Cloud three-dimensional exhibition visualization method

Abstract

The invention relates to the technical field of image generation, in particular to a cloud three-dimensional exhibition visualization method, which comprises the following steps: acquiring a three-dimensional exhibit multi-view image sequence, extracting reflectivity, surface texture and depth information under each view, calculating reflection characteristics, and generating an original optical characteristic matrix. According to the method, a multi-view image sequence is collected, reflectivity, surface texture and depth information is extracted, an original optical feature matrix containing illumination and reflection characteristics is generated, a material feature description set is formed in combination with multi-view illumination changes and reflection parameters, and comprehensive expression of real characteristics of materials is achieved. According to the method, calculation of the curvature change rate and the texture density value is combined with regional feature index analysis, the division precision of a three-dimensional geometric structure is improved, a spatial feature partitioning method is established through extraction of local curved surface features and contour boundary information and construction of an octree index structure, and the data organization efficiency and accuracy of a complex region are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of image generation, and in particular to a cloud-based three-dimensional exhibition visualization method. Background Art

[0002] The field of image generation technology involves technical methods for generating, processing and optimizing images using computer algorithms and mathematical models. This field covers the generation process from two-dimensional images to three-dimensional images, including image rendering, texture generation, material mapping, lighting calculation and other technical means. The cloud-based three-dimensional exhibition visualization method is an application method based on image generation technology, which aims to realize the three-dimensional visualization of exhibits in the cloud by calculating and reconstructing the material, lighting, reflection and other properties of three-dimensional exhibits. However, in the process of space division, the existing technology has limited processing capabilities for complex geometric areas, and simple partitioning strategies are difficult to accurately describe details and local features, affecting the optimization effect of subsequent rendering parameters. In terms of lighting calculation, there is a lack of dynamic adaptability to different lighting conditions and material characteristics, resulting in insufficient consistency in material performance under multiple perspectives. At the same time, the existing technology has weak global and local balance processing of lighting and reflection, and it is difficult to meet the needs of highly complex three-dimensional exhibits in terms of material and lighting detail expression. Therefore, improvements are needed. Summary of the invention

[0003] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a cloud-based three-dimensional exhibition visualization method.

[0004] In order to achieve the above object, the present invention adopts the following technical solution, a cloud-based three-dimensional exhibition visualization method, comprising the following steps:

[0005] Collect multi-view image sequences of three-dimensional exhibits, extract reflectivity, surface texture and depth information at each viewing angle, calculate reflection characteristics, generate an original optical feature matrix, fuse multi-view illumination change characteristics and surface reflection parameters based on the original optical feature matrix, and generate a material feature description set;

[0006] Based on the material feature description set, the curvature change rate and texture density value of each region are calculated to generate a regional feature index, and the surface of the three-dimensional exhibit is divided into regions based on the regional feature index, local surface features and contour boundary information are extracted, an octree index structure is established, and a spatial feature partition map is generated;

[0007] Based on the spatial feature partition map, material reconstruction mapping is performed on each partition, lighting parameters and reflection angles are adjusted, a lighting-material mapping relationship is generated, surface reflection intensity and material rendering parameters are calculated based on the lighting-material mapping relationship, a rendering parameter optimization matrix is ​​constructed, and a material rendering distribution map is generated;

[0008] Based on the material rendering distribution map, the lighting conditions of the cloud exhibition environment are sampled, the material reflection effects at different viewing angles are calculated, and according to the material reflection effects at different viewing angles, the material rendering parameters are integrated to perform material reconstruction and lighting enhancement on the three-dimensional exhibit surface to generate a three-dimensional exhibition rendering result.

[0009] Preferably, the steps of obtaining the original optical characteristic matrix are:

[0010] Collect multi-view image sequences of 3D exhibits, extract reflectivity, surface texture and depth information at each viewpoint, and integrate the data to generate a multi-view optical data set;

[0011] According to the multi-view optical data set, the reflection characteristic score at each viewing angle is calculated, and the calculation formula is:

[0012]

[0013] Among them, P ij Score the reflectance property of the i-th viewing angle and j-th surface area, R ij is the reflectivity at the i-th viewing angle, WT i is the surface texture parameter of the i-th viewing angle, D j is the depth value of the jth region, E ij is the reflection characteristic enhancement factor;

[0014] According to the reflection characteristic scores, all reflection characteristic scores are integrated into a matrix form to generate an original optical characteristic matrix.

[0015] Preferably, the steps of obtaining the material feature description set are:

[0016] Extracting illumination change characteristics and surface reflection parameters in each viewing angle according to the original optical feature matrix, analyzing each element value in the matrix, identifying illumination change and surface reflection rules, calling the analyzed optical feature data, and generating a multi-view illumination feature set;

[0017] Based on the multi-view illumination characteristic set, the interactive relationship between the illumination characteristics and the surface reflection parameters is analyzed for each view angle, the influence of the adjustment of the illumination parameters on the reflection characteristics at each view angle is calculated, and the multiple interaction results are integrated into a comprehensive feature matrix integrating the illumination characteristics and the reflection parameters;

[0018] According to the comprehensive feature matrix, the interaction between each illumination parameter and the reflection characteristic parameter is mapped, the comprehensive feature matrix is ​​parsed row by row, a reflection characteristic mapping matrix is ​​generated, and a material feature description set is constructed according to the matrix data content.

[0019] Preferably, the steps of obtaining the regional characteristic index are:

[0020] Based on the material feature description set, a hierarchical structure of a three-dimensional space is constructed, and the material data is organized into a spatial hierarchical model through hierarchical processing to obtain a three-dimensional spatial hierarchical structure;

[0021] According to the three-dimensional spatial hierarchical structure, the curvature change rate of each area is calculated, and the calculation formula is:

[0022]

[0023] Among them, CR ij is the curvature change rate of the jth region of the i-th layer, K ik is the curvature value of the kth point in the i-th layer, is the average curvature of the i-th layer, θ ik is the texture value of the kth point in the i-th layer, is the average value of the texture of the i-th layer, and n represents the total number of points in the i-th layer;

[0024] Based on the three-dimensional spatial hierarchical structure, the texture density value of each area is calculated, and the calculation formula is:

[0025]

[0026] Among them, WD ij is the texture density value of the jth region of the i-th layer, T jl is the texture value of the lth point in the jth region, and wm represents the total number of points in the jth region;

[0027] The curvature change rate and the texture density value are combined to generate a regional feature index.

[0028] Preferably, the steps of obtaining the spatial feature partition map are:

[0029] Based on the regional characteristic indicators, by analyzing the characteristic indicators in each region and comparing the data differences between adjacent regions, the regions are defined to obtain the regional division results;

[0030] According to the region division result, by analyzing the geometric and texture characteristics of each region, identifying and recording geometric nodes and boundary lines, and obtaining a local feature and boundary information set;

[0031] Based on the local features and boundary information set, an octree index structure is constructed, and a recursive method is used to segment each region and map the three-dimensional spatial relationship of the region to generate a spatial feature partition map.

[0032] Preferably, the steps of obtaining the lighting material mapping relationship are:

[0033] Based on the spatial feature partition map, by analyzing and reconstructing the material data in each partition, the visual and physical properties of the material are reconstructed according to the illumination and reflection characteristics of the area, and the material reconstruction mapping results of each partition are obtained;

[0034] According to the material reconstruction mapping results of each partition, the material properties of each partition are optimized by simulating and adjusting the light intensity and reflection angle under different light sources to obtain the light-material mapping relationship.

[0035] Preferably, the steps of obtaining the material rendering distribution map are:

[0036] Based on the illumination-material mapping relationship, the illumination and material interaction parameters of each partition are analyzed, the surface light reflection intensity within the partition is calculated step by step, the distribution of illumination influence is analyzed, and the surface reflection intensity result is obtained according to the reflection property classification of the material;

[0037] According to the surface reflection intensity result, the material properties of each partition are extracted, the relationship between reflection and material is analyzed layer by layer, the material parameters of each partition are recombined, and a material characteristic matrix is ​​constructed to generate a material rendering parameter set;

[0038] Based on the material rendering parameter set, a multi-dimensional rendering parameter optimization matrix is ​​constructed, and by establishing a relationship between lighting conditions and material characteristics, the material mapping data of each partition is organized to generate a material rendering distribution map.

[0039] Preferably, the steps of obtaining the three-dimensional exhibition rendering result are:

[0040] Based on the material rendering distribution map, multi-point sampling is performed on the lighting conditions of the cloud exhibition environment. By collecting lighting data under different light sources and reflection angles in different regions, the interaction between lighting and materials is analyzed to obtain material reflection data sets from different perspectives.

[0041] According to the material reflection data set, the material reflection effect at each viewing angle is calculated, and the calculation formula is:

[0042]

[0043] Among them, FE ij is the material reflection effect of the i-th partition at the j-th viewing angle, UR ik is the reflection characteristic parameter of the material, L jk is the light intensity, F k is the angular distribution function of the light source, FT k is the surface texture value of the material, P jk is the variation parameter of the interaction between the lighting direction and the material, m is the number of material points associated with the reflection calculation, and a and b are the boundaries of the integration area;

[0044] Based on the material reflection effect, the material rendering parameters and the ambient lighting data are integrated to reconstruct the material on the surface of the three-dimensional exhibit, and the material reflection performance of each viewing angle is adjusted by simulating the lighting effect to generate a three-dimensional exhibition rendering result.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are:

[0046] The present invention collects multi-view image sequences and extracts reflectivity, surface texture and depth information to generate an original optical feature matrix including illumination and reflection characteristics, combines multi-view illumination changes and reflection parameters to form a material feature description set, and realizes a comprehensive expression of the real characteristics of the material. The calculation of the curvature change rate and texture density value is combined with the regional feature index analysis to improve the division accuracy of the three-dimensional geometric structure. Through the extraction of local surface features and contour boundary information and the construction of the octree index structure, a spatial feature partitioning method is established to enhance the data organization efficiency and accuracy of complex areas. The dynamic adjustment of material reconstruction mapping and illumination parameters enables the material rendering parameters to adapt to the detail changes under multi-view conditions, and integrates the rendering parameters with the material properties after illumination enhancement, so that the optical characteristics of the three-dimensional exhibits are more realistic and natural in dynamic observation. The calculation of illumination sampling and material reflection effects ensures the consistency of optical characteristics under different viewing angles. Through the partition optimization and global coordination processing logic, the material expression of the three-dimensional exhibits and the visual appeal of the rendering results are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0049] See also Figure 1 The present invention provides a technical solution, a cloud-based three-dimensional exhibition visualization method, comprising the following steps:

[0050] Collect multi-view image sequences of 3D exhibits, extract reflectivity, surface texture and depth information at each view, calculate reflection characteristics, generate original optical feature matrix, fuse multi-view illumination change characteristics and surface reflection parameters based on the original optical feature matrix, and generate material feature description set;

[0051] Based on the material feature description set, the curvature change rate and texture density value of each region are calculated to generate regional feature indicators. Based on the regional feature indicators, the surface of the three-dimensional exhibit is divided into regions, local surface features and contour boundary information are extracted, an octree index structure is established, and a spatial feature partition map is generated;

[0052] Based on the spatial feature partition map, material reconstruction mapping is performed on each partition, lighting parameters and reflection angles are adjusted, and a lighting-material mapping relationship is generated. Based on the lighting-material mapping relationship, surface reflection intensity and material rendering parameters are calculated, a rendering parameter optimization matrix is ​​constructed, and a material rendering distribution map is generated.

[0053] Based on the material rendering distribution map, the lighting conditions of the cloud exhibition environment are sampled, and the material reflection effects at different viewing angles are calculated. According to the material reflection effects at different viewing angles, the material rendering parameters are integrated, and the material surface of the three-dimensional exhibits is reconstructed and the lighting is enhanced to generate the three-dimensional exhibition rendering results.

[0054] The steps to obtain the original optical feature matrix are:

[0055] Collect multi-view image sequences of 3D exhibits, extract reflectivity, surface texture and depth information at each viewpoint, and integrate the data to generate a multi-view optical data set;

[0056] According to the multi-view optical data set, the reflection characteristic score at each viewing angle is calculated using the following formula:

[0057]

[0058] Among them, P ij Score the reflectance property of the i-th viewing angle and j-th surface area, R ij is the reflectivity at the i-th viewing angle, WT i is the surface texture parameter of the i-th viewing angle, D j is the depth value of the jth region, E ij is the reflection characteristic enhancement factor;

[0059] According to the reflectance characteristic scores, all reflectance characteristic scores are integrated into a matrix form to generate an original optical characteristic matrix.

[0060] Specifically, the process of acquiring multi-view image sequences of three-dimensional exhibits is to arrange several high-resolution cameras around the exhibits, each camera captures image data at different angles under the same lighting conditions, and the exposure time and focal length parameters of the camera are calculated by comparing the distribution of exposure compensation values ​​between 0 and 2EV in conjunction with camera contrast adjustment. After shooting, the image is denoised, and the ambient light interference and sensor noise are processed by a local gradient denoising method. After processing, the image clarity index is checked, and the reflectivity, surface texture, and depth information are recorded with reference to the light energy data provided by the light meter, and the reflectivity is controlled within the range of 0.0 to 1.0. The texture information is extracted within the range of 0 to 255 according to the grayscale gradient and texture direction. The depth information is acquired by means of structured light projection and monitoring by a distance sensor, and the results are written into a matrix data structure by analyzing the distance distribution in multiple measurements and arranged according to the view number. Finally, these reflectivity, texture features, and depth values ​​together constitute a multi-view optical data set.

[0061] The benefit of the formula is that it combines surface texture parameters and depth information with a reflective characteristic enhancement factor, making the comprehensive reflection level evaluation under multi-view shooting conditions more accurate. This formula can be used to obtain a more refined reflective characteristic score, which can be used by subsequent lighting modeling and material rendering to perform more refined processing.

[0062] Parameter R ij It is the reflectivity at the i-th viewing angle, which is obtained by comparing the transmitted light source power with the light energy received by the camera at the same moment. The ratio is mapped to the range of 0.0 to 1.0 and recorded in the matrix with the viewing angle number and area number. This value can describe the light reflection intensity at a specific location.

[0063] Parameter WT i is the surface texture parameter of the i-th viewing angle. By calculating the grayscale difference and edge gradient amplitude for each image block, recording the texture direction distribution and summarizing the average value, and completing the normalization process in the range of 0 to 255, it is bound to the viewing angle number to generate WT i The numerical value of .

[0064] Parameter D j is the depth value of the jth area. The actual measurement results are recorded using structured light projection and distance sensors. The average value of multiple detection data is taken and the spatial reference of the laser interferometer is aligned. The measured value is controlled within the range of 1 mm to 300 mm and then associated with the area number for matrix indexing.

[0065] Parameter E ijIt is the reflection characteristic enhancement factor. By continuously recording the coupling effect of ambient brightness distribution, surface texture and reflectivity, the reflection gain coefficient is extracted and piecewise fitted with reference to the relationship between the incident angle of light and the cosine of the normal line. A value is selected in the range of 0.8 to 1.2, the collection conditions are indicated, and then it is matched with (i, j).

[0066] Calculation process:

[0067] Given the measurement results for a certain viewing angle i = 2 and surface area j = 5, R 2,5 =0.65, WT 2 =180.0, D 5 =45.0, E 2,5 =1.05, the calculation steps are as follows:

[0068] The first step, R 2,5 WT 2 =0.65×180.0=117.0,

[0069] Step 2:

[0070] The third step is to add the two items together to get 117.0+0.0222=117.0222.

[0071] Step 4.

[0072] Step 5, 10.8177×1.05=11.3586, so P 2,5 ≈11.3586;

[0073] The result shows that the reflection characteristic score for viewing angle 2 and surface area 5 is 11.3586, with values ​​between 10 and 12 indicating strong reflection and values ​​below 5 indicating weak reflection.

[0074] For the generated multi-dimensional reflective characteristic scoring data, it is necessary to build a corresponding relationship according to the viewing angle index and the area number, read each scoring item and match it with the previously recorded reflectivity, surface texture and other values, and combine the scoring value with the corresponding (i, j) label. The process includes checking the numerical range of the score one by one. If there is a record exceeding 200 or less than 0, it is marked as abnormal and eliminated. For potential extreme deviation data, the difference between the score and the adjacent viewing angle is detected. If the difference is higher than 10 points for multiple consecutive times, the acquisition process is traced back and the original optical measurement record is re-compared. After the screening is completed, the scoring items that meet the requirements are arranged from 1 to n in the order of viewing angle, and a one-to-one mapping is formed with the corresponding area number. When supplementing the vacant scoring items, refer to the camera shooting timeline to check whether there is a valid measurement record. If available, it is imported into the scoring list. If the record is repeated, the scoring number is compared to abandon the redundant items. The finalized scoring data is merged in the form of multiple array blocks and attached to the end of the reflective characteristic scoring items.

[0075] Based on the above screening and summary of available scores, the score items of each viewing angle and area are converted into a matrix. First, several rows are set according to the viewing angle number, and then the area number is matched to the corresponding column position. After checking whether each row is filled in, the missing scores are completed, and the confirmed reflectivity and surface texture records are combined with the time index to fill in the gaps. For repeated scores, the score number comparison method is used to retain the entries with higher resolution. After completing the row and column mapping of the matrix, a multidimensional score table can be generated and then marked as the original optical feature matrix. Check whether the data header correctly identifies the row and column correspondence, and at the end, the sorted matrix is ​​included in the subsequent processing link.

[0076] The steps to obtain the material feature description set are:

[0077] According to the original optical feature matrix, the illumination change characteristics and surface reflection parameters in each viewing angle are extracted, each element value in the matrix is ​​analyzed, the illumination change and surface reflection law are identified, the analyzed optical feature data is called, and a multi-view illumination feature set is generated;

[0078] Based on the multi-view illumination feature set, the interaction between illumination features and surface reflection parameters is analyzed from each viewpoint, the effect of adjusting illumination parameters on reflection features at each viewpoint is calculated, and the results of multiple interactions are integrated into a comprehensive feature matrix that integrates illumination features and reflection parameters.

[0079] According to the comprehensive feature matrix, the interaction between each illumination parameter and the reflection characteristic parameter is mapped, the comprehensive feature matrix is ​​parsed row by row, the reflection characteristic mapping matrix is ​​generated, and the material feature description set is constructed according to the matrix data content.

[0080] Specifically, based on the original optical feature matrix obtained above, by retrieving the matrix rows corresponding to each viewing angle and parsing the illumination value and surface reflection record contained in each matrix element, the combination information of the viewing angle number and the surface area number is selected to compare the numerical distribution therein, and the segmented comparison method is used to check the changing trend of the illumination value in the range of 10 to 500 row by row and compare the numerical state of the reflection record in the range of 0 to 1. If some element values ​​are not within the pre-configured valid range, the viewing angle element is marked as abnormal and removed. At the same time, the adjacent upstream and downstream rows are checked. After retaining most of the records that meet the requirements, the difference between the illumination and surface reflection values ​​of each row is calculated. The position where the difference is too large is regarded as a mutation point and recorded in a segmented table. Subsequently, according to the statistical results of the number of mutation points that appear multiple times in the same viewing angle, these statistical values ​​are combined with the original illumination and reflection data to identify the illumination change direction. In the specific operation, the correlation coefficient is calculated according to the mutation point index and the corresponding element value of the segmentation table. When the correlation coefficient is greater than the preset threshold, it means that the mutation segment may be strongly correlated with the surface reflection parameter. The preset threshold is obtained by counting and averaging the 100 groups of matrices collected in the early stage, which is an integer range of 5 to 20. If it exceeds this value, the illumination value and reflection value of the position are uniformly put into the candidate list of illumination change to complete the marking, and then the records in the adjacent matrix columns are re-read to compare whether they also have similar change amplitudes. If multiple consecutive columns meet the change amplitude, this area is identified as a feature fragment and its perspective and area label are attached to the list. Finally, after counting all the rows and columns, a parsing result containing the pairing relationship between illumination change and surface reflection is generated. When outputting, the parsing result is used as the comprehensive data of illumination change characteristics and surface reflection laws and retrieved together with the corresponding perspective information, and finally a multi-perspective illumination feature set is obtained.

[0081] Based on the multi-view illumination feature set obtained above, with reference to the illumination and reflection value distribution recorded for each view, the illumination parameters of each view are extracted in order from low to high and compared one-to-one with the aforementioned surface reflection information. The numerical ratio of the illumination parameter increment and the corresponding surface reflection parameter is extracted as the interaction index. When the index exceeds 3, the corresponding view is temporarily stored in the analysis table. When the cumulative interaction degree exceeds 3 for more than 5 items and the difference between the adjacent view numbers does not exceed 2, it is considered that there is a strong correlation between illumination and reflection. The correlation information is recorded in another set of tables and the subsequent view angles are traversed. If the illumination parameter is lower than If the reflection value is still higher than 0.6, the mutation point index registered previously will be compared to determine whether it can be included in the abnormal list. The above operation will accumulate a large number of interaction values ​​and related records after each view angle is completed. By cross-checking and counting these results with the previous candidate list of illumination changes, the number of times the illumination increase and the reflection value increase synchronously, as well as the number of times the illumination decreases and the reflection value is still high, are identified. These statistics are then output as multiple interaction results, and a row-by-row scanning method is performed to merge all the interaction degrees and additional annotations to generate a record of the fusion of illumination characteristics and reflection parameters. Finally, the record is organized into a comprehensive feature matrix.

[0082] According to the comprehensive feature matrix obtained above, read the numerical pairing of illumination parameters and surface reflection parameters in each row and map the pairing relationship. The execution method is to divide the incremental range of illumination parameters into several segments and compare them with the fluctuation of corresponding surface reflection parameters one by one. Set a mapping flag for the records with increments in the range of 5 to 50 and corresponding reflection parameter fluctuations greater than 0.2, and fill the flag in the same row and column position in the mapping matrix. In this way, a preliminary table containing the corresponding relationship between illumination and reflection can be obtained. Then, check row by row whether there is any mapping overlap in the preliminary table. When some adjacent rows are repeatedly marked in the same column, they are merged and adjusted to the same continuous segment. If the mapping segment is too dense during the operation, it is necessary to check whether there is any record conflict between the illumination value and the reflection value in the original feature matrix. After confirming that there is no conflict, all the final mapping entry information is saved in the reflection characteristic mapping matrix and a material feature description set is generated in combination with the column index number.

[0083] The steps to obtain regional characteristic indicators are as follows:

[0084] Based on the material feature description set, a hierarchical structure of three-dimensional space is constructed, and the material data is organized into a spatial hierarchical model through hierarchical processing to obtain a three-dimensional spatial hierarchical structure;

[0085] According to the three-dimensional spatial hierarchical structure, the curvature change rate of each area is calculated using the following formula:

[0086]

[0087] Among them, CR ij is the curvature change rate of the jth region of the i-th layer, K ik is the curvature value of the kth point in the i-th layer, is the average curvature of the i-th layer, θ ik is the texture value of the kth point in the i-th layer, is the average value of the texture of the i-th layer, and n represents the total number of points in the i-th layer;

[0088] Based on the three-dimensional spatial hierarchical structure, the texture density value of each area is calculated using the following formula:

[0089]

[0090] Among them, WD ij is the texture density value of the jth region of the i-th layer, T jl is the texture value of the lth point in the jth region, and wm represents the total number of points in the jth region;

[0091] The curvature change rate and texture density value are combined to generate regional feature indicators.

[0092] Specifically, based on the material feature description set obtained above, the three-dimensional distribution data corresponding to the material is distinguished according to the layer number by reading the spatial coordinate record and texture and reflection information identification of each material entry. First, the upper and lower order of the material in the three-dimensional coordinate system is counted according to the index value of the material position, so as to distinguish the layer range from the 1st layer to the nth layer. Then, the coordinates of all material points in the same layer are aggregated, and the texture and depth observation data of each material point are compared. The discrete distribution of texture values ​​in the range of 0 to 255 is recorded as a local texture mapping table, and the difference information of depth values ​​from 10mm to 500mm is summarized in a distance positioning table. Then, the entries with the same or similar material properties in the same layer are merged. When the merged materials are When the number of quality entries exceeds an empirical threshold, adjacent coordinate groups are further disassembled or merged. This empirical threshold is obtained by statistical analysis of past projects and is usually between 100 and 300 entries. If a segment after stratification still contains multiple discrete fragments in the horizontal direction, the indexing and texture mapping tables of these fragments are checked and the repeated or overlapping parts are merged. If it is found that both the texture value and the depth value have changed significantly, they are recorded in the cross-regional border list, and the processed layers are merged into the initially set three-dimensional space level. The adjacency check is performed on the distribution of material points at the same level, including considering the coordinate difference between adjacent material points as a group of connected fragments when it is no more than 2mm, and recording the texture range and number of illumination measurements of the fragment, to finally obtain a three-dimensional space hierarchical structure.

[0093] formula The benefit of the formula is that by integrating the discreteness of the curvature value and the texture value at the same level, the overall fluctuation of the geometry and texture of the jth region of the i-th layer can be measured in a unified mathematical way, thereby providing a refined numerical reference for the subsequent extraction of spatial feature indicators.

[0094] Parameter K ik The acquisition steps are as follows: extract the surface curvature value of the kth point of the i-th layer from the geometric scanning data obtained by the actual three-dimensional measurement, and record it in a real number range usually from 0 to 10. This range is quantified by the minimum and maximum curvatures measured in the experiment of the actual bending structure. In order to obtain the specific value, it is necessary to first determine the adjacency relationship of each point in the three-dimensional grid during scanning, and calculate the local curvature using the point normal direction offset of the triangular grid. After counting all the observed values, correspond them to the coordinate system of the material. Each curvature value is confirmed by matching the curvature statistics table with the point number. ik The specific value is then stored in an index corresponding array to ensure that all values ​​are within the valid range. When the measured curvature exceeds the established curvature upper limit, repeated detection must be performed and the cause must be investigated. Only after confirming that there is no measurement deviation can the point be officially recorded in the curvature data.

[0095] parameter The acquisition steps are as follows: in the i-th layer, all curvature values ​​are counted, and the K of each material point is calculated. ik After adding, divide by the total number of points n to get the average value. This process requires first confirming whether the number of material points in the i-th layer is greater than 10. If it is greater than 10, record all the curvatures, sum them up and divide by n to get Where n is the number of points in the i-th layer. The entire statistical process is performed by the software automatically reading the curvature measurement results and performing the addition operation, and then using the arithmetic average to obtain The value is then saved in the average curvature list corresponding to layer i, so that the average value can be quickly retrieved in subsequent calls to participate in the calculation of the curvature change rate.

[0096] Parameter θ ik The acquisition steps are as follows: according to the texture analysis results obtained above, for the kth point of the i-th layer, the texture intensity value of the point is derived from the image mapping and local texture detection results, and the texture grayscale information distribution originally in the range of 0 to 255 is mapped to a floating point value of the same resolution through the automatic texture recognition program, usually accurate to the unit between 0 and 255, and these texture values ​​are read in turn for the material points of the i-th layer and stored as index k. When calculations need to be performed, query and directly call θ from the corresponding index. ikBecause texture and curvature have spatial coordinate indexes, it can ensure that the value corresponds to the point position one by one. Each measurement uses the same sampling frequency and shooting angle to ensure that the matching of texture data and three-dimensional coordinates does not cause duplication or omission.

[0097] parameter The acquisition step is to perform an averaging operation on all texture values ​​associated with the i-th layer, by averaging the θ of all material points in this layer ik The sum is then divided by n, where n is the total number of material points in this layer, and the result is recorded as When saving, an index information is added to indicate that the average value belongs to the i-th layer for easy later search. It is the same as the curvature mean value in the calculation process. Used together, it is necessary to ensure that all θ ik All have been determined and are within the valid range. If it is detected that individual texture values ​​deviate too much from 0 to 255, it is necessary to review or supplement the texture detection data of that point to avoid affecting the accuracy of the average value.

[0098] The parameter n represents the total number of points in the i-th layer. The number of points can be counted by pre-sorting the coordinate indexes corresponding to all material points in the i-th layer. When performing 3D measurement, the spatial grid formed by all material points has been recorded and each point has been assigned a layer number and a sequence number in this layer. After loading the information of these points, the total number of points n contained in the layer can be calculated. Usually in a complete 3D scene measurement, the number of points in each layer will range from hundreds to thousands. In order to obtain more accurate area division, it is necessary to ensure that each layer has sufficient points for statistics. When the number of points in a layer is too small, it may be necessary to merge it into the adjacent layer or adjust the acquisition strategy.

[0099] Calculation process:

[0100] First, a specific area is given. Let the i=2 layer contain n=5 material points. The curvature value of all material points in the j=3 region is recorded as K. 2,1 =1.2,K 2,2 =2.1,K 2,3 =3.0,K 2,4 =2.5,K 2,5 =1.7, texture value is θ 2,1 =85,θ 2,2 =92,θ 2,3 =120,θ 2,4 =100,θ 2,5 =80, the calculation process is as follows:

[0101] The first step is to calculate and

[0102]

[0103] The second step is to calculate each and sum:

[0104] To K 2,k :(1.2-2.1) 2 +(2.1-2.1) 2 +(3.0-2.1) 2 +(2.5-2.1) 2 +

[0105] (1.7-2.1) 2 =0.81+0+0.81+0.16+0.16=1.94;

[0106] For θ 2,k :(85-95.4) 2 +(92-95.4) 2 +(120-95.4) 2 +(100-95.4) 2 +(80-95.4) 2 =(-10.4) 2 +(-3.4) 2 +(24.6) 2 +(4.6) 2 +(-15.4) 2 =108.16+11.56+605.16+21.16+237.16=983.2;

[0107] The third step is to add and divide by n=5:

[0108]

[0109] The fourth step is to perform square root operation on the result:

[0110]

[0111] Therefore, we can get CR 2,3 ≈14.0365;

[0112] The result shows that the curvature change rate of the third area of ​​the second layer is about 14, which means that the curvature and texture in this layer and this area have relatively obvious fluctuations for the subsequent calculation of regional characteristic indicators. The larger the value, the greater the discreteness.

[0113] formula The benefit of the formula is that by performing square and square root calculations on the texture values ​​of each point in the same area and averaging them, the density of the texture in the area at the local scale can be measured, providing another quantitative description related to surface details for the regional feature index.

[0114] Parameter T jl The acquisition steps are as follows: for the jth region, all material points are listed and their texture intensities are read. The texture values ​​of the corresponding material points are selected from the acquired image or scanned texture data. Each texture value is between 0 and 255. It is obtained by grayscale or feature analysis of the original image, and then archived into a list and matched with the index of the lth point to form T. jl When it is detected that the texture values ​​of some material points deviate from this range, the texture results of the collection batch of this point should be compared with those of other material points in the same batch, and check whether there are any recording errors or omissions. After confirming that all data are legal, unified calculations can be performed.

[0115] The parameter wm represents the total number of material points in the jth area, which is used to normalize the accumulated result of the texture value. When counting, it is necessary to check the number of points in the area. Usually, the same area may contain dozens to hundreds of points in one scan. The specific number of wm can be obtained by accumulating these points within the area through indexing. If this number is less than 10, it is necessary to confirm in the system whether the divided area is too small. Otherwise, the area is merged or the acquisition strategy is updated to ensure that each area has sufficient data support.

[0116] Calculation process:

[0117] Assume that the j=5th region in layer i=2 contains wm=4 texture points, and their texture values ​​are recorded as T 5,1 =80,T 5,2 =90,T 5,3 =150,T 5,4 =100, the execution steps are as follows:

[0118] The first step is to 5,l Take the square root and square root separately:

[0119]

[0120] In the second step, add the above results and divide them by wm=4:

[0121]

[0122] Therefore, we can get WD 2,5 =105;

[0123] The result shows that the texture density value of the 5th area in the 2nd layer is 105. The larger value indicates that the texture distribution intensity in this area is relatively obvious. This value can be combined with the curvature change rate for overall comparison when generating regional feature indicators.

[0124] Based on the curvature change rate and texture density values ​​obtained previously, the curvature change rate values ​​and the corresponding texture density records of each layer and each area are read in the system, and the two are paired and counted according to the alignment of the layer number and the area number. In the specific operation, a list of curvature change rates with layer numbers 1 to n is first listed, and the list is traversed in order from 1 to m according to the area numbers. Whenever a curvature change rate entry is encountered, the texture density value of the same layer and area is retrieved and the two are combined into a group of paired items. Subsequently, a two-way comparison is performed on each group of paired items to check the distribution of the curvature change rate between 0 and 50 and the distribution of the texture density between 0 and 255. If the curvature change rate is greater than 40 or the texture density exceeds 200, the group of records is marked in a centralized list, and at the same time, it is compared in adjacent areas of the same layer. For comparison, if the adjacent area also has records with high curvature or texture values, it will be merged with the previous item. In this way, potential regional continuity features can be identified, and the cases where the curvature value remains within 0 to 10 but the texture density is still 150 to 255 are also summarized to form corresponding low curvature and high texture groups. After the statistics are completed, all paired items are indexed according to the layer number and the area number, and a matching list is generated together with their curvature change rate and texture density value to record the distribution range and average of each group. If it is found that some data with the same layer number but different area number differ too much in texture density, they will be added to the comparison sequence of adjacent layers, and then the system will sort all layers and areas according to the defined order to obtain a final synthetic list of regional feature indicators.

[0125] The steps to obtain the spatial feature partition map are:

[0126] Based on regional characteristic indicators, by analyzing the characteristic indicators in each region and comparing the data differences between adjacent regions, the regions are defined and the regional division results are obtained;

[0127] According to the region division results, by analyzing the geometric and texture characteristics of each region, the geometric nodes and boundary lines are identified and recorded to obtain the local features and boundary information set;

[0128] Based on the local features and boundary information set, an octree index structure is constructed, and a recursive method is used to segment each region and map the three-dimensional spatial relationship of the region to generate a spatial feature partition map.

[0129] Specifically, based on the regional feature indicators obtained above, we first extract the curvature change value and texture density value in each region, compare the high and low ranges of these values ​​and refer to the threshold standards determined previously. When the curvature change value is between 10 and 50 and the texture density value is between 300 and 1000, the region is identified as a medium level. If the curvature change value is greater than 50 or the texture density value exceeds 1000, it is marked as a high level. If the curvature change value is less than 10 and the texture density value is less than 300, it is marked as a low level. After marking, the data records of adjacent regions are read and compared. The comparison method is to subtract the difference between the curvature and texture of adjacent regions and observe whether it exceeds the range of 5 to 10. If it exceeds this range, it is considered to have a large difference. At the same time, combined with the regional The three-dimensional coordinate positions are checked to see if their intervals in the X-axis and Y-axis directions are less than the preset distance threshold. The threshold is tested on multiple models in 5mm increments and finally fixed at 20mm. When the difference between two adjacent areas is greater than the aforementioned value and the coordinate distance is less than the preset distance threshold, the two areas are separated and marked. If the difference is small, they are merged or the mark of this area is transferred to a more similar neighboring area. After traversing and comparing all areas, a list of regional hierarchical marks is formed and associated with the corresponding regional indexes. If necessary, the high-level or low-level phenomena in multiple consecutive areas will be rechecked and the adjacent relationship will be recorded. When it is confirmed that the marks are consistent and the adjacent difference meets the above standards, the regional classification is locked in the final list to obtain the regional division result.

[0130] According to the area division results obtained previously, the geometric coordinate distribution and texture-related values ​​of each marked area are read one by one to check whether there are obvious surface protrusions or depressions and to count the normal angles with the surrounding points. When the angle is in the range of 0° to 10°, it is considered that the point is close to the surrounding points in geometric direction. If it exceeds 40°, it indicates that there is a more obvious geometric jump at this position. Combined with the texture information, check whether there is a normal angle exceeding 40° in the concentrated section with a grayscale amplitude of more than 150. If this is found, it is recorded as a suspicious geometric node. Then, the 3×3 or 5×5 sampling points around the node are expanded and recorded to find continuous high-contrast textures. When encountering When multiple such nodes are in the same local range, the Euclidean distance and angle distribution between the points are included in a geometric node confirmation table. If there are node pairs with a distance less than 10 mm and an angle difference within 10°, they are merged into a larger feature block. After all feature blocks are confirmed, the outer edges of each block are compared and the boundary lines are recorded in a curve fitting manner. If the boundary line exceeds 10 mm in length, it is maintained as an independent line segment. If the difference between two line segments is only 3 mm, they are merged into the same boundary description. After checking layer by layer in all areas through the above process, the obtained geometric node coordinates and boundary line coordinates are summarized into a local feature and boundary set to obtain a local feature and boundary information set.

[0131] Based on the local features and boundary information set obtained above, a coordinate index is first generated for each boundary line and each geometric node, and then these indexes are divided into several sub-areas according to the spatial coordinate range. When constructing the octree index structure, the three-dimensional space is divided into eight sub-cubes and the feature points therein are located. When the number of feature points in a sub-cube exceeds the specified threshold of 30, the sub-cube is split into eight finer sub-cubes again. Each split will record the split depth and control the depth to be no more than 5 times the first-level division depth. If it is found that the number of feature points in multiple consecutive sub-cubes is less than 10, these sub-cubes are merged into a larger block. After the split is completed, the spatial relationship between each sub-cube and its parent and sibling nodes is coordinate mapped. The splitting and mapping process is repeatedly performed in a recursive manner. If the coordinate interval of the feature point is detected outside the current sub-cube, the point is moved to the upper cube and the cube is re-divided. After traversing all marked areas and locating the feature points in all sub-cubes, a three-dimensional spatial index table is generated. Finally, the index table structure is mapped to the overall spatial scene to generate a spatial feature partition map.

[0132] The steps to obtain the lighting material mapping relationship are:

[0133] Based on the spatial feature partition map, the material data in each partition is analyzed and reconstructed, and the visual and physical properties of the material are reconstructed according to the illumination and reflection characteristics of the area to obtain the material reconstruction mapping results of each partition;

[0134] According to the material reconstruction mapping results of each partition, the material properties of each partition are optimized by simulating and adjusting the light intensity and reflection angle under different light sources to obtain the light-material mapping relationship.

[0135] Specifically, based on the spatial feature partition map obtained above, the material data in each partition is compared item by item and combined with the previously measured light intensity records and surface reflection data, the reflection coefficient in the partition is matched with the texture distribution parameters. If the reflection coefficient of the partition exceeds 0.8 and the surface texture direction is relatively simple, it is judged as a high reflection section. If the texture direction is relatively complex and the reflection coefficient is in the range of 0.3 to 0.6, it is judged as a medium reflection section. According to the above classification, the illumination feature data corresponding to each partition is disassembled, and then the material density and transparency information of the corresponding partition are analyzed. The density information comes from the previous monitoring of the material thickness and is recorded in 10kg / m 3 Up to 3000kg / m 3 If the density value of a certain area is between 2000kg / m 3 Up to 3000kg / m 3 If the transparency is below 0.2, it is regarded as an opaque material. If the transparency is greater than 0.8 and the surface reflection coefficient is relatively high, the refractive index of the section needs to be additionally recorded to reconstruct the material properties. After the above properties are sorted, the material data of each partition is integrated into a mapping table. The comparison method is used to determine whether the visual and physical properties of the material meet the preset interval standards. The interval standards here are based on multiple laboratory measurements and comparison of thickness, density and refractive index. If all parameters fall within a reasonable range, the material characteristics of the partition are marked in the mapping table and a comprehensive material index is calculated as a reconstruction reference. When some characteristics in the partition have extreme values, they are separately included in the group to be subdivided for inspection and written into the corresponding mark column. After all partitions have completed the above inspection and analysis, the material reconstruction mapping results of each partition are sorted out.

[0136] According to the material reconstruction mapping results of each partition, the illumination parameters and reflection angle value lists of each partition are read separately and the illumination adjustment table is established. First, the light source brightness is selected between 100 and 1000 lux as the preliminary range, and the surface reflectivity of each partition is compared with the brightness within this range and recorded in the corresponding column. If the brightness is greater than 600 lux and the reflectivity is above 0.7, the partition is marked as a strong reflection partition. If the brightness is lower than 300 lux and the reflectivity is below 0.4, it is regarded as a weak reflection partition. Then, several different illumination incident angles are tested for the strong reflection partition, and the incident angles are recorded between 0° and 60°. When it is found that the reflection parameters and refractive index of the partition at the incident angle are greatly different, the difference value will be recorded in the result. And determine whether there is a significant change in the range of 5 to 15 degrees. If there is a significant change, the partition is marked as a sensitive area and its material type is checked with reference to the reconstructed mapping result. If the material type is a high-refractive index material, its transparency needs to be corrected. At the same time, check whether the partition has too high a value in specular reflection at an incident angle of 15°. If it exceeds the specified threshold of 0.9, the light loss at this angle is re-recorded and the result is incorporated into the lighting adjustment table. Finally, after all partitions have completed lighting adjustment and reflection angle measurement, these two parts of the value are integrated to obtain a mapping set used to describe the correspondence between lighting and materials, and through the adjustment options set for each partition, an association is established between lighting and reflection behavior to form a summary list and mark it as a lighting-material mapping relationship.

[0137] The steps to obtain the material rendering distribution map are:

[0138] Based on the mapping relationship between light and material, the interaction parameters between light and material in each partition are analyzed, the surface light reflection intensity in the partition is calculated step by step, the distribution of light influence is analyzed, and the surface reflection intensity result is obtained according to the reflection property classification of the material;

[0139] According to the surface reflection intensity results, the material properties of each partition are extracted. By analyzing the relationship between reflection and material layer by layer, the material parameters of each partition are recombined, and the material property matrix is ​​constructed to generate a set of material rendering parameters.

[0140] Based on the material rendering parameter set, a multi-dimensional rendering parameter optimization matrix is ​​constructed. By establishing a relationship between lighting conditions and material properties, the material mapping data of each partition is organized to generate a material rendering distribution map.

[0141] Specifically, based on the illumination-material mapping relationship obtained previously, the illumination parameters and material interaction records corresponding to each partition are read, and the reflection coefficient previously recorded in the range of 0 to 1 is compared with the actually measured incident light energy value. The incident illumination intensity and reflection parameter pairs of each partition are listed in the statistical table. If the incident light intensity of the partition is greater than 400 lux and the reflection coefficient is higher than 0.7, the reflection surface of the partition is measured in sections to determine whether there is a local area with a reflection value exceeding 0.9. If it is found that some of the measurement results exceed the preset threshold of 0.9, this partition is listed as a high-brightness partition and the coordinate range and feature label of the partition are noted in the record. Then, the illumination intensity and reflection coefficient of the remaining areas in the same partition are summarized in steps of 0.1 to analyze whether there is an obvious reflection drop in the local range. If the drop is greater than 0.3, the local range is further split and formed into independent entries. At the same time, the refraction of the material and Texture density value, record the partitions with texture density between 300 and 600 as medium-density materials, if it is greater than 800, it is considered high-density. After completing the disassembly and statistics in the previous step, gradually accumulate the local reflection values ​​of each partition and calculate the average reflection intensity. The average reflection intensity obtained by measurement is compared with the highest and lowest reflection values ​​in the partition. For areas with large differences, the transition sections with reflection coefficients between 0.2 and 0.4 are retrieved again in layers to check whether they are associated with texture features. If it is found that the difference between texture features and reflection parameters is frequently higher than 0.5, this position is marked as a high-change rate section and a mapping relationship is established with the lighting parameters in the overall mapping. Finally, the local reflection calculation results of each partition are added and combined with the verification value of the transition area at the edge of the partition to obtain the surface light reflection intensity of the partition. When all partitions have completed statistics and comparisons, the reflection intensity information of each partition is summarized to obtain the surface reflection intensity result.

[0142] According to the surface reflection intensity results obtained above, the reflection intensity distribution and material attribute entries corresponding to each partition are extracted one by one. First, the high reflection segment and low reflection segment in the partition are marked with different index numbers, and the gradient of the reflection intensity is measured by the difference between adjacent coordinate positions. If the intensity difference between adjacent positions exceeds 0.4, a gradient label is added to the record entry here. Then, the material density value and refractive index information are retrieved in the same partition. If the density value of a certain area is recorded at 1500kg / m 3 Up to 2500kg / m 3 If the refractive index is between 1.3 and 1.6, it is organized into a medium-density material set with medium refractive properties. If the density value exceeds 2500kg / m 3If the refractive index is greater than 1.7, it is identified as a high-refractive material. These two types of materials are listed in different columns of the data table, and then the correlation between the texture direction and the reflection intensity is compared for the data in each column. When the texture direction changes by 90° and the reflection intensity changes in the range of 0.5 to 0.8, the segment here is marked as the texture and reflection transition zone. When the number of transition zones exceeds a certain threshold, such as 5, and the overall area accounts for more than 30% of the partition area, they are merged into a large-area transition zone for further recording. After merging these marks, the reflection intensity, material density, refractive index and texture direction are finally combined through hierarchical retrieval, and the material parameters of each partition are recombined. These parameters are integrated into a two-dimensional table by segment and then expanded to a three-dimensional matrix. The partition coordinates and material property entries are marked to construct a complete material property matrix and generate a material rendering parameter set after the recording is completed.

[0143] Based on the material rendering parameter set obtained above, a multi-dimensional index is first created for each partition to correspond the lighting conditions to the material characteristics. The lighting intensity and incident angle are placed in the row dimension of the matrix, the reflection value and refraction data are placed in the column dimension, and the texture information and density parameters are referenced in the depth dimension for linkage. After these dimensions are combined, it is checked whether there is a situation where the same incident angle corresponds to multiple material reflection values ​​under a certain fixed light intensity. If so, the point is marked in the matrix as a possible layered record, and then the layered record is cross-compared. If the data distribution of the surrounding adjacent units differs from the overall data by more than 0.3, it is marked as a heterogeneous point. If the adjacent difference is in Points between 0.1 and 0.3 are marked as micro-difference points, and the distribution of heterogeneous points and micro-difference points is summarized into the matrix again. They are searched item by item from low to high according to the light intensity and the statistical results are accumulated. If abnormal distribution occurs frequently, it is regarded as a position where the light and material have a greater mutual influence. After counting the differences of all points, the partition coordinates and material type are referred to to determine which partitions have frequent high changes. After determining these key partitions, the light intensity and reflection and refraction information are mapped again to generate a set of parameter lists for multi-angle rendering, which are added to the optimization matrix, and the light and material mapping data of all partitions are organized to finally generate a material rendering distribution map.

[0144] The steps to obtain the 3D exhibition rendering results are as follows:

[0145] Based on the material rendering distribution map, the lighting conditions of the cloud exhibition environment are sampled at multiple points. By collecting lighting data under different light sources and reflection angles in different regions, the interaction between lighting and materials is analyzed to obtain material reflection data sets from different perspectives.

[0146] According to the material reflection data set, the material reflection effect at each viewing angle is calculated using the following formula:

[0147]

[0148] Among them, FE ij is the material reflection effect of the i-th partition at the j-th viewing angle, UR ik is the reflection characteristic parameter of the material, L jk is the light intensity, F k is the angular distribution function of the light source, FT k is the surface texture value of the material, P jk is the variation parameter of the interaction between the lighting direction and the material, m is the number of material points associated with the reflection calculation, and a and b are the boundaries of the integration area;

[0149] Based on the material reflection effect, the material rendering parameters and ambient lighting data are integrated to reconstruct the material on the surface of the three-dimensional exhibit. The material reflection performance of each perspective is adjusted by simulating the lighting effect to generate a three-dimensional exhibition rendering result.

[0150] Specifically, based on the material rendering distribution map obtained above, combined with the coordinate index and lighting conditions indicated in each partition, several sampling points are set in the cloud exhibition environment according to the partition coordinate position. Each sampling point changes the light source direction in sequence within the incident angle range of 0° to 60° and records the actual illuminance value. When comparing the illuminance value with the reference reflection data in the partition, it is first checked whether the light intensity is higher than 100 lux and lower than 800 lux. At the same time, the refractive index and texture density classification of the partition are compared. If the refractive index is between 1.3 and 1.6 and the texture density is higher than 500, it is determined that the partition may have a relatively obvious reflection change under this lighting angle. The brightness and reflection ratio of such partitions are counted to see if it exceeds 0.7. When the ratio exceeds 0.7, the partition is registered as a high-reflection subset for subsequent further screening. If the ratio is between 0.3 and 0.7, only Recorded in the subset of the middle interval, when the reflectivity is lower than 0.3, it is classified as a lower reflective range, and then continue to repeat similar angle and illumination scans at other sampling points under the same partition. When the same partition has a brightness ratio exceeding 0.7 multiple times at different angles, the partition is marked as a multi-angle high-reflective partition. If the brightness ratio continues to be lower than 0.3 after the angle changes, it is marked as a multi-angle low-reflective partition. After the sampling is completed, the data of all partitions are merged and the distribution of multi-angle values ​​of each partition is recorded in the list. On this basis, the illumination direction is associated with the partition coordinates. When it is found that the brightness ratio of some angles in most partitions increases or decreases significantly, this angle and the relevant partition are marked as key observation objects and the number of areas and the average brightness distribution are counted. Finally, the results of all partitions and sampling points are summarized to obtain material reflection data sets from different perspectives.

[0151] The benefit of the formula is that it combines multiple data such as surface reflection characteristic parameters, light source angle distribution function, and material surface texture value, and quantitatively describes the reflection effect of the material by combining summation and integration. This makes it easy to simultaneously measure local reflection and overall texture effects when analyzing the interaction between light and materials at different viewing angles.

[0152] Parameter UR ik The acquisition step is to store the material reflection characteristics under the i-th partition as a sequence {UR i1 ,UR i2 ,…,UR im In the actual environment, the reflectivity distribution is obtained by comparing the projected power of the light source with the light energy received at the sampling point. After weighted average processing of multiple measurement results, a set of reflectivity values ​​floating between 0.0 and 1.0 is obtained. In order to ensure data stability, control tests are carried out under different incident angles and light source intensities in the same partition, and values ​​that obviously deviate from the average level by more than twice are discarded, and finally merged into a relatively stable reflection characteristic sequence.

[0153] Parameter L jk The acquisition steps are as follows: collect the light intensity at the jth viewing angle, control the illumination range of the light source used to be between 100 and 1000 lux, and use an illuminance meter to perform graded measurements of each incident angle and distance on site. When the distance between the light source and the sampling point is between 0.5m and 2m, the illumination will be recorded multiple times at multiple locations. After all the illumination data are sorted out, a small amount of over-brightness or over-darkness values ​​will be corrected. The L jk Satisfy the corresponding viewing angle and distance conditions and correspond to the corresponding k index items in the table.

[0154] Parameter F k The acquisition steps are as follows: record the angular distribution function of the light source, measure the light deflection angles when the light source illuminates the material points in three-dimensional space, collect these angles and discretize them into several segments, digitize each segment according to the light source divergence range and radiation flux, and map it to a range between 0.5 and 1.5 as a correction factor, where 0.5 indicates a low light concentration and 1.5 indicates a high light concentration. If excessive light beam divergence is observed in a local area, the segment needs to be marked in subsequent steps.

[0155] Parameter FT kThe acquisition steps are as follows: record the surface texture value of the material, analyze the image texture gradient and grayscale distribution of the sampling point in the partition, and quantize it in the range of 0 to 255. When the texture gradient is large or the grayscale distribution concentration is high, the value will increase accordingly and occupy a higher weight in the subsequent reflection calculation. The sampling process needs to be combined with multiple frames of images for balanced processing, and the corresponding contrast calibration is performed for the local illumination intensity deviation, so that FT k Can truly reflect the texture amplitude of the material surface.

[0156] Parameter P jk The acquisition steps are as follows: record the changing parameters of the interaction between the illumination direction and the material, convert the angle between the light irradiation direction and the surface normal vector into a numerical value, and measure it in the range of 0 to 90 degrees using the cosine function or the refraction formula. At the same time, additional coefficients are introduced in areas with high refractive index materials to describe the refraction and return situation. If a large difference is detected between the surface texture and the illumination direction, P jk The values ​​are moderately increased and marked between 0.7 and 1.2, which helps to distinguish local highlight or shadow areas later.

[0157] The steps for obtaining the parameter m are to count the number of material points related to the current partition. When the partition area is large or the material distribution difference is significant, a more intensive sampling frequency scheme will be selected to simultaneously record the illumination and reflection information at several coordinate positions within the partition. All valid material points are merged to obtain m, and the m under different partitions is distinguished in the record table. If the number of points in some partitions is too small, the adjacent coordinate positions will be supplemented in the previous step to ensure that m is sufficient to reflect the actual distribution of the entire partition.

[0158] The steps to obtain parameters a and b are as follows: k With P jk When performing integration, determine the upper and lower limits of the integration interval. Combined with the order of magnitude of the sampling points in the partition and the range of illumination angles, a and b are usually selected in the range of 0 to 0.5 or 0 to 1.0, etc., depending on the distribution of illumination and material data obtained from the previous statistics. After determining the interval, the corresponding FT k ·P jk Discretized integration is performed, and symbolic operations are replaced by numerical summation. When the partition area increases or the sampling point density increases, b will be increased accordingly and the number of discrete points for summation will be increased.

[0159] Calculation process:

[0160] In the first step, let m = 3, select the i = 2nd partition and the j = 1st viewing angle, and the material reflection characteristic parameters in the partition are UR 2,1 =0.65,UR 2,2 =0.55,UR 2,3=0.40, the viewing angle illumination intensity is L 1,1 =500lux, L 1,2 =400lux, L 1,3 = 600lux, the light source angle distribution function is recorded as F 1 =1.20, F 2 =0.90, F 3 =1.10;

[0161] The second step is to multiply and sum the above parameters. Specifically, (0.65×500×1.20)+(0.55×400×0.90)+(0.40×600×1.10)=390+

[0162] 198+264=852;

[0163] The third step is to integrate the material surface texture value and the illumination change parameter, set a = 0, b = 1.0, and let FT 1 =150, P 1,1 =1.10, FT 2 =90,P 1,2 =0.80, FT 3 =200, P 1,3 =1.20, and each sampling point is divided into several equal parts for summation. First, for each pair of FT k ·P 1,k By multiplying, we get 150×1.10=165, 90×0.80=72, 200×1.20=240, which can be summed up to 165+72+240=477, which can be regarded as the cumulative value of discrete integrals in the range of 0 to 1.0, recorded as

[0164]

[0165] Step 4: Add the above sum and integral terms to get 852+477=1329, and take the square root to get

[0166]

[0167] Therefore, FE 2,1 =36.44;

[0168] The result shows that the material reflection effect value of partition 2 at viewing angle 1 is 36.44. When the value is higher than 30, it means that there is obvious reflection brightness and texture superposition phenomenon at this viewing angle. If the subsequent rendering stage needs to distinguish the high brightness, it can be determined based on the result of 36.44 that the current viewing angle has a more significant enhancement effect on the material surface.

[0169] Based on the material reflection effect obtained above, the reflection value corresponding to each viewing angle is matched with the material rendering parameters confirmed before. The material information distributed in different coordinate areas on the same three-dimensional exhibit is selected for superposition and arrangement. The light intensity and material refractive index in each partition are compared to see whether they both reach a ratio threshold greater than 0.5. If both exceed 0.5, a mark is added to the matching list so that the corresponding texture value and refraction parameter can be combined later. If only one of the parameters exceeds the predetermined standard, a medium change is noted when marking. After comparing all viewing angles one by one in the list and counting the reflection and refraction distribution corresponding to each material type, it is combined with the already determined The texture intensity data is merged into the rendering parameter table. Then, when simulating the lighting process, the light source incident angle and material reflection performance of each viewing angle are screened. If it is found that the values ​​of some viewing angles rise under large angles of incidence, and the values ​​are obviously low under small angles of incidence, the reflection intensity is adjusted for the coordinate position corresponding to the viewing angle and the final value is recorded. After scanning all viewing angles, the final rendering brightness, texture overlay effect and three-dimensional coordinate sequence are matched, and integrated into a material map index that can reflect the differences in each viewing angle. Combined with the previously statistical refraction data, additional supplements are made to the key partitions. When all partitions are assigned complete visual parameters, the three-dimensional exhibition rendering results are finally generated.

[0170] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A cloud-based three-dimensional exhibition visualization method, characterized in that: The following steps are involved: Collect multi-view image sequences of three-dimensional exhibits, extract reflectivity, surface texture and depth information at each viewing angle, calculate reflection characteristics, generate an original optical feature matrix, fuse multi-view illumination change characteristics and surface reflection parameters based on the original optical feature matrix, and generate a material feature description set; Based on the material feature description set, the curvature change rate and texture density value of each region are calculated to generate a regional feature index, and the surface of the three-dimensional exhibit is divided into regions based on the regional feature index, local surface features and contour boundary information are extracted, an octree index structure is established, and a spatial feature partition map is generated; Based on the spatial feature partition map, material reconstruction mapping is performed on each partition, lighting parameters and reflection angles are adjusted, a lighting-material mapping relationship is generated, surface reflection intensity and material rendering parameters are calculated based on the lighting-material mapping relationship, a rendering parameter optimization matrix is ​​constructed, and a material rendering distribution map is generated; Based on the material rendering distribution map, the lighting conditions of the cloud exhibition environment are sampled, the material reflection effects at different viewing angles are calculated, and according to the material reflection effects at different viewing angles, the material rendering parameters are integrated to perform material reconstruction and lighting enhancement on the three-dimensional exhibit surface to generate a three-dimensional exhibition rendering result.

2. The cloud-based three-dimensional exhibition visualization method according to claim 1, characterized in that: The steps of obtaining the original optical characteristic matrix are: Collect multi-view image sequences of 3D exhibits, extract reflectivity, surface texture and depth information at each viewpoint, and integrate the data to generate a multi-view optical data set; According to the multi-view optical data set, the reflection characteristic score at each viewing angle is calculated, and the calculation formula is: Among them, P ij Score the reflectance property of the i-th viewing angle and j-th surface area, R ij is the reflectivity at the i-th viewing angle, WT i is the surface texture parameter of the i-th viewing angle, D j is the depth value of the jth region, E ij is the reflection characteristic enhancement factor; According to the reflection characteristic scores, all reflection characteristic scores are integrated into a matrix form to generate an original optical characteristic matrix.

3. The cloud-based three-dimensional exhibition visualization method according to claim 1, characterized in that: The steps for obtaining the material feature description set are: Extracting illumination change characteristics and surface reflection parameters in each viewing angle according to the original optical feature matrix, analyzing each element value in the matrix, identifying illumination change and surface reflection rules, calling the analyzed optical feature data, and generating a multi-view illumination feature set; Based on the multi-view illumination characteristic set, the interactive relationship between the illumination characteristics and the surface reflection parameters is analyzed for each view angle, the influence of the adjustment of the illumination parameters on the reflection characteristics at each view angle is calculated, and the multiple interaction results are integrated into a comprehensive feature matrix integrating the illumination characteristics and the reflection parameters; According to the comprehensive feature matrix, the interaction between each illumination parameter and the reflection characteristic parameter is mapped, the comprehensive feature matrix is ​​parsed row by row, a reflection characteristic mapping matrix is ​​generated, and a material feature description set is constructed according to the matrix data content.

4. The cloud-based three-dimensional exhibition visualization method according to claim 1, characterized in that: The steps for obtaining the regional characteristic index are as follows: Based on the material feature description set, a hierarchical structure of a three-dimensional space is constructed, and the material data is organized into a spatial hierarchical model through hierarchical processing to obtain a three-dimensional spatial hierarchical structure; According to the three-dimensional spatial hierarchical structure, the curvature change rate of each area is calculated, and the calculation formula is: Among them, CR ij is the curvature change rate of the jth region of the i-th layer, K ik is the curvature value of the kth point in the i-th layer, is the average curvature of the i-th layer, θ ik is the texture value of the kth point in the i-th layer, is the average value of the texture of the i-th layer, and n represents the total number of points in the i-th layer; Based on the three-dimensional spatial hierarchical structure, the texture density value of each area is calculated, and the calculation formula is: Among them, WD ij is the texture density value of the jth region of the i-th layer, T jl is the texture value of the lth point in the jth region, and wm represents the total number of points in the jth region; The curvature change rate and the texture density value are combined to generate a regional feature index.

5. The cloud-based three-dimensional exhibition visualization method according to claim 1, characterized in that: The steps of obtaining the spatial feature partition map are as follows: Based on the regional characteristic indicators, by analyzing the characteristic indicators in each region and comparing the data differences between adjacent regions, the regions are defined to obtain the regional division results; According to the region division result, by analyzing the geometric and texture characteristics of each region, identifying and recording geometric nodes and boundary lines, and obtaining a local feature and boundary information set; Based on the local features and boundary information set, an octree index structure is constructed, and a recursive method is used to segment each region and map the three-dimensional spatial relationship of the region to generate a spatial feature partition map.

6. The cloud-based three-dimensional exhibition visualization method according to claim 1, characterized in that: The steps for obtaining the illumination material mapping relationship are: Based on the spatial feature partition map, by analyzing and reconstructing the material data in each partition, the visual and physical properties of the material are reconstructed according to the illumination and reflection characteristics of the area, and the material reconstruction mapping results of each partition are obtained; According to the material reconstruction mapping results of each partition, the material properties of each partition are optimized by simulating and adjusting the light intensity and reflection angle under different light sources to obtain the light-material mapping relationship.

7. The cloud-based three-dimensional exhibition visualization method according to claim 1, characterized in that: The steps for obtaining the material rendering distribution map are as follows: Based on the illumination-material mapping relationship, the illumination and material interaction parameters of each partition are analyzed, the surface light reflection intensity within the partition is calculated step by step, the distribution of illumination influence is analyzed, and the surface reflection intensity result is obtained according to the reflection property classification of the material; According to the surface reflection intensity result, the material properties of each partition are extracted, the relationship between reflection and material is analyzed layer by layer, the material parameters of each partition are recombined, and a material characteristic matrix is ​​constructed to generate a material rendering parameter set; Based on the material rendering parameter set, a multi-dimensional rendering parameter optimization matrix is ​​constructed, and by establishing a relationship between lighting conditions and material characteristics, the material mapping data of each partition is organized to generate a material rendering distribution map.

8. The cloud-based three-dimensional exhibition visualization method according to claim 1, characterized in that: The steps for obtaining the three-dimensional exhibition rendering result are: Based on the material rendering distribution map, multi-point sampling is performed on the lighting conditions of the cloud exhibition environment. By collecting lighting data under different light sources and reflection angles in different regions, the interaction between lighting and materials is analyzed to obtain material reflection data sets from different perspectives. According to the material reflection data set, the material reflection effect at each viewing angle is calculated, and the calculation formula is: Among them, FE ij is the material reflection effect of the i-th partition at the j-th viewing angle, UR ik is the reflection characteristic parameter of the material, L jk is the light intensity, F k is the angular distribution function of the light source, FT k is the surface texture value of the material, P jk is the variation parameter of the interaction between the lighting direction and the material, m is the number of material points associated with the reflection calculation, and a and b are the boundaries of the integration area; Based on the material reflection effect, the material rendering parameters and the ambient lighting data are integrated to reflect the three-dimensional exhibits. The surface material is reconstructed and the material reflection performance of each viewing angle is adjusted by simulating the effect of lighting. Generate 3D exhibition rendering results.

Images