A three-dimensional reconstruction method and system

The method employs projection features from a light source generator to construct a three-dimensional mesh model and perform texture mapping, addressing the challenge of feature point extraction in difficult scenes, thereby enabling efficient three-dimensional reconstruction.

CN114067061BActive Publication Date: 2024-12-06CHENGDU RAINPOO TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111451190.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-01
Publication Date
2024-12-06
Estimated Expiration
2041-12-01

AI Technical Summary

Technical Problem

Existing three-dimensional reconstruction methods fail to reconstruct scenes with difficult feature point extraction from two-dimensional images, such as pure color scenes or regularly repeating graphics, due to the inability to locate and describe feature points effectively.

Method used

A three-dimensional reconstruction method that uses a light source generator to project light in the scene, forming projection features, and employs these features to construct a three-dimensional mesh model, followed by texture mapping using a second image without projection features to complete the reconstruction.

Benefits of technology

Enables convenient and effective three-dimensional reconstruction of scenes with difficult feature point extraction by using projection features to construct a three-dimensional mesh model and perform texture mapping, ensuring accurate and efficient reconstruction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114067061B_ABST
    Figure CN114067061B_ABST
Patent Text Reader

Abstract

The present invention discloses a three-dimensional reconstruction method, comprising: obtaining a scene image with projection features as a first image, wherein the projection features are features formed by a light source generator projecting light in the scene; obtaining a scene image without projection features as a second image; using the first image to construct a three-dimensional mesh model; using the second image to perform texture mapping on the three-dimensional mesh model to complete three-dimensional reconstruction. The present invention also discloses a system for implementing the above three-dimensional reconstruction method. The present invention extracts feature points through projection features formed by light projected by a light source generator in the scene, which is easy to operate and can ensure that scenes that are difficult to extract feature points can complete three-dimensional reconstruction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to computer vision technology, and in particular to a three-dimensional reconstruction method and system. Background Art

[0002] With the continuous development of computer technology and image processing technology, the three-dimensional reconstruction technology that reflects a scene or object in the real world is becoming more and more widely used. In the current three-dimensional reconstruction technology, the reconstruction of three-dimensional images using two-dimensional images has become a very common method. In the whole process, the extraction and matching of feature points of two-dimensional images is a very important part. According to the widely used SIFT feature detection algorithm, the process is mainly divided into the steps of constructing scale space, extreme value detection, feature point location, feature direction assignment, feature point description and feature point matching. According to the characteristics of the algorithm, when there are complex textures, contours, boundaries and other forms on the two-dimensional image, it will be easier to locate and describe the feature points.

[0003] However, in actual operations, it is difficult to find feature points in the two-dimensional images of many scenes. For example, in pure color scenes such as white walls, large glass, and pure color tiles, or scenes where large areas of graphics appear regularly and repeatedly, it is impossible to extract feature points, and thus it is impossible to complete three-dimensional reconstruction through feature points. Because the above-mentioned scenes where it is difficult to extract feature points from two-dimensional images are very common in real life, the three-dimensional reconstruction of these scenes cannot be achieved based on two-dimensional image feature points, and can only be achieved by other technical means. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that scenes with difficult extraction of two-dimensional image feature points cannot be reconstructed in a three-dimensional manner based on feature points, and a three-dimensional reconstruction method is provided, which extracts feature points through projection features formed by light projected in the scene by a light source generator, is easy to operate, and can ensure that scenes with difficult extraction of feature points can complete three-dimensional reconstruction. The present invention also discloses a system for implementing the above three-dimensional reconstruction method.

[0005] The purpose of the present invention is mainly achieved through the following technical solutions: A three-dimensional reconstruction method, comprising:

[0006] Acquire a scene image with a projection feature as a first image, wherein the projection feature is a feature formed by the light source generator projecting light in the scene;

[0007] Acquire a scene image without projection features as a second image;

[0008] constructing a three-dimensional mesh model using the first image;

[0009] The second image is used to perform texture mapping on the three-dimensional mesh model to complete the three-dimensional reconstruction. Texture mapping is the automatic extraction and mapping of textures, which can paste the texture image onto the model surface and keep the texture in proper relationship with the attached plane during transformation. Its essence is to establish the correspondence between the two-dimensional texture space point and the three-dimensional object space object surface point, map the color value or grayscale value corresponding to the two-dimensional space point to the three-dimensional object surface, and obtain a three-dimensional model that conforms to the real color vision.

[0010] When the present invention is implemented, for scenes lacking feature points, artificially added images can be used to replace the feature images generated by the light source generator in the scene, that is, some special graphics are artificially pasted in the scene, and the special graphics are used as features, so that feature points can be extracted from the two-dimensional image by adding images later. However, in this way, the larger the scene, the more graphics need to be pasted, and the artificial pasting method is very troublesome. The present invention extracts feature points by forming projection features through the light projected by the light source generator in the scene lacking feature points. The light forms multiple images in the scene. When the first image is obtained, these images are included together with the scene. Feature points are extracted through these images, which is easy to operate and solves the problem of three-dimensional reconstruction of scenes where it is difficult to extract feature points.

[0011] The present invention uses the first image to construct a three-dimensional mesh model and uses the second image to perform texture mapping on the three-dimensional mesh model, which can be achieved by using existing technologies. Since the actual situation of the scene needs to be obtained in the three-dimensional model, it is necessary to remove these images used as features in the three-dimensional model in order to output it as the final product. Feature points are mainly used in the white mold construction stage (i.e., the three-dimensional mesh model construction stage) and are not used in the mapping stage (i.e., the texture mapping stage). Therefore, the present invention uses a scene image with projection features to construct a white mold, and a scene image without projection features as mapping data, so as to avoid the reconstructed three-dimensional model having feature images.

[0012] As a first implementation of the present invention for obtaining the first image and the second image, further, the projection light is visible light, the first image is obtained by shooting, and the second image is obtained by removing the projection features in the first image. In this implementation, the projection features in the first image are removed by using existing image processing software, which is simple to operate. The second image is obtained from the first image, and the position and posture are the same. The images correspond one to one, which is convenient for processing when performing mapping.

[0013] As a second implementation of the present invention for acquiring the first image and the second image, further, the projection light is visible light; the first image and the second image are both acquired by shooting, the light source generator is in an on state when acquiring the first image, and the light source generator is in an off state when acquiring the second image. This implementation is simple to operate, and since the second image and the first image are acquired separately, they should be matched one-to-one as much as possible to avoid slow mapping processing or position correspondence errors in mapping.

[0014] When the first image and the second image of the present invention are both obtained by photography, they can be photographed using the same camera device or different camera devices. This can be implemented regardless of whether the shooting positions of the first image and the second image correspond. When the shooting positions of the first image and the second image do not correspond one-to-one, model construction can also be achieved based on existing technologies.

[0015] As a first implementation mode of shooting the first image and the second image of the present invention, further, the first image and the second image are shot by using the same camera device or different camera devices; the shooting positions of the first image and the second image are not one-to-one corresponding.

[0016] As a second implementation of the present invention for shooting the first image and the second image, further, the first image and the second image are obtained by shooting with the same camera device or different camera devices; the shooting positions of the first image and the second image correspond one to one. This implementation ensures that the second image corresponds one to one with the first image, facilitates data replacement and processing when mapping, and can improve processing efficiency and accuracy.

[0017] As a third implementation manner of acquiring the first image and the second image of the present invention, further, the projection light is non-visible light;

[0018] The first image acquisition method is: using a camera equipped with a first filter device to shoot a projection image with projection features in the scene, and fusing each projection image with a second image corresponding to its shooting position into the first image in a one-to-one correspondence; the first filter device is used to filter out light other than the non-visible light;

[0019] The projection image and the second image are obtained by shooting with the same camera device or different camera devices. The projection image has only projection features, which is equivalent to having only light spots. Only light spots cannot construct an effective three-dimensional grid model. It is necessary to establish a positional relationship between the light spots and the scene in order to construct an effective and real three-dimensional grid model. The first image obtained by fusing the projection image with the second image establishes the positional relationship between the light spots and the scene, and includes a feature image formed by non-visible light and a normal scene image formed by visible light. When shooting the second image, the camera device is not equipped with the first filter device.

[0020] In the present invention, when the projection light is non-visible light, further, the non-visible light is any one of near infrared light, mid-infrared light, far infrared light, ultraviolet light and infrared light. Among them, near infrared light, mid-infrared light, far infrared light, ultraviolet light and infrared light cannot cause vision, but the graphics generated by the projection can be photographed by camera.

[0021] In the present invention, when the projection light is non-visible light, further, the second image acquisition method is: using a camera equipped with a second filter device to shoot the second image; the second filter device is used to filter out light other than visible light. Existing camera equipment usually has a filter or a filter film to filter out light other than visible light to reduce the impact of non-visible light on visible light. In this way, when the present invention is implemented, the second filter device is easy to obtain.

[0022] In order to be more conducive to distinguishing scenes, further, the projection feature shapes and / or sizes formed by the light source generator projecting light in the scene are different.

[0023] A system for implementing the above three-dimensional reconstruction method comprises:

[0024] A light source generator, for projecting light into the scene to form a projection feature;

[0025] A first image acquisition module, used to acquire a scene image with projection features as a first image;

[0026] A second image acquisition module, used to acquire a scene image without projection features as a second image;

[0027] The model construction module is used to use the first image to construct a three-dimensional mesh model, and use the second image to perform texture mapping on the three-dimensional mesh model to complete the reconstruction of the three-dimensional model.

[0028] In summary, compared with the prior art, the present invention has the following beneficial effects: the present invention projects light into a scene lacking feature points through a light source generator to form projection features, and extracts feature points through the projection features to obtain a final matched feature point cloud, and then constructs a white model through the obtained feature point cloud, and then maps the scene image without projection features onto the white model to complete three-dimensional reconstruction. The operation is convenient, and can ensure that three-dimensional reconstruction of scenes where feature points are difficult to extract can be implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0030] Figure 1 A flowchart of a specific embodiment of the present invention;

[0031] Figure 2 A schematic diagram of a light source generator forming a projection feature in a scene;

[0032] Figure 3 The system structure diagram of a specific embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0034] Embodiment 1:

[0035] like Figure 1 As shown, a three-dimensional reconstruction method includes: obtaining a scene image with projection features as a first image, wherein the projection features are features formed by the light source generator projecting light in the scene; obtaining a scene image without projection features as a second image; using the first image to construct a three-dimensional mesh model; using the second image to perform texture mapping on the three-dimensional mesh model to complete the three-dimensional reconstruction. Wherein, texture mapping is to paste a predefined texture or image onto the surface of a three-dimensional space object according to a certain mapping algorithm to establish a one-to-one correspondence between the texture and the space object. The texture mapping of this embodiment is divided into three steps. First, the second image is cut into corresponding texture triangles according to the triangles in the three-dimensional mesh model, and then the correspondence between the three-dimensional mesh model and the texture triangles is established. Each triangle in the three-dimensional mesh model may correspond to multiple texture triangles. It is necessary to establish a one-to-many relationship between the triangles of the three-dimensional mesh model and the texture triangles, select the optimal texture triangle from these texture triangles, and convert the one-to-many relationship into a one-to-one relationship. Finally, a mapping function between the triangles of the three-dimensional mesh model and the texture triangles is established to map the two-dimensional plane texture triangles to the three-dimensional mesh model to complete the texture mapping.

[0036] In this embodiment, the number of first images and second images acquired are both multiple, and there is no order requirement for acquiring the first images and the second images in specific operations. When reconstructing the three-dimensional model of the scene based on the first images and the second images, the first image is used to construct a white mold, that is, a three-dimensional mesh model, which is composed of countless triangular meshes, and the second image is used as mapping data, that is, texture mapping. Figure 2This is an example of the application of this embodiment on a wall. The light source generator forms multiple images on the wall as projection features. When taking a photo, these images are taken together with the wall, and feature points are extracted from the image to complete the three-dimensional reconstruction and build an overall three-dimensional model. The three-dimensional model reconstruction of this embodiment can be implemented using existing three-dimensional modeling software such as Pictometry system, Smart 3D Capture, Vision Map software, Ultramap, DP Modeler, Super MapGIS7C, Pix4Dmapper, PhotoScan, etc., or can be implemented by programming.

[0037] In the specific implementation of this embodiment, due to the influence of various factors such as the time of image acquisition, external lighting conditions, shooting angle, etc., the acquired image may have uneven lighting distribution, large color differences, etc., which affects the production and application of the three-dimensional model, so the first image can be first processed for light and color uniformity. Commonly used algorithms for light and color uniformity include: light uniformity based on MASK, light uniformity based on frequency domain, light uniformity based on adaptive template, and light and color uniformity based on Wallis filter. Among them, the light and color uniformity method based on Wallis filter has better processing effect and higher processing efficiency than the first three methods. This embodiment preferably adopts the light and color uniformity method based on Wallis filter. After the light and color uniformity processing is performed in this embodiment, the SIFT algorithm, BUNDLER algorithm, Harris algorithm, SUSAN algorithm, Foerstner algorithm, Moravec algorithm, etc. can be used to extract feature points of the first image, and then a white model is constructed. Among them, the SIFT algorithm is an image local feature description operator based on scale space, which has scale invariance and rotation invariance, good resistance to light, noise, etc., and has a good feature description effect for areas occluded by some objects. Therefore, this embodiment preferably uses the SIFT algorithm to extract feature points.

[0038] like Figure 3 As shown, the system for implementing the above three-dimensional reconstruction method includes: a light source generator, which is used to project light in the scene to form a projection feature; a first image acquisition module, which is used to acquire a scene image with a projection feature as a first image; a second image acquisition module, which is used to acquire a scene image without a projection feature as a second image; and a model construction module, which is used to use the first image to construct a three-dimensional mesh model, and use the second image to perform texture mapping on the three-dimensional mesh model to complete the three-dimensional model reconstruction. The light source generator is turned on when it is necessary to acquire an image with a projection feature, and when shooting is required during the image acquisition process, a camera device is used to shoot.

[0039] Embodiment 2:

[0040] This embodiment makes the following further limitations on the basis of Embodiment 1: the projection light of this embodiment is visible light, the first image is obtained by shooting, and the second image is obtained by removing the projection features in the first image. The second image of this embodiment is obtained by processing the first image, which shows that there is a sequence in which the first image and the second image are obtained in this embodiment. When this embodiment processes the first image into the second image, it is implemented or programmed using existing image processing software such as Photoshop. When using software to remove the projection features, in order to reduce the impact on the later mapping, other images at the projection features should be avoided as much as possible from being distorted, dislocated, deformed, and other problems.

[0041] Embodiment 3:

[0042] This embodiment makes the following further limitations on the basis of Embodiment 1: the projection light of this embodiment is visible light, the first image and the second image are both obtained by shooting, the light source generator is in the on state when the first image is obtained, and the light source generator is in the off state when the second image is obtained, and there is no order in which the first image and the second image are obtained in this embodiment. The first image and the second image of this embodiment can be obtained by shooting with the same camera device or by different camera devices, and whether the shooting positions when shooting the first image and the second image correspond one to one does not affect the implementation of this embodiment. If the shooting positions of the first image and the second image do not correspond, texture mapping can also be completed according to the principle of texture mapping. When the shooting positions of the first image and the second image do not correspond, one camera device or two camera devices can be used. One camera device can be used to shoot all the first images or the second images first, and then shoot the second image or the first image. Two camera devices can be used to do both at the same time.

[0043] In the specific implementation of this embodiment, in order to improve processing accuracy and efficiency, it is preferred to set the shooting positions when shooting the first image and the second image to be one-to-one corresponding. Ways to achieve a one-to-one correspondence between the shooting positions of the first image and the second image include: 1. Using a camera device, when shooting, first shoot a first image or a second image, then keep the posture and position of the camera device unchanged, control the opening and closing of the light source generator, and then shoot another image. 2. Using two cameras with the same lens orientation, collect the first image and the second image corresponding to the shooting position. When this embodiment is applied, it is necessary to repeatedly turn off or on the light source generator in conjunction with shooting.

[0044] The second image of this embodiment is not obtained by using image processing software, so the operation is more convenient and the accuracy is higher.

[0045] Embodiment 4:

[0046] Based on Example 1, this embodiment further defines the first image acquisition method as follows: This embodiment uses a camera equipped with a first filter device to shoot a projection image with projection features in the scene, and fuses each projection image with the second image corresponding to its shooting position into the first image in a one-to-one correspondence. Among them, the first filter device is used to filter out light other than non-visible light. When acquiring the second image, this embodiment uses a camera equipped with a second filter device to shoot the second image, and the second filter device is used to filter out light other than visible light. It should be noted that ordinary camera devices that shoot visible light are usually equipped with filters or filter films to reduce the impact of non-visible light on visible light. When shooting the second image, this embodiment can be implemented by using an existing conventional camera equipped with filters or filter films. Since the camera device that shoots visible light has a filter device, as described in Examples 1-3, even if this embodiment does not emphasize that the camera device that shoots visible light is equipped with a filter device, based on the common knowledge of those skilled in the art, the method of acquiring the second image in this embodiment can also be implemented. The first image of this embodiment is obtained by fusion of the projection image and the second image, and the modeled image and the texture are Figure 1 One-to-one correspondence can improve mapping efficiency.

[0047] When acquiring images in this embodiment, it is necessary to first acquire the projection image and the second image, and then acquire the first image. The projection image and the second image of this embodiment are obtained by shooting with the same camera device or different camera devices. When specifically acquiring the projection image and the second image, it does not matter which image is acquired first. When this embodiment is applied, in order to facilitate the fusion of the projection image and the second image to obtain the first image, when one camera device is used for shooting, the projection image or the second image is shot first, and then the posture and position of the camera device are kept unchanged, the filter device is switched, and then another image is shot, wherein the camera device is equipped with a first filter device when shooting the projection image, and the camera device is equipped with a second filter device when shooting the second image; when two camera devices are used for shooting, the first filter device and the second filter device are respectively equipped, and the projection image is shot with the camera device equipped with the first filter device, and the second image is shot with the camera device equipped with the second filter device. At this time, the lenses of the two camera devices are facing the same direction, and the projection image and the second image corresponding to the shooting position are collected.

[0048] The projection light of this embodiment is non-visible light, wherein the non-visible light used in this embodiment is any one of near-infrared light, mid-infrared light, far-infrared light, ultraviolet light and infrared light. Here, taking the projection light as near-infrared light as an example, near-infrared light is an electromagnetic wave between visible light and mid-infrared light, with a wavelength in the range of 780 to 2526 nm. The camera equipment uses a CMOS camera. Although the response ability of CMOS cameras to the near-infrared band is generally not as strong as that of the visible light band, the intensity of sunlight in the range from visible light to infrared light also decreases with the increase of wavelength, and the atmosphere has a significant absorption effect on the 940nm spectrum of sunlight, so the influence of sunlight on this band is relatively small, thereby avoiding the reduction of clarity of the characteristic image due to the influence of sunlight. In this embodiment, it is feasible to increase the power of the light source generator or adjust the camera parameters, because the light source generator generates non-visible light and will not affect the actual color of the scene objects. The captured two-dimensional images with characteristic images only capture the characteristic images and do not capture the scene objects. Since the CMOS camera has a weak response ability to this band, the ISO of the camera can be increased to improve the camera's sensitivity, so that the characteristic image formed by the captured near-infrared light is clear, thereby ensuring the shooting effect.

[0049] Embodiment 5:

[0050] In order to make it easier to distinguish scenes, this embodiment makes the following further limitations on the basis of any one of Embodiments 1 to 4: the projection feature shapes and / or sizes formed by the light projected by the light source generator of this embodiment in the scene are different. In this way, it can be ensured that there are differences in the projection features projected in the scene when this embodiment is applied. In specific implementation, light spots of different shapes and patterns can be directly projected, or the light source generator can emit light spots of repeated regular patterns. Since the light source has different positional relationships with different positions in the scene, the projected light spots appear in different shapes and / or sizes, such as Figure 2 As shown, the light source emits a square, and the projection feature in the upper left corner is an irregular square.

[0051] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A three-dimensional reconstruction method, It is characterized in that include: Acquire a scene image with a projection feature as a first image, wherein the projection feature is a feature formed by the light source generator projecting light in the scene; Acquire a scene image without projection features as a second image; constructing a three-dimensional mesh model using the first image; Using the second image to perform texture mapping on the three-dimensional mesh model to complete three-dimensional reconstruction; The projection light is visible light or invisible light; When the projection light is visible light, the first image and the second image are acquired in the following manner: the first image is acquired by photographing, and the second image is acquired by removing the projection feature in the first image; or the first image and the second image are both acquired by photographing, the light source generator is in an on state when acquiring the first image, and the light source generator is in an off state when acquiring the second image; When the projection light is non-visible light; the first image acquisition method is: using a camera equipped with a first filter device to shoot a projection image with projection features in the scene, and merging each projection image with a second image corresponding to its shooting position into the first image in a one-to-one correspondence; the first filter device is used to filter out light other than the non-visible light; the projection image and the second image are obtained by shooting with the same camera device or different camera devices; The texture mapping of the three-dimensional mesh model using the second image includes the following steps: first, the second image is cut into corresponding texture triangles according to the triangles in the three-dimensional mesh model, then the corresponding relationship between the three-dimensional mesh model and the texture triangles is established, and finally a mapping function between the triangles of the three-dimensional mesh model and the texture triangles is established to map the two-dimensional plane texture triangles to the three-dimensional mesh model to complete the texture mapping; wherein, when establishing the corresponding relationship between the three-dimensional mesh model and the texture triangles, a one-to-many relationship between the triangles of the three-dimensional mesh model and the texture triangles is first established, and the optimal texture triangles are selected from the set of these texture triangles to convert the one-to-many relationship into a one-to-one relationship.

2. A three-dimensional reconstruction method according to claim 1, It is characterized in that When the first image and the second image are both obtained by photography, the first image and the second image are obtained by photography using the same camera device or different camera devices; the shooting positions of the first image and the second image are not one-to-one corresponding.

3. A three-dimensional reconstruction method according to claim 1, It is characterized in that When the first image and the second image are both obtained by photography, the first image and the second image are obtained by photography using the same camera device or different camera devices; and the shooting positions of the first image and the second image correspond one to one.

4. A three-dimensional reconstruction method according to claim 1, It is characterized in that The non-visible light is any one of near infrared light, mid infrared light, far infrared light, ultraviolet light and infrared light.

5. A three-dimensional reconstruction method according to claim 1, It is characterized in that When the projection light is non-visible light, the second image is acquired in the following manner: the second image is captured by using a camera equipped with a second filter device; the second filter device is used to filter out light other than visible light.

6. A three-dimensional reconstruction method according to any one of claims 1 to 5, It is characterized in that The projection features formed by the light projected by the light source generator in the scene have different shapes and / or sizes.

7. A system for implementing a three-dimensional reconstruction method according to any one of claims 1 to 6, It is characterized in that include: A light source generator, for projecting light into the scene to form a projection feature; A first image acquisition module, used to acquire a scene image with projection features as a first image; A second image acquisition module, used to acquire a scene image without projection features as a second image; A model construction module, configured to construct a three-dimensional mesh model using the first image, and perform texture mapping on the three-dimensional mesh model using the second image to complete three-dimensional model reconstruction; The projection light is visible light or invisible light; When the projection light is visible light, the first image and the second image are acquired in the following manner: the first image is acquired by photographing, and the second image is acquired by removing the projection feature in the first image; or the first image and the second image are both acquired by photographing, the light source generator is in an on state when acquiring the first image, and the light source generator is in an off state when acquiring the second image; When the projection light is non-visible light, the first image is obtained by: using a camera equipped with a first filter device to shoot a projection image with projection features in the scene, and fusing each projection image with a second image corresponding to its shooting position into the first image in a one-to-one correspondence; the first filter device is used to filter out light other than the non-visible light; the projection image and the second image are obtained by shooting with the same camera device or different camera devices; The texture mapping of the three-dimensional mesh model using the second image includes the following steps: first, the second image is cut into corresponding texture triangles according to the triangles in the three-dimensional mesh model, then the corresponding relationship between the three-dimensional mesh model and the texture triangles is established, and finally a mapping function between the triangles of the three-dimensional mesh model and the texture triangles is established to map the two-dimensional plane texture triangles to the three-dimensional mesh model to complete the texture mapping; wherein, when establishing the corresponding relationship between the three-dimensional mesh model and the texture triangles, a one-to-many relationship between the triangles of the three-dimensional mesh model and the texture triangles is first established, and the optimal texture triangles are selected from the set of these texture triangles to convert the one-to-many relationship into a one-to-one relationship.