TWI798999B - Device and method for buliding three-dimensional video - Google Patents

Abstract

A device and a method for building a three-dimensional video are provided. The device includes a plurality of video capture devices and a processing device. Each video capture device generates a video capture signal. The video capture signal includes image data, depth data, and time stamps with the image data. In a video capture stage, for each of the time stamps, the processing device calculates point cloud data and texture data based on the image data and the depth data of the video capture signal, and creates a three-dimensional model based on the point cloud data and the texture data. The processing device integrates the time stamps and the three-dimensional model corresponding to the time stamps into a video stream.

Description

Apparatus and method for constructing three-dimensional images

The invention relates to a device and method for constructing a three-dimensional image.

In the application scenarios of augmented reality (Augmented Reality; AR) technology or virtual reality (Virtual Reality; VR) technology, it is necessary to play 3D images or dynamic 3D models to allow users to better integrate into the virtual scene or facilitate for interacting with virtual objects.

However, if you want to generate dynamic 3D images (or called "animation"), you need to build a static 3D model first, which is time-consuming and labor-intensive. Moreover, if the static 3D model is to be converted into a dynamic 3D model, the skeleton of the 3D model and the texture map of the surface need to be established separately, thus requiring a lot of design manpower to continue the subsequent animation production. Therefore, if it is possible to reduce the difficulty of creating a 3D model and make it easier for users to build a 3D model, it will be more convenient to develop applications such as AR technology or VR technology.

The present invention provides a device and method for building a 3D image. A 3D image with a 3D model is generated from a real scene image. In addition to recording the movement of the 3D model on the time axis, it can also save the time for building the 3D model.

The invention proposes a device for constructing a three-dimensional image. The device includes a plurality of image capturing devices and a processing device. Each image capture device is used to generate an image capture signal. The image capture signal includes image data, depth data and time stamps, and the image capture signals generated by each image capture device are located on the same timeline through the time stamps. The processing device is used for receiving the respective image capture signals of the image capture devices. In the image capture stage, the image capture device captures the image capture signal, and the processing device receives the image capture signal from the image capture device. For each time stamp, the processing device calculates point cloud data and texture data according to the image data and the depth data in the image capture signal, and establishes a 3D model based on the point cloud data and the texture data . The processing device integrates the time stamp and the 3D model corresponding to the time stamp into an image stream.

The invention proposes a method for constructing a three-dimensional image. The method includes: configuring a plurality of image capture devices, each image capture device is used to generate an image capture signal, the image capture signal includes image data, depth data and time stamp, and each image capture The image capture signals generated by the device are located on the same timeline by the time stamp; in the image capture phase, the image capture signals captured from the plurality of image capture devices are obtained; for For each time stamp, calculate a point cloud data and a texture data according to the image data and the depth data in the image capture signal, and build a 3D model according to the point cloud data and the texture data; and , integrating the time stamp and the 3D model corresponding to the time stamp into an image stream.

Based on the above, the present invention provides a device and method for constructing a 3D image. Real-scene images are captured by multiple image capture devices, and vector data such as point clouds and textures are estimated based on the real-scene images, thereby establishing a time axis. 3D model at each time point above, and generate a 3D image with the 3D model. Therefore, the present invention directly creates a 3D model and a 3D image with the aforementioned 3D model by capturing real-scene images, which can record the movement of the 3D model on the time axis and save the time for building the 3D model. Furthermore, since the 3D image has the dynamic action of the 3D model, it is more conducive to estimating or generating the skeleton information corresponding to the aforementioned 3D model, so as to realize the establishment of the 3D model and its skeleton by using the real scene image, and reduce the construction cost of the 3D model .

FIG. 1 is a block diagram of an apparatus 100 for constructing a 3D image according to an embodiment of the present invention. The device 100 for constructing a 3D image mainly includes a plurality of image capture devices 110 - 1 - 110 -n (n is a positive integer greater than or equal to 2) and a processing device 120 . Each image capture device 110-1~110-n is used for generating image capture signals S110-1~S110-n. The image capturing device 110-n of this embodiment is a depth camera as an example. The processing device 120 can be used for the respective image capture signals S110-1~S110-n of the image capture devices 110-1~110-n, and process these image capture signals S110-1~S110-n to generate An image stream 190 of the 3D model of the shaft.

The image capture devices 110-1~110-n include time synchronization modules 112-1~112-n and transceiver modules 114-1~114-n. The time synchronization modules 112-1~112-n are used to generate time stamps. The processing device 120 can know when each image data and depth data are captured through the time stamps in the image capture signals S110-1˜S110-n. The transceiver modules 114-1~114-n can transmit the image capture signals S110-1~S110-n generated by the image capture devices 110-1~110-n through the network, bus or corresponding communication protocols. to the processing device 120.

The processing device 120 can be used as a main console for controlling the aforementioned image capture devices 110-1~110-n, or a server located in the cloud. The processing device 120 of this embodiment may include a communication module 122 , an image processing module 124 and a central processing unit 126 . The communication module 122 communicates with the aforementioned transceiver modules 114-1~114-n and receives the aforementioned image capture signals S110-1~S110-n. The image processing module 124 and the CPU 126 can process the aforementioned image capture signals S110-1˜S110-n. Those who apply this embodiment can implement the aforementioned modules according to their needs (for example and not limited to, time synchronization modules 112-1~112-n, transceiver modules 114-1~114-n, communication modules 122 and image processing modules Group 124), for example, realize the corresponding functions of the aforementioned modules in the form of central processing unit, graphics processing unit, integrated circuit, firmware device, etc.

The image capture signals S110-1˜S110-n in this embodiment include not only image data and depth data corresponding to the aforementioned image data, but also time stamps. The time stamp is used to record the time point information of capturing the image data. The image capture signals S110-1-S110-n generated by each of the image capture devices 110-1-110-n are located on the same timeline with corresponding time stamps.

In detail, the image capture devices 110-1~110-n include time synchronization modules 112-1~112-n and transceiver modules 114-1~114-n. The time synchronization modules 112-1~112-n are used to generate time stamps. The processing device 120 can know when each image data and depth data are captured through the time stamps in the image capture signals S110-1˜S110-n. For example, in the initialization stage, the processing device 120 controls each image capture device 110-1~110-n to synchronize each time synchronization module 112-1~112-n, that is, to make these time synchronization modules The time stamps generated by 112-1~112-n are mutually synchronized. The transceiver modules 114-1~114-n of this embodiment can transmit the image capture signals S110-1~110-1~110-n generated by the image capture devices 110-1~110-n through a network, a bus or a corresponding communication protocol. S110-n is sent to the processing device 120 .

FIG. 2 is a flowchart of a method 200 for constructing a 3D image according to an embodiment of the present invention. The method 200 shown in FIG. 2 is implemented by using the device 100 for building a three-dimensional image shown in FIG. 1 , and each step in the method 200 is controlled and executed by the processing device 120 . The method 200 includes steps S205 to S230. In step S205, in the initialization stage of the device 100 for constructing a three-dimensional image in FIG. 1, the processing device 120 in FIG. 1 obtains image capture signals S110-1~ S110-n, to obtain the initialized background data and the corresponding initialized depth data. Thereby, in the subsequent image capture stage (step S210 to step S230 ), the processing device 120 in FIG. 1 can capture the background information in the scene before the object is placed. The initialization stage of this embodiment may also include: the synchronization of the time synchronization modules 112-1~112-n in FIG. 1, the relative position correction of the image capture devices 110-1~110-n...etc. The subsequent examples will be described in detail.

In step S210 of the image capture stage, the image capture devices 110-1~110-n capture image capture signals S110-1~S110-n, and the processing device 120 in FIG. ~110-n receives image capturing signals S110-1~S110-n.

In the step S220 of the image capture phase, for each time stamp, the processing device 120 in FIG. Calculate point cloud data and texture data, and build a 3D model based on the aforementioned point cloud data and the aforementioned texture data. In the step S230 of the image capture phase, the processing device 120 in FIG. 1 integrates the time stamps and the 3D models corresponding to the time stamps into an image stream. Therefore, in the embodiment of the present invention, a 3D image with a 3D model is generated from a real-scene image, which not only can record the movement of the 3D model on the time axis, but also saves the construction time of the 3D model.

FIG. 3 is a schematic diagram of configuring multiple image capture devices 110-1~110-8 according to an embodiment of the present invention. 4 and 5 are schematic diagrams for presenting image data captured by the image capture device in FIG. 3 . The image capture device of this embodiment is viewed from multiple angles (for example, image capture devices 110-1 to 110-5 shown in part (A) of FIG. 3 ) and/or multiple heights (for example, in FIG. 3 The image capture devices 110 - 6 - 110 - 8 shown in part (B) capture images of the object 310 (shown as a human body), thereby generating image capture signals S110 - 1 - S110 - 8 in FIG. 1 . The XY plane shown in part (A) of FIG. 3 is used to show that the image capture devices 110 - 1 - 110 - 5 respectively capture images of the object 310 from different angles. The XZ plane shown in part (B) of FIG. 3 is used to show that the image capture devices 110-6-110-8 capture images of the object 310 from different heights. For example, the height of the image capture device 110-6 is higher than The image capture device 110-7, and the height of the image capture device 110-7 is higher than the image capture device 110-8. Those who apply this embodiment can set multiple image capture devices 110 - n at different angles relative to the object 310 , or at different heights, or at different angles and heights according to their requirements. If multiple image capture devices are installed at different angles and heights at the same time (for example, to realize part (A) and part (B) in Figure 3 at the same time), 5 image capture devices will be set at each height and a total of 15 An image capture device is used to realize the embodiment of the present invention.

In detail, the image capture devices 110-1~110-5 shown in part (A) of FIG. The angular difference between the devices is 72 degrees. Therefore, the image data of the image capture signals captured by adjacent image capture devices at the same time point may overlap or be similar to some extent. Fig. 4 is the image data 430-1, 430-2, 430-5 in the adjacent image capture devices 110-1, 110-2, 110-5 in part (A) of Fig. 3 and the splicing of these image data schematic diagram. Here, the image capture device 110-1 is taken as an example to illustrate how to use the image capture devices 110-2 and 110-5 adjacent to the image capture device 110-1 to generate a corresponding image belonging to the image capture device 110-1. Perspective image data captured by the perspective. Part of the image data 430-1 captured by the image capture device 110-1 overlaps with the image data captured by the image capture devices 110-2 and 110-5. Therefore, the image processing module 124 of the processing device 120 in FIG. 1 in this embodiment can use an image stitching algorithm to integrate these partially overlapping image data 430-1, 430-2, and 430-5 in a horizontally connected manner. The viewing angle image data 440-1 belonging to the corresponding viewing angle captured by the image capturing device 110-1 is formed. The aforementioned "image stitching algorithm" can cut and collage multiple images corresponding to different viewing angles to stitch them into a wide-view image. Please refer to the publication No. TWI672670 patent specification. Although the embodiment in FIG. 4 shows that the image data captured by the image capturing devices 110-1, 110-2, and 110-5 can be integrated into the viewing angle image data 440-1 corresponding to the viewing angle of the image capturing device 110-1. However, it can also be known by applying this embodiment that when the image data captured by other image capture devices (eg, image capture devices 110-3, 110-4) can also be combined with the image data captured by the image capture device 110-1 The image data captured overlaps with each other. The processing device 120 in FIG. The image capture device 110 - 1 corresponds to the viewing angle image data of the viewing angle. Moreover, the person applying this embodiment generates the angle-of-view image data corresponding to each of the image capture devices 110 - 1 - 110 - 5 through the aforementioned operations.

On the other hand, the image capture devices 110-6~110-8 shown in part (B) of FIG. 3 are arranged near the object 310 at different heights. Click the captured image to capture the image data of the signal to be slightly overlapped or similar. Figure 5 is the image data 530-6, 530-7, 530-8 in the adjacent image capture devices 110-6, 110-7, 110-8 in (A) of Figure 3 and the splicing of these image data schematic diagram. Because the image capture devices 110-6~110-8 all have the same viewing angle on the XY plane, the image data 530-7 captured by the image capture device 110-7 has a part of the area that is the same as that of the image capture device 110-6. , 110-8 The captured image data 530-6, 530-8 are partially overlapped. Therefore, the image processing module 124 of the processing device 120 in FIG. 1 of this embodiment can use an image stitching algorithm (please refer to the announcement No. TWI672670) to vertically align these partially overlapping 530-6, 530-7, and 530-8 The image data 540-2 of the corresponding viewing angles captured by the image capturing devices 110-6~110-8 are integrated in a concatenated manner.

In this embodiment, the processing device 120 in FIG. 1 can use the image data captured by the image capture devices 110-1~110-8 in FIG. 3 and the corresponding depth data to generate part of the point cloud data required by the 3D model. , and connect these point cloud data into a complete 3D model, and the 3D model is used to describe the object 310 . Alternatively, the processing device 120 in FIG. 1 can use the perspective image data corresponding to the perspectives of the image capture devices 110-1~110-5 (for example, the perspective image data 440-1 in FIG. 4, the perspective image in FIG. 5 Data 540 - 2 ) and corresponding depth data to generate part of the point cloud data required by the 3D model, and connect these point cloud data into a complete 3D model for describing the object 310 . Alternatively, the processing device 120 in FIG. 1 can integrate the viewing angle image data corresponding to the viewing angles of the image capturing devices 110-1~110-5 into a surrounding view image relative to the object 310, by combining the surrounding view image with the The corresponding depth data is used to generate the point cloud data required by the 3D model, thereby generating a complete 3D model for describing the object 310 . In other words, the person applying this embodiment can calculate or estimate part of the point cloud data describing the object 310 based on the image data and depth data captured by the image capture devices 110-1~110-8 according to their needs, so that The aforementioned point cloud data are integrated to generate a 3D model of the object 310 . Moreover, in addition to calculating the point cloud data, the processing device 120 in FIG. 1 of this embodiment will also calculate the position on the object 310 based on the image data and depth data captured by the image capture devices 110-1~110-8 respectively. Texture material, so as to generate corresponding texture data, so as to create a 3D model with both point cloud data and texture data.

FIG. 6 is a schematic diagram showing a 3D model of an object 610 according to the position configuration of the image capture devices 110 - 1 - 110 - 5 in part (A) of FIG. 3 according to an embodiment of the present invention. Part (A) in FIG. 6 is similar to the relative position configuration of the image capture devices 110-1~110-5 in part (A) in FIG. 3 . Parts (B) and (C) of FIG. 6 present image data 630 - 1 , 630 - 5 captured by the image capture devices 110 - 1 , 110 - 5 respectively as an example. Part (D) of FIG. 6 presents the 3D model obtained and integrated after calculation by the processing device 120 of FIG. 1 .

Fig. 7 is a schematic diagram of matching point cloud data with a human skeleton according to an embodiment of the present invention. After the processing device 120 in FIG. 1 calculates the 3D model integrated from the point cloud data 710 , it can generate the skeleton data 720 corresponding to the 3D model according to the point cloud data 710 . For the aforementioned skeleton matching technology, please refer to the paper "Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields, IEEE Transactions on Pattern Analysis and Machine Intelligence, Volume: 43, Issue: 1, Jan. 1 2021". The skeleton matching technology can use the point cloud data 710 to judge the position of the human skeleton, and since the processing device 120 of the present invention in FIG. Dynamic actions are more conducive to estimating or generating the skeleton information 720 corresponding to the aforementioned 3D model, and the skeleton information 720 will have better accuracy.

FIG. 8 is a detailed flowchart of a method 800 for constructing a 3D image according to an embodiment of the present invention. The method 800 in FIG. 8 is implemented by the device 100 for building a 3D image in FIG. 1 , and the method 800 in FIG. 8 is the details of each step in the method 200 for building a 3D image in FIG. 2 . Please refer to FIG. 8 and FIG. 2 , step S201 in FIG. 8 is an initialization phase, including steps S802 , S804 and S806 .

In step S802, the processing device 120 in FIG. 1 determines and corrects the image capture devices 110-1~S110-n through the image capture signals S110-1~S110-n respectively captured by the image capture devices 110-1~110-n. The relative position of 110-n. In one embodiment, the relative positions of the image capture devices 110 - 1 - 110 - n can be adjusted manually or electronically. The aforementioned electronic driving method is, for example, installing adjustable bases on these image capture devices 110-1~110-n, so that the processing device 120 in FIG. relative position.

In step S804, similar to step S205 in FIG. 2, the processing device 120 in FIG. 1 acquires the initialization background data and Its corresponding initialization depth data. In this way, in the subsequent image capture stage, the processing device 120 in FIG. 1 can capture the background information of the scene before the object is placed.

In step S806, the processing device 120 of FIG. 1 controls the image capture devices 110-1~110-n to synchronize the time synchronization modules 112-1~112-n, so that the time synchronization modules 112-n generate The timestamps are synchronized with each other. The time synchronization modules 112-1~112-n are used for generating time stamps in the image capture signals S112-1~S112-n.

After completing the initialization phase (step S201 ), the device 100 for constructing a 3D image in FIG. 1 can enter into an image capture phase. In step S210, in the image capture phase, the processing device 120 in FIG. 1 controls the image capture devices 110-1~110-n to capture image capture signals S110-1~S110-n, and transmits the signals from the image capture device 110- 1~110-n receive image capture signals S110-1~S110-n. Step S220 in FIG. 2 includes multiple steps S822 , S824 , S826 and S828 . In step S822, splicing the image data in the image capture signals S110-1~S110-n in the processing device 120 to generate viewing angle image data corresponding to the viewing angles of the image capturing devices 110-1~110-n, It is as described in the corresponding embodiments of the aforementioned FIGS. 3 to 5 . In step S824 , the aforementioned perspective image data are concatenated in the processing device 120 to form a surround view image, as described in the corresponding embodiments of FIGS. 3 to 5 . In step S826 , the processing device 120 in FIG. 1 generates point cloud data and texture data according to the aforementioned perspective image data and surrounding view image. In this embodiment, the point cloud data can be generated firstly according to the perspective image data and the surrounding view image, and then the aforementioned texture data can be generated by using the aforementioned point cloud data. In this embodiment, the aforementioned texture data can be generated by performing triangular meshing according to the perspective image data. In step S828, the processing device 120 in FIG. 1 creates a 3D model corresponding to each time stamp according to the point cloud data and texture information.

After step S220 is completed, it means that the 3D models corresponding to each time stamp have been established. Therefore, in step S230, the processing device 120 in FIG. 1 integrates these time stamps and the 3D models corresponding to these time stamps into an image stream. Furthermore, in step S840, the processing device 120 in FIG. 1 can also generate skeleton data corresponding to the 3D model according to the aforementioned point cloud data and texture data, as described in the embodiment of FIG. 7 above.

Therefore, the embodiment of the present invention will generate a 3D model corresponding to this timestamp at the first timestamp, and at the next timestamp (that is, the next timestamp relative to the first timestamp on the time axis, which may be The first timestamp passes through a specific time period (for example, the next timestamp after one second or a few tenths of a second) to generate a 3D model corresponding to this timestamp, thereby gradually completing each timestamp correspondence 3D model. Therefore, a corresponding 3D model will be generated at each time stamp, thereby having a dynamic 3D model in a time axis and realizing the function of recording 3D animation.

In summary, the embodiment of the present invention uses multiple image capture devices to capture real-scene images, and estimates vector data such as point clouds and textures based on the real-scene images, thereby establishing a 3D model at each time point on the time axis, Generate 3D images with 3D models. Therefore, the present invention directly creates a 3D model and a 3D image with the aforementioned 3D model by capturing real-scene images, which can record the movement of the 3D model on the time axis and save the time for building the 3D model. Furthermore, since the 3D image has the dynamic action of the 3D model, it is more conducive to estimating or generating the skeleton information corresponding to the aforementioned 3D model, so as to realize the establishment of the 3D model and its skeleton by using the real scene image, and reduce the construction cost of the 3D model .

100: Apparatus for constructing three-dimensional images 110-1~110-n: image capture device 112-1~112-n: Time synchronization module 114-1~114-n: transceiver module 120: processing device 122: Communication module 124: Image processing module 126: CPU 190: Video streaming 310: object 430-1~430-5, 530-6~530-8, 630-1, 630-5: image data 440-1, 540-2: perspective image data 710: Point cloud data 720: Skeleton data S110-1~S110-n: Image capture signal S201-S230, S802-S840: each step of the method for building a three-dimensional image

FIG. 1 is a block diagram of a device\for constructing a 3D image according to an embodiment of the present invention. FIG. 2 is a flowchart of a method for constructing a 3D image according to an embodiment of the invention. FIG. 3 is a schematic diagram of configuring multiple image capture devices according to an embodiment of the present invention. 4 and 5 are schematic diagrams for presenting image data captured by the image capture device in FIG. 3 . FIG. 6 is a schematic diagram showing a 3D model of an object according to the position configuration of the image capture device in part (A) of FIG. 3 according to an embodiment of the present invention. Fig. 7 is a schematic diagram of matching point cloud data with a human skeleton according to an embodiment of the present invention. FIG. 8 is a detailed flowchart of a method for constructing a 3D image according to an embodiment of the invention.

S205~S230: each step of the method for constructing a three-dimensional image

Claims (14)

A device for building a three-dimensional image, comprising: a plurality of image capture devices, each image capture device is used to generate an image capture signal, the image capture signal includes an image data, a depth data and a time stamp by which the image capture signals generated by each image capture device are located on the same timeline; and a processing device for receiving the respective an image capture signal, wherein, in an image capture stage, the plurality of image capture devices capture the image capture signal, the processing device receives the image capture signal from the image capture device, For each time stamp, the processing device calculates point cloud data and a texture data according to the image data and the depth data in the image capture signal, and according to the point cloud data and the texture data building a 3D model, the processing device integrates the time stamp and the 3D model corresponding to the time stamp into an image stream, wherein in an initialization phase, the processing device passes through the image capture device The image capture signals captured respectively are used to determine and correct the relative positions of the plurality of image capture devices. The device according to claim 1, wherein the processing device generates a skeleton data corresponding to the 3D model according to the point cloud data and the texture data, and the 3D model in the image stream has The corresponding skeleton data. The device according to claim 1, wherein the image capture device captures images of an object from multiple angles or multiple heights, thereby generating the image capture signal. The device as described in claim 1, wherein each image capture device includes a time synchronization module and a transceiver module, and the processing device controls these image capture devices to set the time during the initialization phase Synchronization of synchronization modules, wherein the time synchronization module is used to generate the time stamp, and the transceiver module transmits the information generated by the image capture device through a network, a bus or a communication protocol The image capture signal is sent to the processing device, and the processing device includes a communication module, an image processing module and a central processing unit, the communication module communicates with the transceiver module and receives the The image capture signal, the image processing module and the CPU process the image capture signal. The device according to claim 1, wherein in the initialization stage, the processing device obtains the respective image capture signals of the image capture devices to obtain an initialized background data and a corresponding initialized depth data , wherein the processing device calculates the image data, the point cloud data and the texture data corresponding to the depth data based on the initialized background data and the corresponding initialized depth data. The device according to claim 1, wherein the processing device concatenates the image data in the image capture signal to generate a viewing angle image data corresponding to the viewing angle of the image capturing device, concatenating the The perspective image data becomes a surround view image, and the point cloud data and the texture map are generated according to the perspective view image data and the surround view image data, so as to establish the 3D model corresponding to each time stamp according to the point cloud data and the texture information. The device according to claim 1, wherein the processing device concatenates the image data in the image capture signal to generate a view image data corresponding to the view angle of the image capture signal, according to the The perspective image data is triangulated to generate texture data, the point cloud data is generated according to the image data in the image capture signal, and all time stamps corresponding to each time stamp are established according to the point cloud data. 3D model described. A method for building a three-dimensional image, comprising: configuring a plurality of image capture devices, each image capture device is used to generate an image capture signal, the image capture signal includes an image data, a depth data and a time stamp, and the image capture signals generated by each image capture device are located on the same time line by the time stamp; in an image capture stage, the signals from the plurality of image capture devices are obtained The captured image capture signal; for each time stamp, calculate a point cloud data and a texture data according to the image data and the depth data in the image capture signal, according to the point building a 3D model from the cloud data and the texture data; and integrating the 3D model corresponding to the time stamp into an image stream, wherein each image capture device includes a time synchronization module , and the method further includes: in an initialization phase, controlling the image capture devices to set the time synchronization model Group synchronization, wherein the time synchronization module is used to generate the time stamp. The method according to claim 8, further comprising: generating a skeleton data corresponding to the 3D model according to the point cloud data and the texture data, wherein the 3D model in the image stream has a corresponding The skeleton data of . The method according to claim 8, wherein the image capture device captures images of an object from multiple angles or multiple heights, thereby generating the image capture signal. The method according to claim 8, further comprising: in the initialization stage, judging and correcting the relative positions of the plurality of image capture devices through the image capture signals captured by each of the image capture devices . The method according to claim 8, further comprising: in the initialization stage, obtaining the respective image capture signals of the image capture devices to obtain an initialized background data and a corresponding initialized depth data, wherein calculating the image data and the point cloud data and the texture data corresponding to the depth data based on the initialized background data and the corresponding initialized depth data. The method according to claim 8, wherein the point cloud data and the texture map data are calculated according to the image data and the depth data in the image capture signal, and the point cloud data and the texture data are calculated according to the point cloud data and the depth data. The step of creating the 3D model from the texture data includes: splicing the image data in the image capture signal to generate a corresponding image A view image data of the view angle of the captured signal; concatenating the view image data to form a surround view image; generating the point cloud data and the texture map data according to the view view image data and the surround view image ; and establishing the 3D model corresponding to each time stamp according to the point cloud data and the texture information. The method according to claim 8, wherein the point cloud data and the texture map data are calculated according to the image data and the depth data in the image capture signal, and the point cloud data and the texture data are calculated according to the point cloud data and the depth data. The step of creating the 3D model from the texture data includes: splicing the image data in the image capture signal to generate a viewing angle image data corresponding to the viewing angle of the image capture signal, and according to the performing triangulation on the perspective image data to generate a texture data; and generating the point cloud data according to the image data in the image capture signal, and establishing each time stamp correspondence according to the point cloud data The 3D model of .

Images