method of a pressure fabric according to claim 1, wherein after the fusion step, a feature calibration step is further included: comparing the three-dimensional three-dimensional model file with at least one standard three-dimensional three-dimensional model stored in a database The files are compared, so as to distinguish at least a part of the point cloud (point cloud) included in the three-dimensional model file into a plurality of predetermined parts, and to calibrate a part of the plurality of point coordinates included in each of the predetermined parts is a part feature point, and the manufacturing step includes: searching for a plurality of the corresponding part feature points in the three-dimensional three-dimensional model file according to the affected part data; using the found multiple part feature points to calculate The size of the affected part corresponding to the position of the data of the affected part is obtained from the three-dimensional three-dimensional model file. The manufacturing method of a pressure fabric according to claim 4, wherein after the feature calibration step and before calculating the size of the affected part, it further includes a display step, which includes: inputting the three-dimensional model file into a display a device for displaying a corresponding three-dimensional stereoscopic image on the display device, the three-dimensional stereoscopic image includes a plurality of predetermined areas, and the plurality of the predetermined areas correspond to a plurality of the predetermined areas of the three-dimensional stereoscopic model file part, each of the predetermined areas includes a plurality of image points, and the plurality of the image points in each of the predetermined areas correspond to the plurality of point coordinates included in one of the predetermined parts of the three-dimensional model ; A part of the image points in each of the predetermined areas is presented in a first pattern, and the rest of the image points are presented in a second pattern; each of the images presented in the second pattern A point is defined as a feature image point, and each feature image point corresponds to one of the part feature points in the three-dimensional model file; wherein, the display device is electrically connected to an input device, and the input device can According to the operation of the user, at least one of the image points presented in the first style is modified to be presented in the second style, or, at least one of the image points presented in the second style is modified is presented in the first style; if the style of any one of the image points is modified, the input device will correspondingly generate a modified file, and the processing device will modify the three-dimensional file according to the modified file At least one of the part feature points in the solid model file. The manufacturing method of a pressure fabric according to claim 1, wherein after the fusion step, a feature calibration step is further included: loading the three-dimensional model file into a machine learning model to utilize the machine learning A model, dividing at least a part of the point cloud (point cloud) contained in the three-dimensional model file into a plurality of predetermined parts, and marking a part of the plurality of point coordinates contained in each of the predetermined parts as a part feature point, the manufacturing step The steps include: searching for a plurality of the corresponding part feature points in the three-dimensional three-dimensional model file according to the data of the affected part; The size of the affected part at the location of the affected part data. The manufacturing method of a pressure fabric according to claim 6, wherein after the feature calibration step and before calculating the size of the affected part, it further includes a display step, which includes: inputting the three-dimensional model file into a display a device for displaying a corresponding three-dimensional stereoscopic image on the display device, the three-dimensional stereoscopic image includes a plurality of predetermined areas, and the plurality of the predetermined areas correspond to a plurality of the predetermined areas of the three-dimensional stereoscopic model file Each of the predetermined areas includes a plurality of image points; a part of the image points in each of the predetermined areas is presented in a first pattern, and the rest of the image points are presented in a second pattern ; Each of the image points presented in the second style is defined as a characteristic image point, and each of the characteristic image points corresponds to one of the part characteristic points in the three-dimensional model file; wherein, the display The device is electrically connected to an input device, and the input device can modify at least one of the image points presented in the first style to be presented in the second style according to the operation of the user, or at least one The image points presented in the second style are modified to be presented in the first style; if the style of any one of the image points is modified, the input device will generate a corresponding modification file, and the The processing device will modify at least one of the part feature points in the three-dimensional model file according to the modification file. The manufacturing method of a pressure fabric according to claim 1, wherein at least a part of the point cloud (point cloud) included in the three-dimensional model file is divided into a plurality of predetermined parts, and in the fusion step and the manufacturing step between Including: a display step: inputting the three-dimensional model file into a display device, so as to display a corresponding three-dimensional three-dimensional image in the display device, the three-dimensional three-dimensional image includes a plurality of predetermined areas, a plurality of the The predetermined area corresponds to a plurality of the predetermined parts of the three-dimensional stereo model file, each of the predetermined areas includes a plurality of image points, and the plurality of the image points in each of the predetermined areas correspond to the three-dimensional stereo model. wherein a plurality of point coordinates included in the predetermined part; each of the image points is presented in a first pattern; wherein the display device is electrically connected to an input device; a calibration step: using the input device According to the operation of the user, at least one of the image points presented in the first style is modified to be presented in the second style, and a corresponding modification file is generated by using the input device; a modification step: causing the processing device to demarcate the coordinates of each of the points corresponding to the three-dimensional three-dimensional model file as a feature point of a part according to the modification file; wherein, in the manufacturing step, firstly according to the data of the affected part , find out one of the corresponding predetermined parts in the three-dimensional three-dimensional model file, and then use the part feature points included in the predetermined part to calculate the size of the affected part. The manufacturing method of the pressure fabric according to claim 8, wherein each time the second model file generating step is performed, each of the depth cameras and the carrying device are rotated relative to each other by the predetermined amount At the same time, an angle detector is also controlled to detect a real rotation angle of the bearing device relative to the plurality of depth cameras; if the difference between the real rotation angle and the predetermined angle is greater than one predetermined range, the handling device The setting is to bring the real rotation angle into the coordinate conversion relationship, so as to convert the first model file and the second model file into the first conversion model file and the second model conversion file respectively .

Images