CN112652070B - Three-dimensional model surface reduction method, device, equipment and medium - Google Patents

Abstract

The present invention relates to the technical field of three-dimensional models. The present invention discloses a three-dimensional model surface reduction method, device, equipment, and medium. The method includes: constructing three-dimensional model data corresponding to the three-dimensional file to be processed by acquiring the three-dimensional file to be processed ;Recognize the surface reduction of the 3D model data, and obtain the surface reduction result and the corresponding base denomination value of the surface reduction result; when the surface reduction result needs to be reduced, determine the sub-models in the 3D model data according to the base deduction face value The corresponding target surface value; according to all the target surface values corresponding to the sub-models, the corresponding surface reduction is performed on the 3D model data; the 3D model data after surface reduction is compressed into a 3D file for surface reduction and output. The present invention realizes automatic recognition of surface reduction results and reference denomination values, automatically outputs the target surface value, performs corresponding surface reduction according to the target surface value, and compresses the output without manual operation, and improves the efficiency of three-dimensional model surface reduction, improving the The quality of the reduced surface of the 3D model.

Description

Three-dimensional model surface reduction method, device, equipment and medium

technical field

The invention relates to the technical field of three-dimensional models, in particular to a surface reduction method, device, equipment and medium for a three-dimensional model.

Background technique

3D models have been widely used in various fields, such as household industry, construction industry, medical industry, film industry, video game industry, as well as in scientific research and engineering applications. 3D model objects are composed of triangular faces, and the rendering and production of 3D models have extremely high requirements on hardware systems and modeling software. Generally, the number of faces of a 3D model ranges from dozens to millions. Computers with low hardware configuration will experience different degrees of lag when processing slightly complex 3D models. The time for model rendering also varies with the complexity of the model and the size of the model. The number of faces changes, so the number of triangular faces of the 3D model is a key factor affecting the efficiency and frame rate of the model rendering.

At present, the existing technology mainly adopts the following two methods to reduce the surface of the 3D model: the first is to use the related surface reduction plug-ins officially provided by some mainstream modeling software to automatically reduce the surface to a certain extent; the second is to use the original model Contour, reconstruct a simplified model, or manually reduce the surface on the basis of the original model.

However, although the above two methods can reduce the surface of the 3D model, they have different degrees of defects. Specifically, the plug-ins used in method 1 generally need to be paid, and the results generated by the polygon reduction tool cannot meet special needs, and it is difficult to fine-tune the characteristics of the model, and the triangle surface will be different when it is reduced to a certain proportion. The degree of aliasing or even deformation; the second method needs to judge the effect of surface reduction with human eyes. Although this method is effective and the number of patches can be controlled artificially, it is time-consuming and labor-intensive, and the labor cost is extremely high, which cannot adapt to large-scale model processing. .

Therefore, it is necessary to provide a technical solution to solve the above technical problems.

Contents of the invention

The invention provides a three-dimensional model surface reduction method, device, computer equipment and storage medium, which realizes the automatic completion of the three-dimensional model surface reduction operation without manual operation, reduces labor costs, and improves the efficiency of three-dimensional model surface reduction. In addition, the effectiveness and no distortion of the surface reduction of the 3D model are guaranteed, and the quality of the surface reduction of the 3D model is improved.

A method for surface reduction of a three-dimensional model, comprising:

Acquiring the three-dimensional file to be processed, and constructing three-dimensional model data corresponding to the three-dimensional file to be processed;

Perform surface reduction recognition on the three-dimensional model data to obtain a surface reduction result and a reference surface reduction value corresponding to the surface reduction result;

When the surface reduction result is surface reduction, according to the reference denomination reduction value, determine the target denomination value corresponding to each sub-model in the three-dimensional model data;

Perform corresponding surface reduction on the three-dimensional model data according to all target surface values corresponding to the sub-models;

Compressing the three-dimensional model data after surface reduction into a surface-reduction three-dimensional file and outputting it;

The performing surface reduction identification on the 3D model data to obtain a surface reduction result and a reference surface reduction value corresponding to the surface reduction result, including:

Carrying out plane data collection on the three-dimensional model data to obtain a plurality of plane data; the plane data includes plane images, plane numbers and plane dimensions;

Inputting all the planar images into the surface reduction recognition model, performing texture feature extraction on all the planar images through the surface reduction recognition model, and obtaining the reduced surface identified by the surface reduction recognition model according to the extracted texture features. Surface results;

When the surface reduction result is a surface to be reduced, vector conversion is performed on the number of plane surfaces and the plane size in each of the plane data to obtain a vector matrix corresponding to the three-dimensional model data;

The vector matrix is input into a reference recognition model, and the reference face reduction prediction is performed on the vector matrix through the reference recognition model to obtain the reference face reduction value corresponding to the result of the face reduction.

A surface reduction device for a three-dimensional model, comprising:

An acquisition module, configured to acquire a three-dimensional file to be processed, and construct three-dimensional model data corresponding to the three-dimensional file to be processed;

An identification module, configured to perform area reduction identification on the three-dimensional model data, to obtain an area reduction result and a reference area reduction value corresponding to the area reduction result;

A determination module, configured to determine the target denomination corresponding to each sub-model in the 3D model data according to the reference denomination denomination when the denomination result is the denomination required;

A surface reduction module, configured to perform corresponding surface reduction on the 3D model data according to all target surface values corresponding to the sub-models;

An output module, configured to compress the three-dimensional model data after surface reduction into a surface-reduction three-dimensional file and output it;

The identification module is also used for:

Carrying out plane data collection on the three-dimensional model data to obtain a plurality of plane data; the plane data includes plane images, plane numbers and plane dimensions;

Inputting all the planar images into the surface reduction recognition model, performing texture feature extraction on all the planar images through the surface reduction recognition model, and obtaining the reduced surface identified by the surface reduction recognition model according to the extracted texture features. Surface results;

When the surface reduction result is a surface to be reduced, vector conversion is performed on the number of plane surfaces and the plane size in each of the plane data to obtain a vector matrix corresponding to the three-dimensional model data;

The vector matrix is input into a reference recognition model, and the reference face reduction prediction is performed on the vector matrix through the reference recognition model to obtain the reference face reduction value corresponding to the result of the face reduction.

A computer device, comprising a memory, a processor, and a computer program stored in the memory and operable on the processor, when the processor executes the computer program, the steps of the above-mentioned three-dimensional model surface reduction method are realized .

A computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned three-dimensional model surface reduction method are realized.

The three-dimensional model surface reduction method, device, computer equipment, and storage medium provided by the present invention construct three-dimensional model data corresponding to the three-dimensional file to be processed by acquiring the three-dimensional file to be processed; perform surface reduction recognition on the three-dimensional model data , to obtain the surface reduction result and the reference surface reduction value corresponding to the surface reduction result; when the surface reduction result is that surface reduction is required, according to the reference surface reduction value, it is determined to correspond to each sub-model in the three-dimensional model data The value of the target surface; according to all the target surface values corresponding to the sub-models, the corresponding surface reduction is performed on the three-dimensional model data; the three-dimensional model data after the surface reduction is compressed into a surface-reduction three-dimensional file and output, so , to realize the automatic recognition of the surface reduction result and the reference surface value by constructing the 3D model data, and automatically output the target surface value of each sub-model in the 3D model data when the surface reduction result needs to be reduced, and reduce the surface value of each sub-model The surface is reduced according to the corresponding target surface value, and the 3D file of surface reduction is compressed and output, so as to automatically complete the surface reduction operation of the 3D model, without manual operation, which reduces labor costs and improves the efficiency of surface reduction of the 3D model. In addition, the effectiveness and no distortion of the surface reduction of the 3D model are guaranteed, and the quality of the surface reduction of the 3D model is improved.

Description of drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings that need to be used in the description of the embodiments of the present invention. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention , for those skilled in the art, other drawings can also be obtained according to these drawings without paying creative labor.

Fig. 1 is a schematic diagram of the application environment of the surface reduction method of the three-dimensional model in an embodiment of the present invention;

Fig. 2 is a flowchart of a method for reducing the surface of a three-dimensional model in an embodiment of the present invention;

Fig. 3 is a flow chart of step S20 of the surface reduction method of a three-dimensional model in an embodiment of the present invention;

Fig. 4 is a flow chart of step S20 of the surface reduction method of a three-dimensional model in another embodiment of the present invention;

Fig. 5 is a flow chart of step S30 of the surface reduction method of a three-dimensional model in an embodiment of the present invention;

Fig. 6 is a functional block diagram of a conversion module of a surface reduction device for a three-dimensional model in an embodiment of the present invention;

Fig. 7 is a schematic diagram of a computer device in an embodiment of the present invention.

detailed description

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

The surface reduction method of the 3D model provided by the present invention can be applied in the application environment as shown in Fig. 1, wherein the client (computer device) communicates with the server through the network. Wherein, the client (computer device) includes but is not limited to various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices. The server can be implemented by an independent server or a server cluster composed of multiple servers.

In one embodiment, as shown in FIG. 2 , a method for surface reduction of a three-dimensional model is provided, and its technical solution mainly includes the following steps S10-S50:

S10. Acquire a 3D file to be processed, and construct 3D model data corresponding to the 3D file to be processed.

Understandably, the 3D file to be processed is an original 3D model file that has not been processed and needs surface reduction processing, and the process of constructing the 3D model data is to open the 3D file to be processed through application software For example, the application software is 3dsMax software, and the three-dimensional model data can be constructed by the 3dsMax software, and the three-dimensional model data is the data decoded and displayed by the application software, and the three-dimensional model data includes the three-dimensional model and the three-dimensional model Each of the sub-models below, the three-dimensional model is a model with a three-dimensional structure, and the sub-model is a model of the sub-objects constituting the three-dimensional model.

S20. Perform surface reduction recognition on the 3D model data to obtain a surface reduction result and a reference surface reduction value corresponding to the surface reduction result.

Understandably, the processing method of the surface reduction recognition can be set according to requirements, through the processing process of the surface reduction recognition, it can be identified whether the 3D model needs to be reduced, that is, whether the 3D model data needs to be reduced. Face processing, and in the case of needing to reduce the face, identify the base face reduction value, in the case of no need to reduce the face value, set the reference face reduction value to zero, indicating that the three-dimensional model data does not need to be reduced. The 3D model data has already reached the situation where the surface reduction cannot be performed. If the surface reduction operation is performed again, rendering distortion or the consequence of destroying the 3D model will occur. The result of the surface reduction indicates whether the 3D model data needs to be reduced. The reference surface reduction value is the reference value of the surface reduction determined for the 3D model. For example, the processing method of the surface reduction recognition can be to perform surface number statistics on the 3D model data to obtain the total surface area of the 3D model data. According to the total number of faces, it is judged whether it is necessary to reduce the face. If it is necessary to reduce the face, the processing process of the reference face reduction value is determined in combination with the volume value, surface area value and size value of each of the sub-models. The processing method of the face reduction identification It is also possible to collect plane data on the three-dimensional model data to obtain a plurality of plane data, perform texture feature extraction on the plane images in all the plane data through the plane reduction recognition model, and obtain the output plane reduction results. In the case of face reduction, the number of faces and the size of the planes in each of the planar data are vector-converted into a vector matrix, and the reference recognition model is used to predict the reference face reduction of the vector matrix, and the process of predicting the value of the reference face reduction process.

Wherein, the volume value is the value of the volume of the sub-model, the value of the surface area is the value of the surface area of the sub-model, the value of the size is the value of the length, width and height of the sub-model, and the plane data It is the data obtained under the capture of the upper, lower, left, right, front and rear six orientations of the spatial shape of the three-dimensional model. The plane data includes plane images, plane numbers and plane dimensions, and the plane images are The captured binary image, that is, the image after grayscale processing, the number of planes is the number of all planes in the plane image captured in one orientation, and the plane size is the three-dimensional dimension in the plane image captured in one orientation The limit length and limit width of the model, the limit length is the maximum length of the captured 3D model, the limit width is the maximum width of the captured 3D model, and the plane size includes the limit length and the limit width .

In an embodiment, after the step S20, that is, after performing surface reduction recognition on the 3D model, obtaining a surface reduction result and a reference surface reduction value corresponding to the surface reduction result, it further includes: S60, after the When the result of surface reduction is that surface reduction is not required, the 3D model data is compressed into a 3D file with surface reduction and output.

Understandably, in the case that the 3D model does not need to reduce the surface, the 3D model data is directly compressed into the 3D file with reduced surface, that is, the LZMA (Lempel-Ziv-Markov chain-Algorithm) compression algorithm is used to compress the The 3D model data is compressed to obtain the surface-reduced 3D file. The LZMA compression algorithm uses an improved compression algorithm of LZ77 (lossless compression algorithm) supported by interval coding and an algorithm specially used for binary preprocessing programs.

The present invention realizes automatically compressing the three-dimensional model data to obtain the reduced area three-dimensional file when it is automatically judged that the area reduction result is no need for area reduction, which reduces the cost of manual automatic identification and compression, automatically compresses the three-dimensional file to be processed, and reduces the file size.

In one embodiment, as shown in FIG. 3, in the step S20, that is, performing surface reduction recognition on the 3D model, obtaining a surface reduction result and a reference surface reduction value corresponding to the surface reduction result, including:

S201. Perform face count statistics on the 3D model data to obtain the total face count of the 3D model data.

Understandably, the statistics of the number of faces is a process of counting the number of faces of each of the sub-models in the three-dimensional model, and the total number of faces is the sum of the number of faces obtained after the statistical processing of the number of faces, That is, there are as many faces as there are polygons in the 3D model.

S202. Determine whether the total face count is greater than a preset face count threshold.

Understandably, the preset face number threshold is the value of the preset face number, and the preset face number threshold is the minimum face value set by reducing faces, for example, the preset face number threshold is 8, 10, etc. , judging whether the total number of sides is smaller than the preset number of sides threshold, if the total number of sides is smaller than the preset number of sides threshold, it means that the number of sides can no longer be reduced, and at this time, no need to reduce the number of sides.

S203. If the total number of faces is greater than the preset face number threshold, determine the result of face reduction as faces to be reduced.

Understandably, if the total number of faces is greater than the preset face number threshold, it indicates that the 3D model data needs to be reduced, that is, the sub-models in the 3D model need to be reduced. At this time, the The area reduction result is determined as the area to be reduced, and the area reduction result includes the area to be reduced and the area not to be reduced.

S204. Acquire the volume value, surface area value and size value of each of the sub-models in the three-dimensional model data.

Understandably, the volume value, the surface area value, and the size value of each of the sub-models in the three-dimensional model can be obtained from the constructed three-dimensional model data.

S205, sort the volume values of all the sub-models to obtain a first sorting result, sort the surface area values of all the sub-models to obtain a second sorting result, and sort all the sub-models Sort the above size values to get the third sorting result;

Understandably, the volume values of each of the sub-models are sorted in descending order, and all the volume values after sorting are determined as the first sorting result, and the volume values of each of the sub-models are determined as the first sorting result. The surface area values are sorted from large to small, and all the sorted surface area values are determined as the second sorting result, and the size values of each of the sub-models are sorted from large to small Sorting is performed, and all sorted size values are determined as the third sorting result.

S206. According to the first sorting result, the second sorting result, and the third sorting result, determine the reference face reduction value corresponding to the face reduction result.

Understandably, the first three volume values in the first sorting result and the last three volume values in the first sorting result are averaged to obtain the first A sorting mean value, the second sorting result and the third sorting result are processed similarly to obtain the second sorting mean value and the third sorting mean value respectively, and the first sorting mean value, the second sorting mean value and the The third sorting mean value is mapped and converted into a value of the same dimension (i.e., the base-reduction dimension), that is, the numerical range of which section the first sorting mean value falls into is detected, and it is mapped to the value corresponding to the numerical value. The first benchmark value corresponding to the range, similarly do the same mapping conversion process on the second sorting mean value and the third sorting mean value, respectively obtain the second benchmark value and the third benchmark value, and convert the first benchmark value , the second benchmark value and the third benchmark value are input into the benchmark face-reduction function, and the benchmark face-reduction value is calculated by the benchmark face-reduction function, and the benchmark face-reduction function is:

L＝α ₁ L ₁ +α ₂ L ₂ +α ₃ L ₃

in:

L is the benchmark minus face value;

L ₁ is the first reference value;

L ₂ is the second reference value;

L ₃ is the third reference value;

α ₁ is the preset first benchmark weight;

α ₂ is the preset second benchmark weight;

α ₃ is a preset third benchmark weight.

The present invention realizes that the total number of faces of the three-dimensional model data is obtained by counting the number of faces of the three-dimensional model data; judging whether the total number of faces is greater than the preset face number threshold; if the total number of faces is greater than the If the threshold value of the number of faces is preset, then the result of the face reduction is determined as the face to be reduced; the volume value, the surface area value and the size value of each of the sub-models in the three-dimensional model data are obtained; Sort the volume values to obtain the first sorting result, sort the surface area values of all the sub-models to get the second sorting result, sort the size values of all the sub-models to get the third sorting result Result; according to the first sorting result, the second sorting result and the third sorting result, determine the benchmark denomination value corresponding to the deduction result, so that it is realized through face number statistics, preset Face number threshold, automatically judge whether need to reduce the face, and by sorting the volume value, surface area value and size value of each sub-model in the three-dimensional model data, combined with the sorting results to automatically determine the benchmark face value reduction, reduce Manual operation eliminates the need for manual subjective recognition, which improves the accuracy and reliability of recognition and improves the quality of surface reduction of 3D models.

In one embodiment, as shown in FIG. 4, in the step S20, that is, performing surface reduction recognition on the 3D model data to obtain a surface reduction result and a reference surface reduction value corresponding to the surface reduction result, and include:

S207. Collect plane data on the 3D model data to obtain a plurality of plane data; the plane data includes plane images, plane numbers and plane dimensions.

Understandably, the planar data collection is to capture six faces of the three-dimensional model data to obtain the planar images of the three-dimensional model data on six planes; to summarize the faces of each of the planar images, The process of obtaining the number of plane faces corresponding to each of the plane images, and simultaneously measuring the size of each of the plane images to obtain the plane size corresponding to each of the plane images.

In one embodiment, in the step S207, that is, collecting plane data on the 3D model data to obtain a plurality of plane data; the plane data includes plane images, plane numbers and plane dimensions, including:

S2071. Perform six-plane capture on the 3D model data to obtain the plane images of the 3D model data on six planes.

Understandably, the capture process is a process of capturing the plane image from six directions of the top, bottom, left, right, front and back of the spatial body through the application software, and it can also be understood as , down, left, right, front, and back six directions to take the image and then carry out the process of binarizing (ie, grayscale processing) the image, and the binarization process (ie, grayscale processing) is to The process of changing a color image containing brightness and color into a grayscale image.

S2072. Summarize the number of faces of each of the planar images to obtain the number of faces of the plane corresponding to each of the planar images, and at the same time measure the size of each of the planar images to obtain the number of faces corresponding to each of the planar images Flat size.

Understandably, the summary of the number of faces is to calculate the number of faces in the planar image, so as to obtain the number of planar faces of the planar image, and the measurement of the size is to measure the limit of the planar dimension in the planar image long and wide process of the limit.

S2073. Determine one planar image, the number of plane faces and the plane size corresponding to the planar image as one piece of planar data.

Understandably, the plane image of an orientation, the number of plane faces and the plane size corresponding to the plane image are marked as the plane data of the orientation.

S208. Input all the planar images into the surface reduction recognition model, perform texture feature extraction on all the planar images through the surface reduction recognition model, and obtain all the texture features identified by the surface reduction recognition model based on the extracted texture features. Describe the surface reduction results.

Understandably, the surface reduction recognition model is a trained deep neural network model, and the network structure of the surface reduction recognition model can be set according to requirements. Preferably, the network structure of the surface reduction recognition model is a network of VGG19 structure, the surface reduction recognition model is a model that realizes the surface reduction result by extracting the texture features in all the plane images and identifying whether surface reduction is required, and the texture features are polygonal edges in the plane image The length, density or coincidence-related features of the plane image, because the more planes in the plane image, the more obvious the texture features, so by extracting the texture features, it can be more accurately determined whether the surface reduction results need to be reduced. Through the The texture features in each of the planar images extracted by the surface reduction recognition model, and the texture features in each of the planar images are integrated to identify the result of surface reduction.

In one embodiment, before the step S208, that is, before inputting all the plane images into the plane subtraction recognition model, it includes:

S2081. Obtain a training sample set; the training sample set includes a plurality of training samples; each of the training samples is associated with a surface reduction label; the surface reduction label includes surface reduction and non-reduction; one training sample Includes six planar sample images from a historically collected 3D model sample.

Understandably, the training sample set is a collection of training samples, the training samples are historically collected 3D model samples, the 3D model samples are data with a 3D model as samples, and the training samples include 3D model samples. The six plane sample images in the sample, the surface reduction label indicates whether the training sample corresponding to it needs the result of surface reduction history, that is, whether the training sample whose surface reduction label is history has undergone surface reduction For the assigned label, the surface reduction label includes surface reduction and non-reduction surface, and one training sample is associated with one surface reduction label.

S2082. Input the training sample into a deep neural network model including initial parameters.

Understandably, the deep neural network model includes the initial parameters.

S2083. Extract the texture features of all the plane sample images in the training samples by using the deep neural network model.

S2084. Obtain a surface reduction sample result output by the deep neural network model according to the extracted texture feature, and determine a loss value according to a matching degree between the surface reduction sample result and the surface reduction label;

Understandably, the deep neural network model identifies according to the extracted texture features, identifies whether the training sample needs to be reduced, and outputs the result of recognition as the result of the reduced surface sample, and the reduced surface sample result includes face reduction is required and face reduction is not required, the face reduction sample result indicates whether the training sample needs face reduction results, and the face reduction sample result corresponding to the training sample is matched with the face reduction label, Using the binary classification cross-entropy loss algorithm, the degree of difference between the two can be matched, thereby calculating the loss value, which measures the gap between the surface-reduced sample result and the surface-reduced label, Through the loss value, the deep neural network model can be continuously moved closer to an accurate recognition result, thereby improving the recognition accuracy.

S2085. When the loss value does not reach the preset convergence condition, iteratively update the initial parameters of the deep neural network model until the loss value reaches the preset convergence condition, and the depth after convergence The neural network model is recorded as the face reduction recognition model after training.

Understandably, the convergence condition may be a condition that the loss value is very small and will not decrease after 1000 calculations, that is, the loss value is very small and will not decrease after 1000 calculations. When descending again, stop the training, and record the deep neural network model after the convergence as the face-reduction recognition model that the training is completed; When the loss value is less than the set threshold, the training is stopped, and the deep neural network model after convergence is recorded as the face-reduction recognition model that has been trained. In this way, when the loss value does not reach the preset convergence condition, it is continuously adjusted The initial parameters of the deep neural network model, and triggering the step of extracting the texture features of all the plane sample images in the training samples through the deep neural network model, can continuously move closer to accurate results, allowing recognition The accuracy rate is getting higher and higher. In this way, the surface reduction recognition can be optimized, and the accuracy and reliability of the 3D model surface reduction recognition are improved.

The present invention realizes that by acquiring a training sample set; extracting the texture features of all the plane sample images in the training samples through the deep neural network model; acquiring the deep neural network model according to the extracted texture features output surface reduction sample results, and determine the loss value according to the matching degree of the surface reduction sample results and the surface reduction label; when the loss value does not reach the preset convergence condition, iteratively update the deep neural network model The initial parameters of the initial parameter, until the loss value reaches the preset convergence condition, the deep neural network model after convergence is recorded as the face-reduction recognition model that has been trained. In this way, the present invention realizes the Extract the texture features, and identify the surface reduction sample results, and continuously iterate the initial parameters according to the loss value, so as to realize the accurate and fast identification of whether the input image needs surface reduction results, which can improve the accuracy of 3D model surface reduction recognition The efficiency and quality are reduced, and the training efficiency is improved.

S209. When the surface reduction result is that surface reduction is required, perform vector conversion on the number of plane surfaces and the plane size in each of the plane data to obtain a vector matrix corresponding to the three-dimensional model data.

Understandably, if the surface reduction result is that the surface needs to be reduced, the number of planes and the size of the planes in each of the plane data are converted into vectors, and the vector conversion is the number of planes and the number of planes The plane size is converted into a binary code corresponding to the same dimension (number of preset dimensions), such as a 16-bit binary code, and the plane number and the plane size in one plane data are converted The obtained values form a one-dimensional array, and all the planar data are combined into a multi-dimensional array after vector conversion, and the multi-dimensional array is determined as the vector matrix.

S210. Input the vector matrix into a reference recognition model, and perform reference face reduction prediction on the vector matrix through the reference recognition model, to obtain the reference face reduction value corresponding to the result of the face reduction.

Understandably, the benchmark recognition model is a clustering model that has been trained, and the benchmark recognition model performs clustering through historically collected vector matrices, and can learn and predict the hidden mapping relationship between the benchmark minus face value and the vector matrix, The benchmark denomination prediction is performed on the vector matrix through the benchmark recognition model, and the corresponding benchmark denomination value can be predicted.

The present invention realizes that by collecting the plane data of the three-dimensional model data, a plurality of plane data are obtained; all the plane images are input into the surface reduction recognition model, and the texture features of all the plane images are carried out through the surface reduction recognition model Extracting, obtaining the surface reduction result identified by the surface reduction recognition model according to the extracted texture features; Carry out vector conversion with the plane size to obtain a vector matrix corresponding to the three-dimensional model data; input the vector matrix into the benchmark recognition model, and perform benchmark reduction area prediction on the vector matrix through the benchmark recognition model to obtain the same as The reference denomination value corresponding to the surface reduction result, in this way, it is possible to identify whether the 3D model data needs surface reduction by collecting plane data on the 3D model data and extracting texture features through the surface reduction recognition model. Therefore, it can accurately and scientifically automatically identify whether the result of face reduction is needed, and can quickly predict the benchmark denomination value for subsequent The surface reduction of the 3D model provides a data basis, which improves the recognition accuracy and quality of the surface reduction of the 3D model.

S30. When the surface reduction result is that surface reduction is required, determine a target face value corresponding to each sub-model in the three-dimensional model data according to the reference face reduction value.

Understandably, when the surface reduction result is surface reduction, combined with the reference surface reduction value and the volume value, volume ratio, surface area value, and surface area ratio of each of the sub-models in the three-dimensional model data , size value and size ratio to calculate the target surface value of each sub-model in the three-dimensional model data, and the target surface value of a sub-model indicates the number of surfaces that the sub-model needs to reduce to achieve. value.

In one embodiment, as shown in FIG. 5, in the step S30, that is, determining the target denomination corresponding to each sub-model in the three-dimensional model data according to the denomination of the reference, including:

S301. Obtain volume values, volume proportions, surface areas, surface area proportions, dimensions, and dimension proportions of each of the sub-models in the three-dimensional model data.

Understandably, the three-dimensional model data also includes the volume ratio, the surface area ratio and the size ratio of each of the sub-models, and the volume ratio is the volume ratio of the sub-models. The percentage of the total volume of the three-dimensional model, the surface area ratio is the percentage of the surface area value of the sub-model to the total surface area of the three-dimensional model, and the size ratio is the ratio of the size value of the sub-model to the three-dimensional The percentage of the total size of the model, that is, the percentage of the product of the limit length and limit width of each sub-model to the product of the length and width of the three-dimensional model.

S302. Determine, according to the volume value, the surface area value, and the size value of each of the sub-models, a protective surface value corresponding to each of the sub-models.

Understandably, according to the volume value, the surface area value and the size value of the sub-model, the value of the protection surface corresponding thereto can be mapped out, and the value of the protection surface is to ensure that the outline of the sub-model is not The number of surfaces damaged by surface reduction, that is, which section the volume value, the surface area value and the size value fall into respectively, and then map a protection surface value according to the three sections falling into it respectively, among which three A segment of dimensions corresponds to a protection surface value map.

S303. Multiply the base denomination value by the volume ratio of each of the sub-models to obtain a first proportion value corresponding to each of the sub-models, and multiply the reference denomination value by the volume proportion of each of the sub-models multiplying the surface area proportions to obtain a second proportion corresponding to each of the sub-models, multiplying the base minus face value by the size proportions of each of the sub-models to obtain a second proportion corresponding to each of the sub-models The corresponding third proportion value.

S304. Determine the target corresponding to the sub-model according to the protection surface value, the first proportion, the second proportion, and the third proportion corresponding to the same sub-model face value.

Understandably, the estimated face value corresponding to the sub-model is determined through the first proportion value, the second proportion value and the third proportion value corresponding to the sub-model, and by combining with the The predicted face value corresponding to the sub-model is compared with the protected face value, so as to determine the target face value corresponding to the sub-model.

The present invention realizes that by acquiring the volume value, volume ratio, surface area value, surface area ratio, size value and size ratio of each of the sub-models in the three-dimensional model data; according to the volume of each of the sub-models value, the surface area value and the size value, determine the protective surface value corresponding to each of the sub-models; multiply the base minus face value with the volume ratio of each of the sub-models to obtain the corresponding For the first proportion corresponding to the sub-models, multiply the base minus face value by the surface area proportions of each of the sub-models to obtain a second proportion corresponding to each of the sub-models, and multiply the base The deduction face value is multiplied by the size ratio of each of the sub-models to obtain the third ratio corresponding to each of the sub-models; according to the protection face value corresponding to the same sub-model, the first A proportion value, the second proportion value and the third proportion value determine the target denomination value corresponding to the sub-model, so that scientifically and accurately automatically determine the denomination value corresponding to each sub-model The target surface value can ensure the quality of surface reduction for each sub-model, and provides the target surface value of each sub-model for surface reduction, which improves the accuracy of subsequent surface reduction and avoids distortion and destruction of surface reduction.

In one embodiment, in the step S304, that is, according to the value of the protection area, the first proportion value, the second proportion value and the third proportion value corresponding to the same sub-model Ratio, determine the target face value corresponding to the sub-model, including:

S3041. Round up the largest value among the first proportion value, the second proportion value, and the third proportion value corresponding to the sub-model, and determine it as the predicted value corresponding to the sub-model. face value.

Understandably, the rounding is rounding down the largest value among the first percentage value, the second percentage value and the third percentage value, and the rounded value is determined as The estimated face value corresponding to the sub-model, the estimated face value is the estimated face number that the sub-model needs to reduce to achieve.

S3042. Compare the predicted face value corresponding to the sub-model with the protected face value.

S3043. If the predicted face value corresponding to the sub-model is less than or equal to the protected face value, determine the protected face value as the target face value corresponding to the sub-model.

S3044. If the estimated face value corresponding to the sub-model is greater than the protected face value, determine the estimated face value as the target face value corresponding to the sub-model.

The present invention realizes that by rounding the largest value among the first proportion value, the second proportion value and the third proportion value corresponding to the sub-model, it is determined to be the same as the sub-model Corresponding estimated face value; compare the estimated face value corresponding to the sub-model with the protected face value; if the estimated face value corresponding to the sub-model is less than or equal to the protected face value face value, then determine the protected face value as the target face value corresponding to the sub-model; if the estimated face value corresponding to the sub-model is greater than the protected face value, then set the predicted face value The estimated surface value is determined as the target surface value corresponding to the sub-model, thus, a method for determining the target surface value of each sub-model is provided, the target of subsequent three-dimensional model reduction is provided, and the three-dimensional model reduction is improved. the quality of.

S40. According to all the target surface values corresponding to the sub-models, corresponding surface reduction is performed on the 3D model data.

Understandably, through the application program to reduce the area of each of the sub-models according to the target area value of each of the sub-models, the corresponding area reduction operation can be automatically performed, and the non-compliant and unnecessary display Surfaces are subtracted, so that the number of surfaces retained by each sub-model approaches the corresponding target surface value until the target surface value is reached, thereby completing the corresponding surface reduction operation, and finally the corresponding surface reduction operations of all the sub-models are completed After that, the 3D model data after surface reduction is obtained.

S50. Compress the 3D model data after plane reduction into a plane-reduced 3D file and output it.

Understandably, compressing the 3D model data after plane reduction into the plane-reduced 3D file, that is, using the LZMA (Lempel-Ziv-Markov chain-Algorithm) compression algorithm to compress the 3D model data after plane reduction Compress to obtain the reduced plane three-dimensional file, the LZMA compression algorithm uses the improved compression algorithm of LZ77 (lossless compression algorithm) supported by interval coding and the algorithm specially used for the binary preprocessing program, and the described reduced plane three-dimensional file Output, complete the surface reduction process of the 3D file to be processed, through the experimental data, the 3D file with reduced surface ensures the effect of subsequent rendering, and the capacity of the 3D file with reduced surface is 10% less than the capacity of the 3D file to be processed above.

The present invention achieves the construction of three-dimensional model data corresponding to the three-dimensional file to be processed by acquiring the three-dimensional file to be processed; performing surface reduction recognition on the three-dimensional model data to obtain the surface reduction result and the benchmark corresponding to the surface reduction result denomination; when the result of the deduction is that the denomination is required, according to the benchmark denomination, determine the target denomination corresponding to each sub-model in the three-dimensional model data; according to all the targets corresponding to the sub-models surface value, corresponding surface reduction is performed on the three-dimensional model data; the three-dimensional model data after surface reduction is compressed into a surface-reduction three-dimensional file and output, so that by building three-dimensional model data, automatic recognition of surface reduction results and Base face value reduction, when the face reduction result is the face reduction, the target face value of each sub-model in the 3D model data is automatically output, and the face reduction of each sub-model is correspondingly reduced according to the corresponding target face value, and compressed Output the 3D surface reduction file to automatically complete the surface reduction operation of the 3D model without manual operation, reduce labor costs, improve the efficiency of 3D model surface reduction, and ensure the effectiveness and undistortion of the 3D model surface reduction, improving Improve the quality of the 3D model minus the surface.

In one embodiment, a device for reducing the surface of a three-dimensional model is provided, and the device for reducing the surface of the three-dimensional model corresponds to the method for reducing the surface of the three-dimensional model in the above-mentioned embodiments. As shown in FIG. 6 , the surface reduction device for the 3D model includes an acquisition module 11 , an identification module 12 , a determination module 13 , a surface reduction module 14 and an output module 15 . The detailed description of each functional module is as follows:

An acquisition module 11, configured to acquire a three-dimensional file to be processed, and construct three-dimensional model data corresponding to the three-dimensional file to be processed;

The identification module 12 is configured to perform area reduction identification on the three-dimensional model data, and obtain an area reduction result and a reference area reduction value corresponding to the area reduction result;

The determination module 13 is used to determine the target denomination corresponding to each sub-model in the three-dimensional model data according to the reference denomination denomination when the deduction result is denomination required;

A surface reduction module 14, configured to perform corresponding surface reduction on the three-dimensional model data according to all target surface values corresponding to the sub-models;

The output module 15 is configured to compress the 3D model data after plane reduction into a plane-reduced 3D file and output it.

For the specific limitations of the surface reduction device for a 3D model, refer to the above definition of the surface reduction method for a 3D model, which will not be repeated here. Each module in the above-mentioned three-dimensional model surface reduction device can be fully or partially realized by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, and can also be stored in the memory of the computer device in the form of software, so that the processor can invoke and execute the corresponding operations of the above-mentioned modules.

In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure may be as shown in FIG. 7 . The computer device includes a processor, memory, network interface and database connected by a system bus. Wherein, the processor of the computer device is used to provide calculation and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs and databases. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by a processor, a method for reducing the surface of a three-dimensional model is realized.

In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored on the memory and operable on the processor. When the processor executes the computer program, the surface reduction of the three-dimensional model in the above-mentioned embodiments is realized. method.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the method for reducing the surface of a three-dimensional model in the above-mentioned embodiments is implemented.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented through computer programs to instruct related hardware, and the computer programs can be stored in a non-volatile computer-readable memory In the medium, when the computer program is executed, it may include the processes of the embodiments of the above-mentioned methods. Wherein, any reference to memory, storage, database or other media used in the various embodiments provided by the present invention may include non-volatile and/or volatile memory. Nonvolatile memory can include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Those skilled in the art can clearly understand that for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used for illustration. In practical applications, the above-mentioned functions can be assigned to different functional units, Completion of modules means that the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still carry out the foregoing embodiments Modifications to the technical solutions recorded in the examples, or equivalent replacement of 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 various embodiments of the present invention, and should be included in within the protection scope of the present invention.

Claims (9)

1. A method for reducing the surface of a three-dimensional model, characterized in that, comprising: Acquiring the three-dimensional file to be processed, and constructing three-dimensional model data corresponding to the three-dimensional file to be processed; Perform surface reduction recognition on the three-dimensional model data to obtain a surface reduction result and a reference surface reduction value corresponding to the surface reduction result; When the surface reduction result is surface reduction, according to the reference denomination reduction value, determine the target denomination value corresponding to each sub-model in the three-dimensional model data; Perform corresponding surface reduction on the three-dimensional model data according to all target surface values corresponding to the sub-models; Compressing the three-dimensional model data after surface reduction into a surface-reduction three-dimensional file and outputting it; The performing surface reduction identification on the 3D model data to obtain a surface reduction result and a reference surface reduction value corresponding to the surface reduction result, including: Carrying out plane data collection on the three-dimensional model data to obtain a plurality of plane data; the plane data includes plane images, plane numbers and plane dimensions; Inputting all the planar images into the surface reduction recognition model, performing texture feature extraction on all the planar images through the surface reduction recognition model, and obtaining the reduced surface identified by the surface reduction recognition model according to the extracted texture features. Surface results; When the surface reduction result is a surface to be reduced, vector conversion is performed on the number of plane surfaces and the plane size in each of the plane data to obtain a vector matrix corresponding to the three-dimensional model data; The vector matrix is input into a reference recognition model, and the reference face reduction prediction is performed on the vector matrix through the reference recognition model to obtain the reference face reduction value corresponding to the result of the face reduction. 2. The surface reduction method of a three-dimensional model as claimed in claim 1, wherein said three-dimensional model is carried out for surface reduction recognition to obtain a surface reduction result and a reference surface reduction value corresponding to the surface reduction result, and include: Perform surface number statistics on the three-dimensional model data to obtain the total number of surfaces of the three-dimensional model data; judging whether the total face count is greater than a preset face count threshold; If the total number of faces is greater than the preset number of faces threshold, the face reduction result is determined as the face to be reduced; Acquiring the volume value, surface area value and dimension value of each of the sub-models in the three-dimensional model data; sorting the volume values of all the sub-models to obtain a first sorting result, sorting the surface area values of all the sub-models to obtain a second sorting result, and sorting the dimensions of all the sub-models Values are sorted to get the third sorted result; According to the first sorting result, the second sorting result and the third sorting result, determine the reference denomination value corresponding to the deduction result. 3. The surface reduction method of a three-dimensional model as claimed in claim 1, wherein, before inputting all the plane images into the surface reduction recognition model, it comprises: Obtain a training sample set; the training sample set includes a plurality of training samples; each of the training samples is associated with a face reduction label; the face reduction label includes face reduction and need not face reduction; one training sample includes a Six planar sample images from historically collected 3D model samples; The training samples are input into a deep neural network model comprising initial parameters; extracting the texture features of all the plane sample images in the training samples through the deep neural network model; Obtaining the surface reduction sample result output by the deep neural network model according to the extracted texture feature, and determining a loss value according to the matching degree between the surface reduction sample result and the surface reduction label; When the loss value does not reach the preset convergence condition, iteratively update the initial parameters of the deep neural network model, until the loss value reaches the preset convergence condition, the deep neural network after convergence The model is recorded as a face reduction recognition model that has been trained. 4. The surface reduction method of a three-dimensional model as claimed in claim 1, wherein the plane data collection is carried out to the three-dimensional model data to obtain a plurality of plane data; the plane data comprises plane images, plane faces and plane dimensions, including: Capturing the three-dimensional model data on six planes to obtain the plane images of the three-dimensional model data on six planes; Summarizing the number of faces of each of the planar images to obtain the number of faces of the plane corresponding to each of the planar images, and at the same time measuring the size of each of the planar images to obtain the plane size corresponding to each of the planar images ; One planar image, the number of plane faces and the plane size corresponding to the planar image are determined as one piece of planar data. 5. The surface reduction method of a three-dimensional model as claimed in claim 1, wherein said denomination according to said reference, determining a target surface value corresponding to each sub-model in said three-dimensional model data, comprises: Acquiring the volume value, volume ratio, surface area value, surface area ratio, size value, and size ratio of each of the sub-models in the three-dimensional model data; According to the volume value, the surface area value and the size value of each of the sub-models, the value of the protective surface corresponding to each of the sub-models is determined; Multiplying the base denomination by the volume proportion of each sub-model to obtain the first proportion corresponding to each sub-model, and multiplying the base deduction by the volume proportion of each sub-model multiply the surface area proportions to obtain the second proportions corresponding to each of the sub-models, multiply the base minus face value with the size proportions of each of the sub-models to obtain the second proportions corresponding to each of the sub-models the third proportion; According to the protection face value, the first percentage value, the second percentage value and the third percentage value corresponding to the same sub-model, the target face value corresponding to the sub-model is determined . 6. The surface reduction method of a three-dimensional model as claimed in claim 5, wherein said protection surface value corresponding to the same sub-model, said first ratio, and said second ratio Ratio and the third proportion, determine the target denomination corresponding to the sub-model, including: Rounding the largest value among the first proportion, the second proportion and the third proportion corresponding to the sub-model, and determining it as the estimated surface corresponding to the sub-model value; comparing said estimated face value and said protected face value corresponding to said sub-model; If the estimated denomination value corresponding to the sub-model is less than or equal to the protection denomination value, then determining the protection denomination value as the target denomination value corresponding to the sub-model; If the estimated face value corresponding to the sub-model is greater than the protected face value, then determine the estimated face value as the target face value corresponding to the sub-model. 7. A surface reducing device for a three-dimensional model, characterized in that it comprises: An acquisition module, configured to acquire a three-dimensional file to be processed, and construct three-dimensional model data corresponding to the three-dimensional file to be processed; An identification module, configured to perform area reduction identification on the three-dimensional model data, to obtain an area reduction result and a reference area reduction value corresponding to the area reduction result; A determination module, configured to determine the target denomination corresponding to each sub-model in the 3D model data according to the reference denomination denomination when the denomination result is the denomination required; A surface reduction module, configured to perform corresponding surface reduction on the 3D model data according to all target surface values corresponding to the sub-models; An output module, configured to compress the three-dimensional model data after surface reduction into a surface-reduction three-dimensional file and output it; The identification module is also used for: Carrying out plane data collection on the three-dimensional model data to obtain a plurality of plane data; the plane data includes plane images, plane numbers and plane dimensions; Inputting all the planar images into the surface reduction recognition model, performing texture feature extraction on all the planar images through the surface reduction recognition model, and obtaining the reduced surface identified by the surface reduction recognition model according to the extracted texture features. Surface results; When the surface reduction result is a surface to be reduced, vector conversion is performed on the number of plane surfaces and the plane size in each of the plane data to obtain a vector matrix corresponding to the three-dimensional model data; The vector matrix is input into a reference recognition model, and the reference face reduction prediction is performed on the vector matrix through the reference recognition model to obtain the reference face reduction value corresponding to the result of the face reduction. 8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and operable on the processor, characterized in that, when the processor executes the computer program, the computer program according to claim The surface reduction method of the three-dimensional model described in any one of 1 to 6. 9. A computer-readable storage medium, the computer-readable storage medium stores a computer program, characterized in that, when the computer program is executed by a processor, the three-dimensional model according to any one of claims 1 to 6 is realized surface reduction method.

Images