Evaluation Method, Device and Terminal Device for Compressed Neural Radiance Field Method

Abstract

This application is applicable to the field of computer application technology, and provides an evaluation method for a compressed neural radiance field method, including: collecting data of a preset scene, obtaining the neural radiance field corresponding to the preset scene, then using a preset compressed neural radiance field method to compress the neural radiance field to obtain a compressed neural radiance field, then calculating the redundancy of the compressed neural radiance field according to the compressed neural radiance field, and finally evaluating the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field, thereby improving the accuracy and generalization of the evaluation of the compressed neural radiance field method.

Description

Technical Field

[0001] This application belongs to the field of computer application technologies, and particularly relates to an evaluation method, device, and terminal device for a compressed neural radiance field method. Background Art

[0002] Neural radiance fields have been developing vigorously in recent years. It is a type of learnable model that constructs three-dimensional scene representations from multiple camera observations. As a novel view synthesis and three-dimensional reconstruction method, neural radiance field models have been widely applied in fields such as robotics, urban maps, autonomous navigation, and virtual reality.

[0003] In related technologies, due to the overly large selection space for the parameters of the neural radiance field model, the existing evaluation criteria for neural radiance fields are not applicable to compressed neural radiance fields, thus affecting the accuracy and generalization of the evaluation of compressed neural radiance field methods. Summary of the Invention

[0004] Embodiments of this application provide an evaluation method, device, and terminal device for a compressed neural radiance field method, which can solve the problem that the overly large selection space for general neural radiance field parameters leads to the inapplicability of existing neural radiance field evaluation criteria to compressed neural radiance fields, ultimately reducing the accuracy and generalization of the compressed neural radiance field evaluation method.

[0005] In a first aspect, embodiments of this application provide an evaluation method for a compressed neural radiance field method, including: collecting data of a preset scene to obtain the neural radiance field corresponding to the preset scene; using the preset compressed neural radiance field method to compress the neural radiance field to obtain the compressed neural radiance field; calculating the redundancy of the compressed neural radiance field according to the compressed neural radiance field; and evaluating the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field.

[0006] In a possible implementation manner of the first aspect, before calculating the redundancy of the compressed neural radiance field according to the compressed neural radiance field, it includes:

[0007] Geometrically transform the compressed neural radiance field;

[0008] Copy and superimpose the geometrically transformed compressed neural radiance field according to a preset complexity multiple to generate a mixed scene with the preset complexity multiple, where the complexity multiple is a positive integer greater than 1;

[0009] Collect data according to the mixed scene with the preset complexity multiple to obtain sampling data corresponding to the mixed scene with the preset complexity multiple.

[0010] Optionally, in another possible implementation of the first aspect, the above-mentioned data acquisition according to the mixed scenario of the preset complexity multiple to obtain the sampling data corresponding to the mixed scenario of the preset complexity multiple includes:

[0011] Preset the sampling range for data acquisition of the mixed scenario of the preset complexity multiple, where the sampling range is composed of the upper plane of the mixed scenario of the preset complexity multiple and the surrounding surfaces of the mixed scenario of the preset complexity multiple;

[0012] Uniformly set sampling devices on the upper plane of the mixed scenario of the preset complexity multiple, and adjust the viewing angle of the sampling devices to face the center of the object closest to the sampling devices in the mixed scenario of the preset complexity multiple to obtain the sampling data of the upper plane of the mixed scenario of the preset complexity multiple;

[0013] Uniformly set sampling devices on the surrounding surfaces of the mixed scenario of the preset complexity multiple, and adjust the viewing angle of the sampling devices to face the center of the mixed scenario of the preset complexity multiple to obtain the sampling data of the surrounding surfaces of the mixed scenario of the preset complexity multiple;

[0014] According to the ratio of the area of the upper plane of the mixed scenario of the preset complexity multiple to the area of the surrounding surfaces of the mixed scenario of the preset complexity multiple, set the ratio of the number of sampling data of the upper plane of the mixed scenario of the preset complexity multiple to the number of sampling data of the surrounding surfaces of the mixed scenario of the preset complexity multiple.

[0015] Optionally, in another possible implementation of the first aspect, the above-mentioned calculation of the redundancy of the compressed neural radiance field according to the compressed neural radiance field includes:

[0016] Acquire the first peak signal-to-noise ratio corresponding to the neural radiance field according to the data of the preset scene collected;

[0017] Acquire the second peak signal-to-noise ratio corresponding to the mixed scenario of the preset complexity multiple according to the sampling data corresponding to the mixed scenario of the preset complexity multiple;

[0018] Generate a rectangular coordinate system, with the abscissa being the preset complexity multiple and the ordinate being the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio;

[0019] Generate a curve in the rectangular coordinate system according to the preset complexity multiple, and calculate the enclosed area between the curve and the X-axis and Y-axis of the rectangular coordinate system to determine the redundancy corresponding to the compressed neural radiance field.

[0020] Optionally, in another possible implementation of the first aspect, the above-mentioned evaluation of the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field includes:

[0021] Judge the quality of the preset compressed neural radiance field method according to the redundancy of the corresponding compressed neural radiance field. Among them, the smaller the redundancy of the compressed neural radiance field, the better the compressed neural radiance field method.

[0022] In a second aspect, an evaluation device for a compressed neural radiance field method provided by an embodiment of the present application includes: a first acquisition module, configured to collect data of a preset scene and obtain a neural radiance field corresponding to the preset scene; a second acquisition module, configured to use a preset compressed neural radiance field method to compress the neural radiance field and obtain a compressed neural radiance field; a first calculation module, configured to calculate the redundancy of the compressed neural radiance field according to the compressed neural radiance field; a first evaluation module, configured to evaluate the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field.

[0023] In a possible implementation manner of the second aspect, the above device includes:

[0024] A geometric transformation module, configured to geometrically transform the compressed neural radiance field;

[0025] A first generation module, configured to copy and superimpose the geometrically transformed compressed neural radiance field according to a preset complexity multiple to generate a mixed scene with the preset complexity multiple, where the complexity multiple is a positive integer greater than 1;

[0026] A third acquisition module, configured to collect data according to the mixed scene with the preset complexity multiple to obtain sampling data corresponding to the mixed scene with the preset complexity multiple.

[0027] Optionally, in another possible implementation manner of the second aspect, the above third acquisition module includes:

[0028] A first preset unit, configured to preset a sampling range for collecting data of the mixed scene with the preset complexity multiple, where the sampling range is composed of the upper plane of the mixed scene with the preset complexity multiple and the surrounding surfaces of the mixed scene with the preset complexity multiple;

[0029] A first acquisition unit, configured to uniformly set sampling devices on the upper plane of the mixed scene with the preset complexity multiple, and adjust the viewing angle of the sampling devices to face the center of the object closest to the sampling devices in the mixed scene with the preset complexity multiple to obtain sampling data of the upper plane of the mixed scene with the preset complexity multiple;

[0030] A second acquisition unit, configured to uniformly set the sampling devices on the surrounding surfaces of the mixed scene with the preset complexity multiple, and adjust the viewing angle of the sampling devices to face the center of the mixed scene with the preset complexity multiple to obtain sampling data of the surrounding surfaces of the mixed scene with the preset complexity multiple;

[0031] A first setting unit, configured to set a ratio of the number of sampling data of the upper plane of a mixed scenario with a preset complexity multiple to the number of sampling data of the surrounding surfaces of the mixed scenario with the preset complexity multiple according to a ratio of the area of the upper plane of the mixed scenario with the preset complexity multiple to the area of the surrounding surfaces of the mixed scenario with the preset complexity multiple.

[0032] Optionally, in another possible implementation manner of the second aspect, the above first calculation module includes:

[0033] A third obtaining unit, configured to obtain a first peak signal-to-noise ratio corresponding to a neural radiance field according to data of a preset scenario collected;

[0034] A fourth obtaining unit, configured to obtain a second peak signal-to-noise ratio corresponding to a mixed scenario with a preset complexity multiple according to sampling data corresponding to the mixed scenario with the preset complexity multiple;

[0035] A first generating unit, configured to generate a rectangular coordinate system, where the abscissa is the preset complexity multiple and the ordinate is the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio;

[0036] A first determining unit, configured to generate a curve in the rectangular coordinate system according to the preset complexity multiple, and calculate an enclosed area of the curve with the X-axis and Y-axis of the rectangular coordinate system to determine the redundancy corresponding to the compressed neural radiance field.

[0037] Optionally, in another possible implementation manner of the second aspect, the above device further includes:

[0038] A first judging module, configured to judge the pros and cons of a preset method for compressing a neural radiance field according to the magnitude of the redundancy corresponding to the compressed neural radiance field, where the smaller the redundancy corresponding to the compressed neural radiance field, the better the method for compressing the neural radiance field.

[0039] In a third aspect, an embodiment of the present application provides a terminal device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the evaluation method of the method for compressing a neural radiance field as described above when executing the computer program.

[0040] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, characterized in that the computer program implements the evaluation method of the method for compressing a neural radiance field as described above when being executed by a processor.

[0041] In a fifth aspect, an embodiment of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device is caused to execute the evaluation method of the compressed neural radiance field method described in any one of the above first aspects.

[0042] The beneficial effects of the embodiments of the present application compared with the prior art are as follows: By evaluating the redundancy of the compressed neural radiance field, the advantages and disadvantages of the compressed neural radiance field method are judged, thereby improving the accuracy and generalization of the evaluation of the compressed neural radiance field method. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 is a schematic flowchart of an evaluation method for a compressed neural radiance field method provided by an embodiment of the present application;

[0045] Figure 2 is a schematic flowchart of an evaluation method for a compressed neural radiance field method provided by another embodiment of the present application;

[0046] Figure 3 is a schematic diagram of the sampling range of a mixed scene provided by another embodiment of the present application;

[0047] Figure 4 is a schematic flowchart of an evaluation method for a compressed neural radiance field method provided by another embodiment of the present application;

[0048] Figure 5 is a schematic diagram of the peak signal-to-noise ratio curve of a mixed scene provided by another embodiment of the present application;

[0049] Figure 6 is a schematic structural diagram of an evaluation device for a compressed neural radiance field method provided by an embodiment of the present application;

[0050] Figure 7 is a schematic structural diagram of a terminal device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] In the following description, specific details such as specific system architectures, technologies, etc. are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0052] It should be understood that when used in the specification and the appended claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0053] It should also be understood that the term "and / or" used in the specification and the appended claims of the present application refers to any combination and all possible combinations of one or more of the related listed items, and includes these combinations.

[0054] As used in the specification and the appended claims of the present application, the term "if" can be interpreted as "when", "once", "in response to determining", or "in response to detecting" depending on the context. Similarly, the phrase "if determined" or "if [the described condition or event] is detected" can be interpreted as meaning "once determined", "in response to determining", "once [the described condition or event] is detected", or "in response to detecting [the described condition or event]" depending on the context.

[0055] In addition, in the description of the specification and the appended claims of the present application, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0056] The reference to "one embodiment" or "some embodiments" etc. described in the specification of the present application means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in another way. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in another way.

[0057] The following provides a detailed description of an evaluation method, device, terminal device, storage medium, and computer program for a compressed neural radiance field method with reference to the accompanying drawings.

[0058] Figure 1 The flowchart shows an evaluation method for a compressed neural radiance field method provided by an embodiment of the present application.

[0059] Step 101: Collect data of a preset scene and obtain the neural radiance field corresponding to the preset scene.

[0060] It should be noted that the evaluation method of the compressed neural radiance field method in the embodiment of the present application can be executed by the evaluation device of the compressed neural radiance field method in the embodiment of the present application. The evaluation device of the compressed neural radiance field method in the embodiment of the present application can be configured in any terminal device to execute the evaluation method of the compressed neural radiance field method in the embodiment of the present application.

[0061] Among them, the data of the preset scene can refer to the preset five-dimensional vector for generating the neural radiance field corresponding to the target scene.

[0062] It should be noted that the data of the preset scene, that is, the five-dimensional vector, can refer to the three-dimensional coordinates of the preset scene or object space points and the perspective data of the observation direction of the sampling device, which can be expressed as (x, y, z, θ, φ), where x, y, z are the three-dimensional coordinates of the space points, and θ, φ are the perspective data of the observation direction of the sampling device.

[0063] Among them, the neural radiance field can refer to a model that can output a four-dimensional vector corresponding to the data of the preset scene by collecting the data of the preset scene.

[0064] It should be noted that after collecting the data of the preset scene and passing it through the neural radiance field, the output four-dimensional vector refers to the spatial point color and opacity corresponding to the data of the preset scene, which can be expressed as (r, g, b, σ). Among them, the output r, g, b are the three channels of the spatial point color, and σ is the opacity of this spatial point.

[0065] In the embodiment of the present application, first collect the data of the preset scene, that is, the five-dimensional vector, input it into the neural radiance field, and a four-dimensional vector corresponding to the data of the preset scene can be output. The above process can be expressed as (x, y, z, θ, φ) → (r, g, b, σ), and finally the neural radiance field corresponding to the preset scene data is obtained.

[0066] Step 102: Compress the neural radiance field using a preset compressed neural radiance field method to obtain a compressed neural radiance field.

[0067] Among them, the compressed neural radiance field method can refer to a model compression method that compresses the neural radiance field to make it lighter, while the effect of the compressed neural radiance field does not decrease too much.

[0068] For example, currently, with the rapid improvement of computing power, existing models are getting larger and their networks and structures are becoming more and more complex. The advantage is that it brings an improvement in model accuracy. However, the disadvantage is that it increases the computational complexity. Therefore, model compression methods such as knowledge distillation have emerged, which can avoid the computational complexity caused by the model being too large and at the same time ensure that the performance of the compressed model does not degrade too much. In knowledge distillation, first, a large and complex model, namely the Teacher model, is required. At this time, it is compressed to obtain a smaller model called the student model. During the distillation process, the knowledge learned in the Teacher model is transferred to the student model, so that the student model is lightweight while having the same performance as the Teacher model.

[0069] Among them, the compressed neural radiance field can refer to the neural radiance field compressed by the compressed neural radiance field method.

[0070] In the embodiments of the present application, since the compressed neural radiance field method needs to be evaluated, first, the compressed neural radiance field method is preset, and then the neural radiance field is compressed by the compressed neural radiance field method to obtain the compressed neural radiance field to be evaluated.

[0071] Step 103, calculate the redundancy of the compressed neural radiance field according to the compressed neural radiance field.

[0072] Among them, the redundancy of the compressed neural radiance field can refer to a value that can represent the information expression ability of the compressed neural radiance field.

[0073] In the embodiments of the present application, in order to judge the superiority and inferiority of the compressed neural radiance field method, the redundancy is used for evaluation. According to the obtained compressed neural radiance field above, it is calculated to obtain the redundancy corresponding to the compressed neural radiance field.

[0074] Step 104, evaluate the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field.

[0075] In the embodiments of the present application, the compressed neural radiance field is calculated to obtain the redundancy corresponding to the compressed neural radiance field, and then the compressed neural radiance field method is evaluated according to the magnitude of the redundancy. When the redundancy corresponding to the compressed neural radiance field is lower, it indicates that the preset compressed neural radiance field method is a more preferred compressed neural radiance field method.

[0076] The evaluation method of the compressed neural radiance field method provided by this application first collects data of a preset scene to obtain the neural radiance field corresponding to the preset scene, then compresses the neural radiance field using the preset compressed neural radiance field method to obtain the compressed neural radiance field, then calculates the redundancy of the compressed neural radiance field based on the compressed neural radiance field, and finally evaluates the preset compressed neural radiance field method based on the redundancy of the compressed neural radiance field. Thus, the compressed neural radiance field is obtained by compressing the neural radiance field using the preset compressed neural radiance field method, then the redundancy of the compressed neural radiance field is calculated, and the preset compressed neural radiance field method is evaluated using the redundancy to judge the advantages and disadvantages of the compressed neural radiance field method, thereby improving the accuracy and generalization of the evaluation of the compressed neural radiance field method.

[0077] In a possible implementation form of this application, since calculating the redundancy of the compressed neural radiance field needs to be performed under mixed scenes of different complexities, the compressed neural radiance fields can be superimposed to form mixed scenes of different complexities for calculating the redundancy, so as to further improve the accuracy of calculating the redundancy of the compressed neural radiance field.

[0078] The following combines Figure 2 , and further illustrates the evaluation method of the compressed neural radiance field method provided by the embodiments of this application.

[0079] Figure 2 The flowchart shows another evaluation method of the compressed neural radiance field method provided by the embodiments of this application.

[0080] Step 201, collect data of a preset scene to obtain the neural radiance field corresponding to the preset scene.

[0081] Step 202, compress the neural radiance field using the preset compressed neural radiance field method to obtain the compressed neural radiance field.

[0082] For the specific implementation process and principle of the above steps 201 - 202, reference can be made to the detailed description of the above embodiments, and details are not described here again.

[0083] Step 203, geometric transformation compresses the neural radiance field.

[0084] Among them, the geometric transformation may refer to transforming the five-dimensional coordinates composed of the three-dimensional coordinates of the spatial points in the compressed neural radiance field and the viewing angle data of the observation direction of the sampling device using a coordinate transformation matrix.

[0085] It should be noted that the compressed neural radiance field is denoted as f, and using L T Geometric transformation compresses the neural radiance field f, denoted as L Tf(x, u) = f(Tx, Tu), where T is the coordinate transformation matrix, and x and u are the three-dimensional coordinates of the spatial points in the compressed neural radiance field and the perspective data of the observation directions of the sampling devices, respectively.

[0086] In the embodiments of the present application, since the compressed neural radiance field is a field, it is necessary to perform geometric transformation on the compressed neural radiance field before superimposing them on each other. Therefore, it is necessary to perform geometric transformation on the compressed neural radiance field first.

[0087] Step 204: Copy and superimpose the geometrically transformed compressed neural radiance field according to a preset complexity multiple to generate a mixed scene with the preset complexity multiple.

[0088] Among them, the preset complexity multiple may refer to the multiple of copying and superimposing the geometrically transformed compressed neural radiance field, and the complexity multiple is a positive integer greater than 1.

[0089] For example, the geometrically transformed compressed neural radiance field can be copied and superimposed 4 times, 6 times, or 8 times, and the above 4 times, 6 times, and 8 times can all be the preset complexity multiples.

[0090] Among them, the mixed scene with the preset complexity multiple may refer to a new scene generated after the geometrically transformed compressed neural radiance field is copied and superimposed according to the preset complexity multiple.

[0091] For example, since the geometric transformation of the compressed neural radiance field f is denoted as L T f(x, u) = f(Tx, Tu), so when the complexity multiple is 2, the mixed scene can be expressed as L T f + L T1 f, similarly, when the complexity multiple is 4, the mixed scene can be expressed as L T2 f + L T1 f + L T2 f + L T3 f + L T4 f, when the complexity multiple is 8, the mixed scene can be expressed as L T1 f + L T2 f + L T3 f + L T4 f + L T5 f + L T6 f + L T7 f + L T8 f.

[0092] In the embodiments of the present application, in order to better test the information-carrying capacity of the compressed neural radiance field under the condition of unchanged parameters and the redundancy corresponding to the precisely compressed neural radiance field, the compressed neural radiance field is replicated and superimposed according to a preset complexity multiple, and finally a mixed scene with the preset complexity multiple is generated.

[0093] Step 205: Collect data according to the mixed scene with the preset complexity multiple to obtain sampling data corresponding to the mixed scene with the preset complexity multiple.

[0094] In the embodiments of the present application, a sampling device is used to collect data from the mixed scene with the preset complexity multiple to obtain the data in the mixed scene with the preset complexity multiple.

[0095] Furthermore, in order to make the sampling data corresponding to the mixed scene with the preset complexity multiple obtained by the sampling device more accurate and comprehensive, the sampling rules of the sampling device in the mixed scene can be set. That is, in another possible implementation manner of the embodiments of the present application, the above step 205 may further include:

[0096] Preset the sampling range for collecting data from the mixed scene with the preset complexity multiple, where the sampling range is composed of the upper plane of the mixed scene with the preset complexity multiple and the surrounding surfaces of the mixed scene with the preset complexity multiple;

[0097] In the embodiments of the present application, the sampling range of the mixed scene with the preset complexity multiple can be preset in advance. The sampling range is composed of the enclosed surfaces above and around the mixed scene. For example, Figure 3 As shown, the sampling range is composed of the circular plane above the mixed scene and the annular curved surface around it.

[0098] Uniformly set sampling devices on the upper plane of the mixed scene with the preset complexity multiple, and adjust the viewing angle of the sampling devices to face the center of the object closest to the sampling devices in the mixed scene with the preset complexity multiple to obtain the sampling data of the upper plane of the mixed scene with the preset complexity multiple;

[0099] In the embodiments of the present application, in order to improve the accuracy of the obtained sampling data, sampling devices are uniformly set on the upper plane of the mixed scene with the preset complexity multiple. At the same time, each sampling device is set to face the object in the mixed scene closest to it. For example, Figure 3 As shown, the sampling devices on the circular plane above the mixed scene select different objects to face according to their distances from different objects in the mixed scene. For example, the sampling device on the left of the circular plane above the mixed scene faces the object on the left in the mixed scene, and the sampling device on the right faces the object on the right in the mixed scene.

[0100] Uniformly arrange sampling devices on the peripheral surfaces of a mixed scenario with a preset complexity multiple, and adjust the viewing angles of the sampling devices to face the center of the mixed scenario with the preset complexity multiple, so as to obtain sampling data of the peripheral surfaces of the mixed scenario with the preset complexity multiple;

[0101] In the embodiment of the present application, in order to obtain sampling data with high accuracy, sampling devices are uniformly arranged on the peripheral surfaces of a mixed scenario with a preset complexity multiple, and the above-mentioned sampling devices all face the center of the mixed scenario. For example, as Figure 3 shown, sampling devices are uniformly arranged on the peripheral annular curved surface of the mixed scenario, and at the same time, the above-mentioned sampling devices all face the center position of the mixed scenario.

[0102] Set the ratio of the number of sampling data of the upper plane of the mixed scenario with the preset complexity multiple to the number of sampling data of the peripheral surfaces of the mixed scenario with the preset complexity multiple according to the ratio of the area of the upper plane of the mixed scenario with the preset complexity multiple to the area of the peripheral surfaces of the mixed scenario with the preset complexity multiple.

[0103] In the embodiment of the present application, first obtain the ratio of the area of the upper plane of the mixed scenario with the preset complexity multiple to the area of the peripheral surfaces of the mixed scenario, and then use the obtained ratio as the ratio of the number of sampling data of the upper plane of the mixed scenario with the preset complexity multiple to the number of sampling data of the peripheral surfaces of the mixed scenario. For example, as Figure 3 shown, first obtain the ratio of the area of the circular plane above the mixed scenario to the area of the peripheral annular curved surface of the mixed scenario, and then use the above ratio as the ratio of the number of sampling data of the circular plane above the mixed scenario to the number of sampling data of the peripheral annular curved surface of the mixed scenario.

[0104] Step 206, calculate the redundancy of the compressed neural radiance field according to the compressed neural radiance field.

[0105] Step 207, evaluate the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field.

[0106] For the specific implementation process and principle of the above steps 206-207, reference can be made to the detailed description of the above embodiments, and details are not described herein again.

[0107] The evaluation method of the compressed neural radiance field method provided in this application first collects data of a preset scene, obtains the neural radiance field corresponding to the preset scene, compresses the neural radiance field using the preset compressed neural radiance field method to obtain a compressed neural radiance field, then geometrically transforms the compressed neural radiance field, and then duplicates and superimposes the geometrically transformed compressed neural radiance field according to a preset complexity multiple to generate a mixed scene with the preset complexity multiple. Then, data is collected according to the mixed scene with the preset complexity multiple to obtain sampling data corresponding to the mixed scene with the preset complexity multiple. Further, the redundancy of the compressed neural radiance field is calculated based on the compressed neural radiance field, and finally, the preset compressed neural radiance field method is evaluated according to the redundancy of the compressed neural radiance field. Thus, by superimposing the complexity multiple on the compressed neural radiance field without changing the parameters of the compressed neural radiance field to form a corresponding mixed scene, and at the same time using the preset sampling rule for the mixed scene, the data of the mixed scene composed of the compressed neural radiance field collected by the acquisition device is more accurate and has higher generalization ability.

[0108] In a possible implementation form of this application, the redundancy of the compressed neural radiance field can be calculated by obtaining the peak signal-to-noise ratio between the neural radiance field and the mixed scene formed by superimposing the compressed neural radiance fields.

[0109] The following combines Figure 4 , and further illustrates the evaluation method of the compressed neural radiance field method provided in the embodiments of this application.

[0110] Step 401: Collect data of a preset scene and obtain the neural radiance field corresponding to the preset scene.

[0111] Step 402: Compress the neural radiance field using the preset compressed neural radiance field method to obtain a compressed neural radiance field.

[0112] For the specific implementation process and principle of the above steps 401-402, reference can be made to the detailed description of the above embodiments, which will not be elaborated here.

[0113] Step 403: Obtain the first peak signal-to-noise ratio corresponding to the neural radiance field according to the data of the collected preset scene.

[0114] Among them, the peak signal-to-noise ratio can be an evaluation criterion for evaluating the neural radiance field.

[0115] It should be noted that the difference between the picture predicted by the neural radiance field and the real picture can be described using PSNR (Peak Signal-to-Noise Ratio), that is:

[0116]

[0117]

[0118] Among them, MSE represents the true value minus the predicted value, then squared and summed and averaged, that is, the pixel values at the same position in each of the two images of the true image and the predicted image are subtracted, squared, summed, and then averaged. MSE represents the average of the differences in pixel values at each position of the two images. The larger its value, the lower the similarity between the two images. Since MSE is in the denominator of the PSNR expression, the larger the value of PSNR, the higher the similarity between the two images.

[0119] In the embodiment of the present application, first, data of a preset scene is collected, and a corresponding predicted image is generated through a neural radiance field, and then the first peak signal-to-noise ratio corresponding to the neural radiance field is obtained by using the true image of the preset scene and the predicted image generated from the data of the preset scene.

[0120] Step 404, obtain the second peak signal-to-noise ratio corresponding to the mixed scene with a preset complexity multiple according to the sampling data corresponding to the mixed scene with a preset complexity multiple.

[0121] In the embodiment of the present application, first, sampling data corresponding to the mixed scene with a preset complexity multiple is collected, and a corresponding predicted image is generated through the compressed neural radiance field corresponding to the above mixed scene, and then the second peak signal-to-noise ratio corresponding to the mixed scene is obtained by using the true image of the mixed scene and the predicted image generated from the sampling data corresponding to the mixed scene.

[0122] Step 405, generate a rectangular coordinate system, where the abscissa is the preset complexity multiple and the ordinate is the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio.

[0123] In the embodiment of the present application, in order to numerically represent the redundancy of the compressed neural radiance field, a rectangular coordinate system is generated, where the abscissa is the preset complexity multiple and the ordinate is the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio.

[0124] For example, let the PSNR value of the neural radiance field be P base , and the preset complexity multiple be N (N is a positive integer greater than 1), then the PSNR value of the mixed scene with N times complexity is P N , then in the rectangular coordinate system, the abscissa is N and the ordinate is As Figure 5 shown, when the complexity multiples are 2, 4, and 8, their corresponding coordinates in the rectangular coordinate system are

[0125] Step 406: Generate a curve in the rectangular coordinate system according to a preset complexity multiple, and calculate the enclosed area between the curve and the X-axis and Y-axis of the rectangular coordinate system to determine the redundancy corresponding to the compressed neural radiance field.

[0126] In the embodiment of the present application, corresponding coordinate points are generated according to a preset complexity multiple, and then the coordinates in the rectangular coordinate system corresponding to their respective complexity multiples are connected. The value of the enclosed area formed by the curve connecting the above coordinates and the X-axis and Y-axis in the rectangular coordinate system is used as the redundancy corresponding to the compressed neural radiance field.

[0127] For example, as Figure 5 shown, when the complexity multiples are 2, 4, and 8, the corresponding coordinates generated are Connect the above three coordinates with a curve. The value of the enclosed area formed by the above straight line and the X-axis and Y-axis is the redundancy corresponding to the compressed neural radiance field.

[0128] Step 407: Judge the pros and cons of the preset method for compressing the neural radiance field according to the size of the redundancy corresponding to the compressed neural radiance field.

[0129] In the embodiment of the present application, since the redundancy can clearly distinguish the differences in information expression of different models, the redundancy is used to evaluate the method for compressing the neural radiance field. Among them, the smaller the redundancy corresponding to the compressed neural radiance field, the better the method for compressing the neural radiance field.

[0130] The evaluation method of the compressed neural radiance field method provided by this application first collects data of a preset scene to obtain the neural radiance field corresponding to the preset scene, then uses the preset compressed neural radiance field method to compress the neural radiance field to obtain the compressed neural radiance field. Next, according to the data collected from the preset scene, the first peak signal-to-noise ratio corresponding to the neural radiance field is obtained. Then, according to the sampling data of the mixed scene with a preset complexity multiple, the second peak signal-to-noise ratio corresponding to the mixed scene with the preset complexity multiple is obtained. Further, a rectangular coordinate system is generated, with the abscissa being the preset complexity multiple and the ordinate being the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio. Then, a curve is generated in the rectangular coordinate system according to the preset complexity multiple, and the enclosed area between the curve and the X-axis and Y-axis of the rectangular coordinate system is calculated to determine the redundancy corresponding to the compressed neural radiance field. Finally, the pros and cons of the preset compressed neural radiance field method are judged according to the magnitude of the redundancy corresponding to the compressed neural radiance field. Thus, by calculating the peak signal-to-noise ratios corresponding to the neural radiance field and the compressed neural radiance field and establishing a rectangular coordinate system, the redundancy of the compressed neural radiance field can be represented in the rectangular coordinate system by the peak signal-to-noise ratios corresponding to the neural radiance field and the compressed neural radiance field, so as to calculate the redundancy corresponding to the compressed neural radiance field more intuitively and simply, and then use the redundancy to evaluate the method of the compressed neural radiance field.

[0131] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of this application.

[0132] Corresponding to the evaluation method of the compressed neural radiance field method in the above embodiments, Figure 6 The structural block diagram of the evaluation device for the compressed neural radiance field method provided by the embodiment of this application is shown. For the convenience of description, only the parts related to the embodiment of this application are shown.

[0133] Referring to Figure 6 , the device 60 includes:

[0134] The first acquisition module 61 is used to collect data of a preset scene and obtain the neural radiance field corresponding to the preset scene;

[0135] The second acquisition module 62 is used to compress the neural radiance field by using the preset compressed neural radiance field method to obtain the compressed neural radiance field;

[0136] The first calculation module 63 is used to calculate the redundancy of the compressed neural radiance field according to the compressed neural radiance field;

[0137] The first evaluation module 64 is used to evaluate the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field.

[0138] During actual use, the evaluation device for the compressed neural radiance field method provided by the embodiments of the present application can be configured in any terminal device to execute the evaluation method of the aforementioned compressed neural radiance field method.

[0139] The evaluation device for the compressed neural radiance field method provided by the present application first collects data of a preset scene, obtains the neural radiance field corresponding to the preset scene, then compresses the neural radiance field using a preset compressed neural radiance field method to obtain a compressed neural radiance field, then calculates the redundancy of the compressed neural radiance field based on the compressed neural radiance field, and finally evaluates the preset compressed neural radiance field method based on the redundancy of the compressed neural radiance field. Thus, by compressing the neural radiance field using a preset compressed neural radiance field method to obtain a compressed neural radiance field, then calculating the redundancy of the compressed neural radiance field, and using the redundancy to evaluate the preset compressed neural radiance field method, the superiority and inferiority of the compressed neural radiance field method are judged, thereby improving the accuracy and generalization of the evaluation of the compressed neural radiance field method.

[0140] In a possible implementation manner of the embodiments of the present application, the above-mentioned device 60 includes:

[0141] A geometric transformation module for geometrically transforming the compressed neural radiance field;

[0142] A first generation module for replicating and superimposing the geometrically transformed compressed neural radiance field according to a preset complexity multiple to generate a mixed scene with the preset complexity multiple, where the complexity multiple is a positive integer greater than 1;

[0143] A third acquisition module for collecting data according to the mixed scene with the preset complexity multiple to obtain sampling data corresponding to the mixed scene with the preset complexity multiple.

[0144] Further, in another possible implementation manner of the embodiments of the present application, the above-mentioned third acquisition module includes:

[0145] A first preset unit for presetting a sampling range for collecting data of the mixed scene with the preset complexity multiple, where the sampling range is composed of the upper plane of the mixed scene with the preset complexity multiple and the surrounding surfaces of the mixed scene with the preset complexity multiple;

[0146] A first acquisition unit for uniformly arranging sampling devices on the upper plane of the mixed scene with the preset complexity multiple, adjusting the viewing angle of the sampling devices to face the center of the object closest to the sampling devices in the mixed scene with the preset complexity multiple to obtain sampling data of the upper plane of the mixed scene with the preset complexity multiple;

[0147] A second acquisition unit, configured to evenly arrange the sampling device on the peripheral surfaces of a mixed scenario with a preset complexity multiple, and adjust the viewing angle of the sampling device to face the center of the mixed scenario with the preset complexity multiple, so as to acquire sampling data of the peripheral surfaces of the mixed scenario with the preset complexity multiple;

[0148] A first setting unit, configured to set the ratio of the number of sampling data of the upper plane of the mixed scenario with the preset complexity multiple to the number of sampling data of the peripheral surfaces of the mixed scenario with the preset complexity multiple according to the ratio of the area of the upper plane of the mixed scenario with the preset complexity multiple to the area of the peripheral surfaces of the mixed scenario with the preset complexity multiple.

[0149] Further, in another possible implementation manner of the embodiment of the present application, the above first calculation module includes:

[0150] A third acquisition unit, configured to acquire a first peak signal-to-noise ratio corresponding to a neural radiance field according to data of a preset scenario;

[0151] A fourth acquisition unit, configured to acquire a second peak signal-to-noise ratio corresponding to the mixed scenario with the preset complexity multiple according to the sampling data corresponding to the mixed scenario with the preset complexity multiple;

[0152] A first generation unit, configured to generate a rectangular coordinate system, where the abscissa is the preset complexity multiple and the ordinate is the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio;

[0153] A first determination unit, configured to generate a curve in the rectangular coordinate system according to the preset complexity multiple, and calculate the enclosed area of the curve with the X-axis and Y-axis of the rectangular coordinate system to determine the redundancy corresponding to the compressed neural radiance field.

[0154] Further, in another possible implementation manner of the embodiment of the present application, the above device 60 further includes:

[0155] A first judgment module, configured to judge the pros and cons of a preset method for compressing a neural radiance field according to the magnitude of the redundancy corresponding to the compressed neural radiance field, where the smaller the redundancy corresponding to the compressed neural radiance field, the better the method for compressing the neural radiance field.

[0156] It should be noted that the information interaction, execution process, etc. between the above devices / units, due to being based on the same concept as the method embodiment of the present application, for their specific functions and the technical effects brought, please refer to the method embodiment part specifically, and will not be elaborated here.

[0157] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of the functional units and modules are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiment and will not be elaborated here.

[0158] To implement the above embodiment, the present application also proposes a terminal device.

[0159] Figure 7 It is a schematic structural diagram of a terminal device according to an embodiment of the present application.

[0160] As Figure 7 shown, the above terminal device 200 includes:

[0161] A memory 210 and at least one processor 220, a bus 230 connecting different components (including the memory 210 and the processor 220), and the memory 210 stores a computer program. When the processor 220 executes the program, it implements the evaluation method of the compressed neural radiance field method described in the embodiment of the present application.

[0162] The bus 230 represents one or more of several types of bus structures, including a memory bus or a memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any bus structure in a variety of bus structures. For example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

[0163] The terminal device 200 typically includes a variety of electronically readable media. These media can be any available media that can be accessed by the terminal device 200, including volatile and non-volatile media, removable and non-removable media.

[0164] The memory 210 may also include a computer system readable medium in the form of volatile memory, such as random access memory (RAM) 240 and / or cache memory 250. The terminal device 200 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 260 may be used for reading and writing on non-removable, non-volatile magnetic media ( Figure 7 not shown, commonly referred to as a "hard disk drive"). Although Figure 7 not shown, a disk drive for reading and writing on a removable non-volatile disk (such as a "floppy disk"), and an optical disk drive for reading and writing on a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM or other optical media) may be provided. In these cases, each drive may be connected to the bus 230 through one or more data media interfaces. The memory 210 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present application.

[0165] A program / utility 280 having a set (at least one) of program modules 270 may be stored, for example, in the memory 210. Such program modules 270 include - but are not limited to - an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include the implementation of a network environment. The program modules 270 generally perform the functions and / or methods in the embodiments described in the present application.

[0166] The terminal device 200 may also communicate with one or more external devices 290 (such as a keyboard, a pointing device, a display 291, etc.), and may also communicate with one or more devices that enable a user to interact with the terminal device 200, and / or communicate with any device that enables the terminal device 200 to communicate with one or more other computing devices (such as a network card, a modem, etc.). Such communication may be carried out through an input / output (I / O) interface 292. Moreover, the terminal device 200 may also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN) and / or a public network, such as the Internet) through a network adapter 293. As shown in the figure, the network adapter 293 communicates with other modules of the terminal device 200 through the bus 230. It should be understood that although not shown in the figure, other hardware and / or software modules may be used in conjunction with the terminal device 200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0167] The processor 220 executes various functional applications and data processing by running the programs stored in the memory 210.

[0168] It should be noted that for the implementation process and technical principle of the terminal device in this embodiment, refer to the foregoing explanation of the evaluation method of the compressed neural radiance field method of the embodiments of the present application, which will not be elaborated here.

[0169] The embodiments of the present application also provide a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0170] The embodiments of the present application provide a computer program product, and when the computer program product runs on a terminal device, the terminal device is enabled to execute the steps in the above-mentioned method embodiments.

[0171] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-mentioned method embodiments of the present application, a computer program can be used to instruct relevant hardware to complete. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the photographing device / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disc, etc. In some jurisdictions, according to legislation and patent practice, the computer-readable medium cannot be an electrical carrier signal and a telecommunication signal.

[0172] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0173] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0174] In the embodiments provided in the present application, it should be understood that the disclosed device / terminal device and method can be implemented in other ways. For example, the device / terminal device embodiments described above are merely illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in electrical, mechanical or other forms.

[0175] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0176] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. An evaluation method for a method of compressing neural radiance fields, characterized in that , including: Collect data of a preset scene and obtain the neural radiance field corresponding to the preset scene; Compress the neural radiance field by using the preset compressed neural radiance field method to obtain the compressed neural radiance field; Calculate the redundancy of the compressed neural radiance field according to the compressed neural radiance field; Evaluate the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field; The calculating the redundancy of the compressed neural radiance field according to the compressed neural radiance field includes: Obtain the first peak signal-to-noise ratio corresponding to the neural radiance field according to the data collected from the preset scene; Obtain the second peak signal-to-noise ratio corresponding to the mixed scene with a preset complexity multiple according to the sampling data of the mixed scene with the preset complexity multiple; Generate a rectangular coordinate system, with the abscissa being the preset complexity multiple and the ordinate being the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio; Generate a curve in the rectangular coordinate system according to the preset complexity multiple, and calculate the enclosed area between the curve and the X-axis and Y-axis of the rectangular coordinate system to determine the redundancy corresponding to the compressed neural radiance field.

2. The method according to claim 1, before calculating the redundancy of the compressed neural radiance field according to the compressed neural radiance field, includes: Perform geometric transformation on the compressed neural radiance field; Copy and superimpose the geometrically transformed compressed neural radiance field according to a preset complexity multiple to generate the mixed scene with the preset complexity multiple, where the complexity multiple is a positive integer greater than 1; Collect data according to the mixed scene with the preset complexity multiple to obtain the sampling data corresponding to the mixed scene with the preset complexity multiple.

3. The method according to claim 2, the collecting data according to the mixed scene with the preset complexity multiple to obtain the sampling data corresponding to the mixed scene with the preset complexity multiple includes: Preset the sampling range for collecting data of the mixed scene with the preset complexity multiple, where the sampling range is composed of the upper plane of the mixed scene with the preset complexity multiple and the surrounding surfaces of the mixed scene with the preset complexity multiple; Uniformly set sampling devices on the upper plane of the mixed scene with the preset complexity multiple, and adjust the viewing angle of the sampling devices to face the center of the object closest to the sampling devices in the mixed scene with the preset complexity multiple to obtain the sampling data of the upper plane of the mixed scene with the preset complexity multiple; Uniformly set the sampling devices on the surrounding surfaces of the mixed scene with the preset complexity multiple, and adjust the viewing angle of the sampling devices to face the center of the mixed scene with the preset complexity multiple to obtain the sampling data of the surrounding surfaces of the mixed scene with the preset complexity multiple; Set the ratio of the number of sampling data of the upper plane of the mixed scene with the preset complexity multiple to the number of sampling data of the surrounding surfaces of the mixed scene with the preset complexity multiple according to the ratio of the area of the upper plane of the mixed scene with the preset complexity multiple to the area of the surrounding surfaces of the mixed scene with the preset complexity multiple.

4. The method according to any one of claims 1-3, wherein the evaluation of the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field comprises: Judging the quality of the preset compressed neural radiance field method according to the magnitude of the redundancy corresponding to the compressed neural radiance field, wherein the smaller the redundancy corresponding to the compressed neural radiance field, the better the compressed neural radiance field method.

5. An evaluation device for a compressed neural radiance field method, characterized in that, Comprising: A first acquisition module, configured to collect data of a preset scene and obtain the neural radiance field corresponding to the preset scene; A second acquisition module, configured to compress the neural radiance field by using the preset compressed neural radiance field method to obtain the compressed neural radiance field; A first calculation module, configured to calculate the redundancy of the compressed neural radiance field according to the compressed neural radiance field; A first evaluation module, configured to evaluate the preset compressed neural radiance field method according to the redundancy of the compressed neural radiance field; A third acquisition unit, configured to obtain the first peak signal-to-noise ratio corresponding to the neural radiance field according to the data of the preset scene collected; A fourth acquisition unit, configured to obtain the second peak signal-to-noise ratio corresponding to the mixed scene with the preset complexity multiple according to the sampling data corresponding to the mixed scene with the preset complexity multiple; A first generation unit, configured to generate a rectangular coordinate system, with the abscissa being the preset complexity multiple and the ordinate being the ratio of the second peak signal-to-noise ratio to the first peak signal-to-noise ratio; A first determination unit, configured to generate a curve in the rectangular coordinate system according to the preset complexity multiple, and calculate the enclosed area of the curve with the X-axis and Y-axis of the rectangular coordinate system to determine the redundancy corresponding to the compressed neural radiance field.

6. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the method according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the method according to any one of claims 1 to 4 is implemented.

Images