Method, device and storage medium for evaluating quality of three-dimensional scene reconstruction

Abstract

The application relates to the technical field of computer vision, and discloses a three-dimensional scene reconstruction quality evaluation method, which comprises the following steps: collecting multiple images of a real three-dimensional scene, wherein a graphic code is arranged in the real three-dimensional scene, and the multiple images comprise an image containing the graphic code; reconstructing a virtual three-dimensional scene corresponding to the real three-dimensional scene by using the multiple images, wherein the virtual three-dimensional scene comprises a virtual graphic code corresponding to the graphic code; determining the virtual graphic code in the virtual three-dimensional scene; performing scanning identification on the virtual graphic code to obtain an identification result of the virtual graphic code; and determining the quality of the virtual three-dimensional scene according to the identification result. In the above manner, the quality of the reconstructed three-dimensional scene can be objectively and accurately evaluated.

Description

Technical Field

[0001] This application relates to the field of computer vision technology, specifically to a method, device, and storage medium for evaluating the quality of 3D scene reconstruction. Background Technology

[0002] 3D reconstruction is a technique that reconstructs the three-dimensional structure of an object or scene by analyzing and processing two-dimensional images or scanned data taken from different perspectives. With the development of computer vision technology, 3D reconstruction has been widely used in fields such as Virtual Reality (VR) and Augmented Reality (AR).

[0003] For virtual 3D scenes reconstructed from real-world 3D scenes, it is generally required that the virtual 3D scene highly replicate the real-world 3D scene in terms of geometric structure, texture details, and spatial layout to ensure visual realism and spatial consistency. Therefore, quality assessment of virtual 3D scenes is crucial; it is not only a key means of measuring the reconstruction effect but also an important basis for guiding technology optimization and practical application. Currently, there is a lack of a method that can objectively and accurately evaluate virtual 3D scenes. Summary of the Invention

[0004] In view of the above problems, this application provides a method, device and storage medium for evaluating the quality of 3D scene reconstruction, which solves the problem that the quality of reconstructed 3D scenes cannot be objectively and accurately evaluated in the prior art.

[0005] According to one aspect of the embodiments of this application, a method for evaluating the quality of three-dimensional scene reconstruction is provided. The method includes: acquiring multiple images of a real three-dimensional scene, wherein graphic codes are arranged in the real three-dimensional scene, and the multiple images include an image containing the graphic codes; reconstructing a virtual three-dimensional scene corresponding to the real three-dimensional scene using the multiple images, wherein the virtual three-dimensional scene includes virtual graphic codes corresponding to the graphic codes; determining the virtual graphic codes in the virtual three-dimensional scene; scanning and recognizing the virtual graphic codes to obtain a recognition result of the virtual graphic codes; and determining the quality of the virtual three-dimensional scene based on the recognition result.

[0006] In one optional approach, the recognition result includes successful recognition of the virtual graphic code and failure to recognize the virtual graphic code; determining the quality of the virtual 3D scene based on the recognition result includes: if the recognition result is successful recognition of the virtual graphic code, then the quality of the virtual 3D scene is determined to be a first level; if the recognition result is failure to recognize the virtual graphic code, then the quality of the virtual 3D scene is determined to be a second level, wherein the first level is superior to the second level.

[0007] In one alternative approach, a plurality of graphic codes are arranged in the real three-dimensional scene, and the plurality of graphic codes are arranged at multiple locations in the real three-dimensional scene and have multiple orientations.

[0008] In one alternative approach, the plurality of locations include the ground, walls, and top.

[0009] In one optional embodiment, the virtual 3D scene includes multiple virtual graphic codes corresponding to multiple graphic codes; the recognition result includes successful virtual graphic code recognition and failed virtual graphic code recognition; determining the quality of the virtual 3D scene based on the recognition result includes: determining a first number of recognition results with successful virtual graphic code recognition from the multiple recognition results obtained by scanning and recognizing the multiple virtual graphic codes; determining a ratio between the first number and a second number, wherein the second number is the number of virtual graphic codes included in the virtual 3D scene; determining a target ratio range from multiple preset ratio ranges; and determining the target quality level corresponding to the target ratio range as the quality of the virtual 3D scene based on the preset correspondence between multiple ratio ranges and multiple quality levels.

[0010] In one optional approach, the multiple graphic codes have different sizes; the virtual 3D scene includes multiple virtual graphic codes of different sizes corresponding to the multiple graphic codes, and the virtual graphic codes of different sizes correspond to different scores; the recognition result includes successful virtual graphic code recognition and failed virtual graphic code recognition; determining the quality of the virtual 3D scene based on the recognition result includes: determining the target virtual graphic code whose recognition result is a successful virtual graphic code recognition based on the recognition result; determining the sum of the scores corresponding to all the target virtual graphic codes to obtain a total score; determining the target score interval to which the total score belongs from multiple preset score intervals; and determining the target quality level corresponding to the target score interval as the quality of the virtual 3D scene based on the correspondence between the preset multiple score intervals and multiple quality levels.

[0011] In one alternative approach, among the plurality of graphic codes, the smallest graphic code has a different color than the other graphic codes, and / or the smallest graphic code is marked in the real-world 3D scene.

[0012] In one alternative approach, the graphic code is a QR code.

[0013] According to another aspect of the embodiments of this application, a three-dimensional scene reconstruction quality assessment device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the three-dimensional scene reconstruction quality assessment method as described above.

[0014] According to another aspect of the embodiments of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the three-dimensional scene reconstruction quality assessment method as described above.

[0015] If the virtual graphic code can be successfully recognized, it indicates that during the 3D reconstruction of the graphic code in the real-world 3D scene, not only was the shape, position, and orientation of the graphic code accurately captured, but the integrity and recognizability of its key modules were also preserved. This results in a high-fidelity restoration of the graphic code in the real-world 3D scene in terms of geometric structure, texture details, and spatial layout, indicating a high reconstruction quality of the virtual graphic code. Since the graphic code, as a marker with clear structural and informational features, is used in the 3D reconstruction of the entire real-world 3D scene, its reconstruction quality can indirectly reflect the reconstruction accuracy and detail restoration capability of the entire real-world 3D scene. Therefore, when the virtual graphic code can be accurately recognized, it can be reasonably inferred that the reconstructed virtual 3D scene has also achieved a high level of restoration in terms of spatial structure, detail representation, and visual consistency. Moreover, the recognition result obtained by scanning and recognizing the graphic code is objective. Therefore, in this embodiment, after 3D reconstruction using the above method, by determining whether the virtual graphic code can be successfully recognized, the quality of the virtual 3D scene can be objectively and accurately determined based on the recognition result.

[0016] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0017] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0018] Figure 1 A flowchart of the three-dimensional scene reconstruction quality assessment method provided in the embodiments of this application is shown;

[0019] Figure 2 The embodiments provided in this application are shown. Figure 1 A flowchart illustrating the sub-steps of step 150 in the diagram;

[0020] Figure 3 Another embodiment of this application is shown. Figure 1 A flowchart illustrating the sub-steps of step 150 in the middle section;

[0021] Figure 4 A schematic diagram of the structure of the three-dimensional scene reconstruction quality assessment device provided in the embodiments of this application is shown. Detailed Implementation

[0022] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. Although exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein.

[0023] Currently, conventional 3D reconstruction methods are used to reconstruct real-world 3D scenes, resulting in virtual 3D scenes. The quality of these virtual 3D scenes is typically evaluated by manually assessing their fidelity to the real-world scene in terms of geometry, texture detail, and spatial layout. However, this quality assessment method struggles to achieve fair comparisons between different reconstruction models or algorithms. Furthermore, because it relies on subjective human evaluation, it lacks a scientific basis for processes such as engineering acceptance, performance optimization, and technological iteration.

[0024] To address the aforementioned issues, this application proposes a method for evaluating the quality of 3D scene reconstruction. This method involves acquiring multiple images of a real-world 3D scene, in which graphic codes are arranged, and including images containing these graphic codes among the acquired images. Then, a virtual 3D scene corresponding to the real-world 3D scene is reconstructed using these images. The virtual graphic codes within the virtual 3D scene are identified and scanned for recognition. The quality of the virtual 3D scene is determined based on the recognition results. Since evaluating the quality of the virtual 3D scene based on the recognition results of the virtual graphic codes does not rely on subjective human judgment, it allows for an objective and accurate assessment of the virtual 3D scene's quality.

[0025] Figure 1A flowchart of a three-dimensional scene reconstruction quality assessment method provided in an embodiment of this application is shown. This method is executed by a terminal device, which may include one or more processors, such as a 3D scanning device, a touchscreen phone, a smartphone, a tablet computer, a portable electronic device, or other electronic devices including a camera. The processor may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application; no limitation is made here. The one or more processors included in the terminal device may be processors of the same type, such as one or more CPUs; or they may be processors of different types, such as one or more CPUs and one or more ASICs; no limitation is made here. Figure 1 As shown, the method includes the following steps:

[0026] Step 110: Collect multiple images of the real 3D scene.

[0027] The real-world 3D scene includes graphic codes, which can be barcodes or QR codes. Specifically, graphic codes can be pre-placed in the real-world 3D scene. For example, the graphic codes can be printed out in advance and then placed in the real-world 3D scene; or objects containing graphic codes (such as product packaging printed with graphic codes) can be placed in the real-world 3D scene.

[0028] In this step, multiple images (e.g., RGB images) of the real-world 3D scene can be captured using the camera device in the terminal equipment. Since these multiple images are used for subsequent 3D reconstruction, it is necessary to capture the real-world 3D scene from multiple angles, ensuring that every object and every region in the scene is captured, thus obtaining multiple images. The number of images captured can be determined based on the size of the real-world 3D scene and the number of objects within it. Generally, the larger the real-world 3D scene and the more objects it includes, the more images need to be captured.

[0029] Since the graphic codes placed in the real-world 3D scene are used to evaluate the quality of the virtual 3D scene obtained from the 3D reconstruction, the multiple images acquired in this step must include images containing graphic codes. For example, at least one image among the multiple images must contain a complete graphic code.

[0030] Step 120: Reconstruct a virtual 3D scene corresponding to the real 3D scene using multiple images.

[0031] In this step, existing 3D reconstruction techniques (such as 3DGS technology) can be used to reconstruct the virtual 3D scene. It is worth noting that since the multiple images acquired in step 110 include images containing graphic codes, the virtual 3D scene reconstructed in this step will correspondingly include virtual graphic codes, and the virtual graphic codes in the virtual 3D scene correspond to the graphic codes arranged in the real 3D scene.

[0032] Step 130: Determine the virtual graphic code in the virtual 3D scene.

[0033] In this step, the location of the virtual graphic code within the virtual 3D scene is determined. For example, the coordinates of the graphic code in the real 3D scene can be mapped to the coordinate system of the virtual 3D scene based on its actual placement, thus determining the virtual graphic code within the virtual 3D scene. Alternatively, each area in the virtual 3D scene can be examined individually to locate the virtual graphic code.

[0034] Step 140: Scan and recognize the virtual graphic code to obtain the recognition result of the virtual graphic code.

[0035] After determining the virtual graphic code in step 130, this step allows the decoding device in the terminal settings to scan and recognize the virtual graphic code. The recognition result includes successful virtual graphic code recognition and virtual graphic code recognition failure.

[0036] Step 150: Determine the quality of the virtual 3D scene based on the recognition results.

[0037] In this step, if there is only one graphic code in the real-world 3D scene, and correspondingly only one virtual graphic code in the virtual 3D scene, then if the recognition result is that the virtual graphic code is successfully recognized, the quality of the virtual 3D scene is determined to be at level one; if the recognition result is that the virtual graphic code fails to be recognized, the quality of the virtual 3D scene is determined to be at level two, where level one is superior to level two. Specifically, a correspondence between recognition results and quality levels can be established in advance and stored in the terminal device. After scanning and recognizing the virtual graphic code in step 140, the quality level of the virtual 3D scene can be determined in this step based on this correspondence.

[0038] Taking a QR code as an example, a QR code consists of multiple parts, including a positioning pattern, an alignment pattern, a timing pattern, format information, version information, and a data module. The positioning pattern is located in the three corners of the QR code, helping scanning devices quickly identify its position and orientation. The alignment pattern assists in image correction, ensuring accurate reading from different viewing angles. The timing pattern, composed of alternating black and white modules, determines the row and column coordinates of the QR code. The format information includes the error correction level and mask mode of the QR code, ensuring accurate decoding. The version information indicates the size specifications of the QR code. The data module is the core of the QR code, storing actual information through the arrangement and combination of black and white modules, and utilizing error correction code technology to improve robustness. These structures work together to enable the QR code to efficiently and reliably store and transmit information. Therefore, only when the structure of the virtual graphic code is complete can it be successfully recognized.

[0039] If the virtual graphic code can be successfully recognized, it indicates that during the 3D reconstruction of the graphic code in the real-world 3D scene, not only was the shape, position, and orientation of the graphic code accurately captured, but the integrity and recognizability of its key modules were also preserved. This results in a high-fidelity restoration of the graphic code in the real-world 3D scene in terms of geometric structure, texture details, and spatial layout, indicating a high reconstruction quality of the virtual graphic code. Since the graphic code, as a marker with clear structural and informational features, is used in the 3D reconstruction of the entire real-world 3D scene, its reconstruction quality can indirectly reflect the reconstruction accuracy and detail restoration capability of the entire real-world 3D scene. Therefore, when the virtual graphic code can be accurately recognized, it can be reasonably inferred that the reconstructed virtual 3D scene has also achieved a high level of restoration in terms of spatial structure, detail representation, and visual consistency. Moreover, the recognition result obtained by scanning and recognizing the graphic code is objective. Therefore, in this embodiment, after 3D reconstruction using the above method, by determining whether the virtual graphic code can be successfully recognized, the quality of the virtual 3D scene can be objectively and accurately determined based on the recognition result.

[0040] Furthermore, if the quality of a virtual 3D scene is evaluated by manually determining the fidelity of virtual objects to real-world objects, it typically requires comparing the size, shape, and texture information of each virtual object with that of the real-world object. This process is cumbersome and time-consuming. In this embodiment, by directly scanning and recognizing the virtual graphic code, the reconstruction quality of the virtual 3D scene can be quickly determined based on the recognition results, thereby improving the efficiency of quality assessment.

[0041] It's worth noting that while graphic tags in real-world 3D scenes are placed to evaluate the quality of virtual 3D scenes, their generally small size means that even if a reconstructed virtual 3D scene includes virtual graphic tags, it won't necessarily affect its normal use in certain situations. For example, in applications like virtual displays, digital twins, or scene simulations, users prioritize the realism and interactivity of the overall environment, and graphic tags, as tiny markers, often don't interfere with the main visual experience or functionality. However, if high standards are required for the virtual 3D scene and the presence of virtual graphic tags is unacceptable, they can be removed using image processing or model editing techniques to meet high visual presentation requirements.

[0042] In some embodiments, steps 110 to 150 above may be partially performed by the terminal device and partially performed manually. For example, steps 110 to 120 may be performed by the terminal device, and the specific execution method is as described in the above embodiments. Steps 130 to 150 may be performed manually, which can also objectively and accurately determine the quality of the virtual 3D scene.

[0043] Specifically, in step 130, the virtual graphic code in the virtual 3D scene can be found by manually operating the terminal device. For example, the 3D scene reconstruction software used to reconstruct the virtual 3D scene in the terminal device can be used to find the virtual graphic code from the virtual 3D scene.

[0044] In step 140, if the terminal device does not have a decoding device, the virtual graphic code can be scanned and recognized by manually operating the decoding device in an electronic device (such as a smartphone or smartwatch) to obtain the recognition result of the virtual graphic code.

[0045] In step 150, the quality of the virtual 3D scene is determined manually based on the recognition results. If the virtual graphic code is successfully recognized, the quality of the virtual 3D scene is determined to be at the first level; if the virtual graphic code fails to be recognized, the quality of the virtual 3D scene is determined to be at the second level. Although the aforementioned prior art also uses manual evaluation of the quality of the virtual 3D scene, the prior art subjectively determines the degree of fidelity of the virtual 3D scene to the real 3D scene in terms of geometric structure, texture details, and spatial layout to evaluate the quality of the virtual 3D scene. This differs from the method of manual evaluation of the quality of the virtual 3D scene in this application. In this application, since the recognition result of the virtual graphic code only includes two cases—one is successful recognition, and the other is failure—the result of the virtual graphic code obtained through step 140 is objective and accurate. Because the recognition result of the virtual graphic code is objective and accurate, even if the quality of the virtual 3D scene is evaluated manually based on the recognition result of the virtual graphic code in this application, the evaluation result is still objective and accurate.

[0046] To more accurately assess the quality of the virtual 3D scene, in this embodiment, multiple graphic codes (e.g., 2, 5, or 8) are arranged in the real 3D scene. These graphic codes are placed in multiple locations within the real 3D scene (e.g., the ground, walls, and ceiling), and each graphic code has multiple orientations. Correspondingly, the virtual 3D scene obtained from the 3D reconstruction includes multiple virtual graphic codes corresponding to the multiple graphic codes. When the virtual 3D scene includes multiple virtual graphic codes, steps 130-140 involve identifying all virtual graphic codes and scanning and recognizing each one to obtain multiple recognition results. Figure 2 The embodiments provided in this application are shown. Figure 1 A flowchart illustrating the sub-steps of step 150 in the diagram. (See attached flowchart.) Figure 2 As shown in the embodiment of this application, step 150 includes steps 1511 to 1514.

[0047] Step 1511: Based on the multiple recognition results obtained by scanning and recognizing multiple virtual graphic codes, determine the first number of recognition results in which the virtual graphic codes are successfully recognized.

[0048] If a virtual 3D scene contains n virtual graphic codes, and m of these n codes are successfully recognized, then the first quantity is m. Here, n and m are both positive integers, 1 < n and m ≤ n.

[0049] Step 1512: Determine the ratio of the first quantity to the second quantity.

[0050] The second quantity is n. This step determines the ratio of m to n.

[0051] Step 1513: Determine the target ratio interval to which the ratio belongs from multiple preset ratio intervals.

[0052] Among them, multiple preset ratio ranges are pre-set and used to quantify the quality level of virtual 3D scenes.

[0053] Step 1514: Based on the pre-defined correspondence between multiple ratio intervals and multiple quality levels, determine the target quality level corresponding to the target ratio interval as the quality of the virtual 3D scene.

[0054] This system pre-sets and stores the correspondence between multiple ratio intervals and multiple quality levels. After determining the target ratio interval in step 1513, the quality level of the virtual 3D scene can be determined in this step based on this correspondence. Specifically, in this correspondence, for any two ratio intervals, the larger the ratio interval, the better the corresponding quality level.

[0055] For scenarios where only one graphic code is placed in a real-world 3D scene, improper placement or severe lighting effects on the captured image containing the graphic code may prevent successful identification of the virtual graphic code, potentially leading to misjudgments when assessing the quality level of the virtual 3D scene. To address this, this embodiment of the application places multiple graphic codes in multiple different locations with multiple different orientations. This effectively reduces the probability of misjudgments due to local errors or accidental factors. Furthermore, by statistically analyzing the recognition success rate of multiple virtual graphic codes, a more objective and comprehensive evaluation of the virtual 3D scene's performance in terms of overall structure, detail reproduction, and spatial consistency can be achieved, thereby improving the reliability and accuracy of quality assessment.

[0056] It is understandable that the more graphic codes are placed, the more virtual graphic codes are included in the virtual 3D scene, and the higher the accuracy of the quality of the virtual 3D scene determined based on the virtual graphic codes. Since determining and scanning virtual graphic codes takes time, the more graphic codes are placed in the real 3D scene, the lower the efficiency of evaluating the quality of the virtual 3D scene. Therefore, in this embodiment, the number of graphic codes placed in the real 3D scene can be determined as needed and is not limited here.

[0057] To further improve the accuracy of evaluating the quality of virtual 3D scenes, multiple graphic codes (e.g., 2, 5, or 8) are arranged in the real 3D scene. These graphic codes have different sizes and are placed in multiple locations within the real 3D scene (e.g., the ground, walls, and ceiling), and they have multiple orientations. Correspondingly, the virtual 3D scene obtained from the 3D reconstruction includes multiple virtual graphic codes corresponding to these multiple graphic codes. When the virtual 3D scene includes multiple virtual graphic codes, steps 130-140 involve identifying all the virtual graphic codes and scanning and recognizing each one to obtain multiple recognition results. Figure 3 Another embodiment of this application is shown. Figure 1 A flowchart illustrating the sub-steps of step 150. (See attached flowchart.) Figure 3 As shown in the embodiment of this application, step 150 includes the following steps 1521 to 1524.

[0058] Step 1521: Based on the recognition results, determine the target virtual graphic code as successfully recognized.

[0059] In this context, if a virtual 3D scene contains r virtual graphic codes, and s of these r virtual graphic codes are successfully recognized, then all s virtual graphic codes are considered target virtual graphic codes. Here, r and s are both positive integers, 1 < r and s ≤ r.

[0060] Step 1522: Determine the sum of the scores corresponding to all target virtual graphic codes to obtain the total score.

[0061] The system pre-sets and stores the correspondence between the size of the graphic codes and their scores; the smaller the graphic code, the higher the corresponding score. In this step, the sum of the scores corresponding to the s target virtual graphic codes determined in step 1521 is determined based on this correspondence to obtain the total score.

[0062] Step 1523: Determine the target score range to which the total score belongs from multiple preset score ranges.

[0063] Among them, multiple preset score ranges are pre-set and used to quantify the quality level of virtual 3D scenes.

[0064] Step 1524: Based on the preset correspondence between multiple score ranges and multiple quality levels, determine the target quality level corresponding to the target score range as the quality of the virtual 3D scene.

[0065] This system pre-sets and stores the correspondence between multiple score ranges and multiple quality levels. After determining the target score range in step 1523, the quality of the virtual 3D scene can be determined in this step based on this correspondence. Specifically, for any two score ranges, the larger the score range, the better the corresponding quality level.

[0066] Smaller-sized virtual graphic codes are more susceptible to insufficient resolution, loss of detail, or viewpoint deviation during reconstruction. Therefore, successful recognition of smaller virtual graphic codes better reflects the terminal device's ability to restore fine structures and complex textures when reconstructing real-world 3D scenes. Thus, in this embodiment, multiple graphic codes of different sizes are arranged, each corresponding to a different score. The recognition results of these multiple virtual graphic codes are then combined, and the total score corresponding to the successfully recognized target virtual graphic code is calculated. Based on this total score, the quality of the virtual 3D scene is evaluated. This allows for a more comprehensive and detailed assessment of the virtual 3D scene's performance at different levels of precision, effectively distinguishing between different reconstruction quality levels. This makes the quality assessment results more objective, accurate, and discriminative, avoiding the bias or insufficient sensitivity issues caused by using only a single-sized graphic code.

[0067] for Figure 3 In the provided embodiments, if the graphic code is small in size, it may be ignored when capturing images of the real-world 3D scene. Alternatively, if a smaller graphic code is placed close to a larger graphic code, the smaller graphic code may also be ignored, resulting in the captured image not including images containing the smaller graphic code. This negatively impacts the subsequent quality assessment of the virtual 3D scene. Therefore, to avoid these problems, in this embodiment, preferably, the smallest graphic code is colored differently from the other graphic codes. This facilitates the differentiation of the smaller graphic code from the others by color, making it less likely to be missed.

[0068] In some embodiments, to facilitate the differentiation of smaller graphic codes from other graphic codes, a marker is provided for the smallest graphic code in the real 3D scene. This marker distinguishes the smaller graphic code from the others, making it less likely to miss it. For example, multiple arrows pointing to the smallest graphic code can be placed near it, or a rectangle can be used to enclose the virtual graphic code.

[0069] Figure 4 The diagram shows a structural schematic of the three-dimensional scene reconstruction quality assessment device provided in the embodiment of this application. The specific embodiments of this application do not limit the specific implementation of the three-dimensional scene reconstruction quality assessment device.

[0070] like Figure 4 As shown, the 3D scene reconstruction quality assessment device 200 may include a processor 202 and a memory 204.

[0071] The memory 204 is used to store the computer program 206. The memory 204 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk drive. The computer program 206 may include computer-executable instructions.

[0072] The processor 202 is used to execute the computer program 306 to implement the above-described embodiment of the three-dimensional scene reconstruction quality assessment method.

[0073] The processor 202 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The one or more processors included in the 3D scene reconstruction quality assessment device 200 may be processors of the same type, such as one or more CPUs; or they may be processors of different types, such as one or more CPUs and one or more ASICs.

[0074] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described embodiment of the three-dimensional scene reconstruction quality assessment method.

[0075] This application provides a computer program that can be executed by a processor to implement the above-described three-dimensional scene reconstruction quality assessment method.

[0076] This application provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described embodiment of the three-dimensional scene reconstruction quality assessment method.

[0077] In the several embodiments provided in this application, any function, if implemented as a software functional module / unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the technical solution of this application can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or other electronic device) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing computer program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0078] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, the embodiments of this application are not directed to any particular programming language. It should be understood that the content of this application described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of this application.

[0079] It should be noted that the above embodiments are illustrative of this application and not restrictive, and those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In claims enumerating several means, several units or modules of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

[0080] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method of quality assessment of a three-dimensional scene reconstruction, characterized in that, The method comprises: collecting multiple images of a real three-dimensional scene, wherein the real three-dimensional scene is arranged with a graphic code, and the multiple images include an image containing the graphic code; reconstructing a virtual three-dimensional scene corresponding to the real three-dimensional scene by using the multiple images, wherein the virtual three-dimensional scene includes a virtual graphic code corresponding to the graphic code; determining the virtual graphic code in the virtual three-dimensional scene; scanning and identifying the virtual graphic code to obtain an identification result of the virtual graphic code; determining the quality of the virtual three-dimensional scene according to the identification result; wherein the real three-dimensional scene is arranged with multiple graphic codes, the virtual three-dimensional scene includes multiple virtual graphic codes corresponding to the multiple graphic codes, and the identification result includes virtual graphic code identification success and virtual graphic code identification failure; the determination of the quality of the virtual three-dimensional scene according to the identification result comprises: determining a first number of identification results of virtual graphic code identification success in multiple identification results obtained by scanning and identifying multiple virtual graphic codes; determining a ratio of the first number to a second number, wherein the second number is the number of virtual graphic codes included in the virtual three-dimensional scene; determining a target ratio interval to which the ratio belongs from multiple preset ratio intervals; determining a target quality level corresponding to the target ratio interval as the quality of the virtual three-dimensional scene according to a preset corresponding relationship between multiple ratio intervals and multiple quality levels.

2. The method of claim 1, wherein, the determination of the quality of the virtual three-dimensional scene according to the identification result comprises: if the identification result is virtual graphic code identification success, the quality of the virtual three-dimensional scene is determined as a first level; if the identification result is virtual graphic code identification failure, the quality of the virtual three-dimensional scene is determined as a second level, wherein the first level is better than the second level.

3. The method of claim 1, wherein, The multiple graphic codes are arranged at multiple positions in the real three-dimensional scene and have multiple orientations.

4. The method of claim 3, wherein, The multiple positions include the ground, the wall and the top.

5. The method according to claim 3 or 4, characterized in that, The sizes of the multiple graphic codes are different; the virtual three-dimensional scene includes multiple virtual graphic codes of different sizes corresponding to the multiple graphic codes, and virtual graphic codes of different sizes correspond to different scores; the identification result includes virtual graphic code identification success and virtual graphic code identification failure; the determination of the quality of the virtual three-dimensional scene according to the identification result comprises: determining a target virtual graphic code whose identification result is virtual graphic code identification success according to the identification result; determining a sum value of scores corresponding to all the target virtual graphic codes to obtain a total score; determining a target score interval to which the total score belongs from multiple preset score intervals; determining a target quality level corresponding to the target score interval as the quality of the virtual three-dimensional scene according to a preset corresponding relationship between multiple score intervals and multiple quality levels.

6. The method of claim 5, wherein, Among the multiple graphic codes, the color of the smallest graphic code is different from the colors of other graphic codes, and / or The smallest size graphical code in the real three-dimensional scene is provided with an identifier.

7. The method of any one of claims 1-3, wherein, The graphical code is a two-dimensional code.

8. A three-dimensional scene reconstruction quality assessment device comprising a memory, a processor and a computer program stored on the memory, characterized in that, The processor executes the computer program to implement the three-dimensional scene reconstruction quality evaluation method of any one of claims 1-7.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the three-dimensional scene reconstruction quality evaluation method of any one of claims 1-7.

Images