XR virtual scene evaluation method and system based on virtual-real fusion data analysis

Abstract

The invention relates to the technical field of virtual scenes, in particular to an XR virtual scene evaluation method and system based on virtual-real fusion data analysis, and the method comprises the following steps: obtaining the change of an XR point cloud detection node, generating a structure block, mapping an interaction region, extracting an interaction and real number difference value, marking an offset node, and modeling interference in combination with performance parameters. And correcting the weight and outputting a matching label set. According to the method, the path consistency and the response action execution sequence are accurately compared by combining the serial number sequence difference and the logic offset result of the component in the virtual path and the real path, the logic conflict and the abnormal triggering position are positioned, and the interference intensity model is constructed by combining the frame rate fluctuation and the CPU occupancy value of the component in the interaction task; the aggregation identification of key interference points is realized by counting the relationship between fluctuation abnormity and task frequency, a confrontation correction factor set is constructed in an evaluation index dimension, and the influence of interference components on evaluation precision is effectively eliminated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual scene, in particular to an XR virtual scene evaluation method and system based on virtual-real fusion data analysis. BACKGROUND

[0002] The virtual scene technical field belongs to the cross direction of information processing and human-computer interaction, covering digital modeling, graphics rendering, space calculation, behavior construction and other specific technologies. This field mainly studies the generation of realistic three-dimensional scenes through computers, and realizes the simulation and response of human perception system by combining user interaction logic. Its implementation includes building three-dimensional models, setting lighting materials, simulating environmental characteristics, and combining space tracking, perception feedback and other means to realize immersive visual, auditory and behavioral interaction experience in virtual environment. Virtual scenes are commonly used in industrial simulation, teaching and training, cultural display, intelligent driving testing, building visualization and other tasks, emphasizing real-time, immersion, operability and visual logic integrity of the scene, especially in augmented reality, mixed reality and extended reality systems, which have wide engineering value and application depth.

[0003] Among them, the XR virtual scene evaluation method is a method for analyzing and judging the virtual environment construction quality and interaction adaptability in the XR system. Its purpose is to systematically evaluate the virtual scene based on scene complexity, interaction logic rationality, response delay, spatial consistency, user cognitive load and other quantitative indicators to determine whether it meets the set functional goals and scene requirements. This method can be used to improve the construction efficiency and running quality of XR systems, assist developers in optimizing schemes and resource allocation, and is widely used in immersive teaching, simulation training, medical assistance, industrial design and other XR system development and deployment processes involving high interactivity and scene simulation.

[0004] The evaluation method only performs static evaluation based on surface feature quantitative dimensions such as scene complexity, interaction logic rationality and user cognitive load. The existing technology lacks a mechanism for constructing precise mapping relationships between scene structure and interaction response, and cannot timely locate abnormal point positions when the virtual component structure fluctuates significantly or the interaction trigger logic differs, resulting in problems such as index mismatch or evaluation ambiguity in the evaluation conclusion. For example, in complex interaction training, it is difficult to identify the specific component nodes of interaction response sequence deviation, making it difficult for the evaluation result to support subsequent correction and optimization operations. Especially in scenes with high frame rate fluctuations and uneven resource occupation, it is difficult to effectively measure the impact of interaction interference on the overall task flow, making it difficult for the evaluation system to have dynamic and accurate adaptability. SUMMARY

[0005] The purpose of the present application is to solve the shortcomings in the prior art and to propose an XR virtual scene evaluation method and system based on virtual-real fusion data analysis.

[0006] To achieve the above object, the present application adopts the following technical scheme: An XR virtual scene evaluation method based on virtual-real fusion data analysis, comprising the following steps:

[0007] S1: Obtain multi-frame three-dimensional structure point cloud data of interactive construction in the XR virtual scene, perform comparison based on inter-node construction stability value, identify boundary identification nodes, associate and map the structure block with the interactive event response area in the XR virtual scene, and obtain a virtual-real structure mapping block set;

[0008] S2: Based on the virtual-real structure mapping block set, perform number sequence difference value calculation and logical offset position comparison, logically mark and summarize the missing positions of the path segment number appearance sequence dislocation, response action sequence reversal or absence to form an offset marker record table, and obtain a virtual-real path offset component node set;

[0009] S3: Call the virtual-real path offset component node set, calculate the product of frame rate fluctuation rate and CPU occupancy rate, establish the ratio between the interaction frequency of the interference component appearing in the XR scene and the abnormal frame response fluctuation, and generate a structure evaluation interference node grouping table;

[0010] S4: Based on the structure evaluation interference node grouping table, retrieve the evaluation items associated with the interference nodes in the same dimension, compare the deviation direction and amplitude value, extract the difference maximum pair of index values to construct an antagonistic correction factor, and obtain a same-dimension index antagonistic correction coefficient set.

[0011] As a further scheme of the present application, the virtual-real structure mapping block set includes spatial boundary identification, interactive area association label, three-dimensional structure division segment and virtual response positioning index, the virtual-real path offset component node set includes path segment dislocation number, response sequence abnormal label, action logic offset node number and spatial coordinate mapping information, the structure evaluation interference node grouping table includes frame rate response mutation node, processor resource occupancy identification, interaction frequency fluctuation node number and interference behavior frequency factor, and the same-dimension index antagonistic correction coefficient set includes path class index weighting factor, interaction class index suppression factor, action execution offset weight and index direction contrast coefficient.

[0012] As a further scheme of the present application, the acquisition step of the virtual-real structure mapping block set is specifically:

[0013] Obtain multi-frame three-dimensional structure point cloud data of interactive construction in the XR virtual scene, extract the spatial displacement value, connection vector angle value and relative position offset between each structure space node and adjacent nodes in consecutive frames, construct a three-dimensional data feature set of each node, calculate the value domain change interval of the structure space node between different frames, and obtain node stable fluctuation interval information;

[0014] According to the node stability fluctuation interval information, the standard deviation value and the range value of each structural space node in a time sequence are calculated, the standard deviation value and the range value are called to construct a stability comparison item, a structural stability division threshold value is taken as a threshold reference, structural instability of each node is obtained by operation, and the node whose structural instability is greater than the stability division threshold value is identified as a space boundary identification node to obtain a space boundary node identification result.

[0015] Based on the space boundary node identification result, the space point cloud set is regionally segmented according to the position index of the structural boundary identification node, an independent structural block boundary line group is demarcated, and the demarcated space structure block is bound to a point position mapping with a set interactive event response region in the XR virtual scene to obtain a pairing relationship between each structure block and the response region, and a virtual-real structure mapping block set is established.

[0016] As a further scheme of the present application, the formula for obtaining the structural instability of each node is specifically:

[0017]

[0018] Where ΔS represents the structural instability, σ represents the standard deviation normalized value of the node stability fluctuation interval value, θ represents the structural stability division threshold value, R represents the range normalized value of the node fluctuation value in the continuous frame, n represents the frame acquisition quantity, and γ represents the normalized value of the adjacent structure conflict quantity.

[0019] As a further scheme of the present application, the obtaining step of the virtual-real path offset component node set is specifically:

[0020] Based on the virtual-real structure mapping block set, the response trigger number, the path segment stay duration and the interactive action logic sequence number of the component in the interactive task chain in the virtual scene are extracted, a trigger parameter sequence set corresponding to the task chain is constructed, the interactive action number is indexed and sorted according to the task node order, the distribution interval value of the number sequence in the chain is calculated, and task path response sequence distribution information is obtained;

[0021] The task path response sequence distribution information is called to obtain the component execution order number and the response segment number, and the number difference calculation and the sequence offset position comparison are performed with the corresponding virtual path sequence, the path segment number dislocation, the reverse response action sequence or the missing position are screened, and a logical offset marker is given, a marker grouping set is established and the corresponding number paragraph index position is recorded, and path number logical offset record information is obtained.

[0022] According to the path number logic offset record information, the component number corresponding to the inner powder number paragraph is extracted, and the mapped structure node coordinate points in the three-dimensional space are located. The node point set is summarized and the offset corresponding index set is established. The component node set with offset relationship under the task response chain is integrated to establish the virtual-real path offset component node set.

[0023] As a further scheme of the present application, the obtaining step of the structure evaluation interference node grouping table is specifically:

[0024] The virtual-real path offset component node set is called, the CPU occupancy rate, interaction time length and frame rate fluctuation rate data of the corresponding component in the XR virtual scene task library are obtained according to the component number index, the three parameter corresponding sequence set is constructed according to the component number, and the product calculation item between the CPU occupancy rate and the frame rate fluctuation rate is established to obtain the load fluctuation value sequence.

[0025] According to the load fluctuation value sequence, the average change amplitude value of each component corresponding value is calculated according to the component number and is set as a judgment reference. The load fluctuation value change amplitude of each component corresponding to the component is compared with the judgment reference, the component number set whose change amplitude exceeds twice the judgment reference is screened, and the abnormal fluctuation node number set is obtained.

[0026] Based on the abnormal fluctuation node number set, the number of times of triggering interaction of the corresponding component node in the XR scene is extracted and the interaction frequency is calculated. The proportion of the frame rate fluctuation times to the total frame number is calculated, the ratio between the proportion value and the interaction frequency is established, the response interference intensity level of the component in the XR task is judged, and the structure evaluation interference node grouping table is established.

[0027] As a further scheme of the present application, the obtaining step of the same dimension index countermeasure correction coefficient set is specifically:

[0028] According to the structure evaluation interference node grouping table, the task dimension label corresponding to each node is extracted, the dimension categories are divided according to the path navigation type, the interaction accuracy type and the component triggering frequency type, the index set of the node number and the task dimension type is constructed, and the task dimension classification index set is established.

[0029] The task dimension classification index set is called, the evaluation items involved by the corresponding interference node are retrieved according to the index value under each task dimension, the index value pairs with opposite mutual deviation directions are extracted, the deviation direction vector included angle and the deviation amplitude value are calculated respectively, the offset contrast score between the index pairs is obtained by operation, the index pair with the maximum score is selected, the corresponding adjustment factor is constructed, and the countermeasure correction evaluation result is obtained.

[0030] The formula for obtaining the offset contrast score between the index pairs is specifically:

[0031]

[0032] Wherein, F is the offset contrast score between index pairs, V1 is the normalized value of the first evaluation index, V2 is the normalized value of the second evaluation index, D1 is the offset amplitude normalized value of the first evaluation index, D2 is the offset amplitude normalized value of the second evaluation index, and alpha is the association frequency normalized value of the node in the current task dimension.

[0033] Based on the evaluation result of the adversarial correction, the offset contrast score is used as a coefficient to act on the weight value of the original index under the corresponding interference node, the participating indexes are respectively superimposed and updated in the dimension division mode, the updated adjustment coefficient is integrated and associated with the original index system number, and the same dimension index adversarial correction coefficient set is established.

[0034] As a further scheme of the application, the method further comprises the following steps:

[0035] S5: calling the same dimension index adversarial correction coefficient set, updating the weight of the component operation logic, path consistency expression and interactive timing cycle index according to the corresponding correction coefficient, performing comparison between the virtual scene evaluation result and the real operation reference value in the XR system, calculating the evaluation deviation of each link in the virtual and real scenes, and outputting a distinguishable matching level label according to the operation scene, and establishing a virtual and real evaluation matching level label set.

[0036] The virtual and real evaluation matching level label set is specifically an operation consistency level label, a path reconstruction matching level, a behavior synchronization evaluation label and a structure alignment effectiveness label.

[0037] As a further scheme of the application, the acquisition step of the virtual and real evaluation matching level label set is specifically:

[0038] The same dimension index adversarial correction coefficient set is called to extract the original weight values corresponding to the three types of evaluation indexes of component operation logic, path consistency expression and interactive timing cycle, perform product calculation on the original weight values and correction coefficients according to the index attribution dimension, perform proportional update on the indexes in the original dimension, and generate evaluation index updated weight information.

[0039] According to the evaluation index updated weight information, the component operation value, the path matching score value and the interactive response cycle value in the virtual scene are obtained, the corresponding real operation reference value sequence is obtained, the numerical difference value calculation is performed on the same type of indexes, and the reference value mean is divided to obtain the evaluation item deviation amplitude, the index deviation amplitude is summarized and matched with the corresponding operation link, and the virtual and real index deviation value group is obtained.

[0040] Based on the virtual-real index deviation value group, the deviation amount of each evaluation link is compared with the matching threshold range, the operation scene type information is called and the index matching state is extracted according to the scene grouping, the evaluation state level label of each group is output according to the grade division rule, and the virtual-real evaluation matching level label set is established.

[0041] An XR virtual scene evaluation system based on virtual-real fusion data analysis is used to implement the XR virtual scene evaluation method based on virtual-real fusion data analysis.

[0042] The structure mapping processing module obtains multi-frame three-dimensional structure point cloud data of interactive construction in the XR virtual scene, performs comparison based on the inter-node structure stability value, identifies the boundary identification node, associates the structure block with the interactive event response region in the XR virtual scene, and obtains the virtual-real structure mapping block set.

[0043] The offset component identification module performs number sequence difference value calculation and logical offset position comparison based on the virtual-real structure mapping block set, logically marks and aggregates the position of the reverse or missing position of the path segment number appearance order and the response action order, forms an offset mark record table, and obtains the virtual-real path offset component node set.

[0044] The interference node analysis module calls the virtual-real path offset component node set, calculates the frame rate fluctuation rate and the product of the CPU occupation rate, establishes the ratio between the interaction frequency of the interference component appearing in the XR scene and the abnormal frame response fluctuation, and generates a structure evaluation interference node grouping table.

[0045] The correction coefficient evaluation module retrieves the evaluation items associated with the interference node in the same dimension based on the structure evaluation interference node grouping table, compares the deviation direction and the amplitude value, extracts a pair of index values with the largest difference to construct an antagonistic correction factor, and obtains the same-dimension index antagonistic correction coefficient set.

[0046] The evaluation matching analysis module calls the same-dimension index antagonistic correction coefficient set, updates the weight of the component operation logic, path consistency expression and interaction timing period index according to the corresponding correction coefficient, performs comparison between the virtual scene evaluation result in the XR system and the real operation reference value, calculates the evaluation deviation amount of each link in the virtual-real scene, and outputs the discriminable matching level label according to the operation scene aggregation, and establishes the virtual-real evaluation matching level label set.

[0047] Compared with the prior art, the advantages and positive effects of the present application are:

[0048] In the application, by extracting the displacement, angle and position change between the space nodes in the multi-frame structure point cloud, a stability index is constructed, which can dynamically identify the node area with significant fluctuations in the virtual structure, map to the interactive event response space to realize the correlation of virtual and real space structures, combine the number sequence difference and logical offset results of the components in the virtual and real paths, and accurately compare the path consistency and response action execution sequence, locate the logical conflict and trigger abnormal position, and construct an interference intensity model combined with the frame rate fluctuation and CPU occupation value of the components in the interactive task, and through the relationship between the statistical fluctuation anomaly and the task frequency, the key interference points are aggregated and recognized, the counter correction factor set is constructed in the evaluation index dimension, and the original index weight is adjusted according to the set, which effectively eliminates the influence of the interference components on the evaluation accuracy, and outputs the evaluation deviation result and the matching level label, so that the virtual scene evaluation result has high adaptability and interference immunity to the real operation response. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0050] Figure 1 The workflow diagram of the present application;

[0051] Figure 2 The S1 refinement flowchart of the present application;

[0052] Figure 3 The S2 refinement flowchart of the present application;

[0053] Figure 4 The S3 refinement flowchart of the present application;

[0054] Figure 5 The S4 refinement flowchart of the present application;

[0055] Figure 6 The S5 refinement flowchart of the present application;

[0056] Figure 7 The system flowchart of the present application. DETAILED DESCRIPTION

[0057] The technical solutions in the present application will be described below in combination with the drawings.

[0058] In the embodiments of the present application, the words such as "exemplary", "for example", etc. are used to represent an example, illustration, or description. Any embodiment or design scheme described as "exemplary" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "exemplary" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0059] In the embodiments of the present application, "image" and "picture" can be used interchangeably at times, and it should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding" and "corresponding" can be used interchangeably at times, and it should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0060] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1, and when the distinction is not emphasized, the meanings expressed are consistent.

[0061] In order to make the technical problems, technical schemes and advantages to be solved by the present application more clear, the following will be described in detail in conjunction with the drawings and specific embodiments.

[0062] Please refer to Figure 1 The present application provides a technical scheme: an XR virtual scene evaluation method based on virtual-real fusion data analysis, comprising the following steps:

[0063] S1: Obtain multi-frame three-dimensional structure point cloud data of interactive construction in an XR virtual scene, detect the displacement change value between each space node and the surrounding adjacent node in the continuous acquisition frame, the connection vector angle change sequence and the relative position offset, perform comparison based on the node construction stability value, identify the nodes with fluctuation amplitude exceeding the structure stability demarcation value as boundary identification nodes, generate the space structure block interval divided by the boundary identification nodes, and associate and map the structure block and the interactive event response area in the XR virtual scene, to obtain a virtual-real structure mapping block set;

[0064] S2: Based on the virtual-real structure mapping block set, extract the response trigger number, path segment stay duration and interactive action logic sequence number of the component in the virtual scene interactive task chain, obtain the component execution sequence and response segment number in the corresponding real system, perform number sequence difference calculation and logic offset position comparison, logically mark and summarize the path segment number order dislocation, response action sequence reversal or missing position to form an offset mark record table, and extract the spatial node coordinate group corresponding to the component number, to obtain a virtual-real path offset component node set;

[0065] S3: Call the virtual-real path offset component node set, calculate the frame rate fluctuation rate and CPU occupancy rate product according to the CPU occupancy rate, interaction time and frame rate fluctuation rate in the virtual scene task library indexed by component number, and combine the interaction time to form a sequence, extract the corresponding node number according to the interval where the product value change amplitude in the sequence exceeds twice the average change amplitude, establish the ratio between the interaction frequency and abnormal frame response fluctuation of the interference component appearing in the XR scene, and generate the structure evaluation interference node grouping table;

[0066] The CPU occupancy rate represents the system processor resource usage when a specific component or node is called during the evaluation process, which can be obtained through the conventional system monitoring API; The interaction time is the continuous time period that the user triggers, operates or responds to the target component in the interaction stage, which is automatically generated by the event recording system; The frame rate fluctuation rate refers to the unit time change amplitude of the video output frame rate in the interaction process, which is calculated by the difference between the current frame rate and the sliding average value of the historical frame rate divided by the sliding average value, which is commonly used to judge the rendering stability;

[0067] S4: Based on the structure evaluation interference node grouping table, according to the node belonging to the task dimension label classification information, the XR evaluation index set is divided into three categories: path navigation, interaction accuracy and component trigger frequency, the evaluation items associated with the interference nodes in the same dimension are retrieved, and the evaluation index value pairs with opposite direction deviation are obtained in the dimension, the deviation direction and amplitude value are compared, the pair of index values with the largest difference is extracted to construct the counter correction factor, and the factor is applied to the interference source index original weight adjustment coefficient to obtain the same dimension index counter correction coefficient set;

[0068] S5: Call the same dimension index counter correction coefficient set, update the weight of the component operation logic, path consistency expression and interaction timing period index according to the corresponding correction coefficient, execute the comparison between the virtual scene evaluation result in the XR system and the real operation reference value, calculate the evaluation deviation of each link in the virtual and real scenes, and output the discriminable matching level label set according to the operation scene, and establish the virtual-real evaluation matching level label set;

[0069] The virtual-real structure mapping block set includes space boundary identification, interaction area association label, three-dimensional structure division section and virtual response positioning index, the virtual-real path offset component node set includes path segment dislocation number, response sequence abnormal label, action logic offset node number and space coordinate mapping information, the structure evaluation interference node grouping table includes frame rate response mutation node, processor resource occupancy identifier, interaction frequency fluctuation node number and interference behavior frequency factor, the same dimension index counter correction coefficient set includes path class index weight adjustment factor, interaction class index suppression factor, action execution offset weight and index direction contrast coefficient, and the virtual-real evaluation matching level label set is specifically operation consistency level label, path reconstruction matching level, behavior synchronization evaluation label and structure alignment effectiveness label.

[0070] Referring to Figure 2 The obtaining step of the virtual-real structure mapping block set is specifically as follows:

[0071] The multi-frame three-dimensional structure point cloud data of the interactive construction in the XR virtual scene is obtained, the spatial displacement value, the connection vector angle value and the relative position offset between each structure space node and the adjacent node in the continuous frames are extracted, the three-dimensional data feature set of each node is constructed, the value range change interval of the structure space node between the differentiated frames is calculated, and the node stable fluctuation interval information is obtained;

[0072] The multi-frame three-dimensional structure point cloud data of the interactive construction in the XR virtual scene is obtained, which means that 10 frames of data are collected from the depth camera bound with the scene, each frame generates a point cloud set, the set is described by spatial coordinates (x, y, z) and is attached with the adjacent relationship of each node in the current frame, the spatial displacement value of each structure space node and its adjacent node between the current frame and the previous and subsequent frames is extracted according to the adjacent information, the calculation method is the Euclidean distance of the position difference of the same node number between adjacent frames, and the connection vector angle value is calculated, the calculation method is the inverse cosine value of the direction difference of the connection vector between the nodes, and the relative position offset between each node and the adjacent node in the frame sequence is represented as the path projection deviation ratio, the above three types of values constitute the three-dimensional data feature set, and are recorded as the dynamic structure feature information of each node in each frame, the three-dimensional data features of the same node are accumulated frame by frame to construct the change range of each node in the 10 frame sequence, and finally the difference interval of the maximum value and the minimum value is summarized and extracted, the fluctuation range of the node in the space structure evolution is obtained, and the node stable fluctuation interval information is obtained;

[0073] According to the node stable fluctuation interval information, the standard deviation value and the range value of each structure space node in the time sequence are calculated, the standard deviation value and the range value are called to construct the stability comparison item, the structure stability division value is taken as the threshold reference, the structure instability of each node is calculated and obtained, the node whose structure instability is greater than the stability division threshold is identified as a space boundary identification node, and the space boundary node identification result is obtained;

[0074] The formula for calculating the structure instability of each node is specifically as follows:

[0075]

[0076] wherein, ΔS represents the structural instability, used to measure the degree of deviation of node stability, σ represents the standard deviation of the normalized value of the node stability fluctuation interval, reflecting the state fluctuation degree of the node in the time series, θ represents the structural stability threshold, which is a constant set according to the system tolerance, used to divide stable and unstable nodes, R represents the normalized value of the range of the node fluctuation value in the continuous frame, used to represent the maximum amplitude of the spatial position change of the node, n represents the number of frame acquisition, participating in the standard deviation and range normalization processing, γ represents the normalized value of the number of adjacent structure conflicts, used to improve the sensitivity of the structure boundary judgment, reflecting the inconsistency degree of the node surrounding structure;

[0077] According to the node stability fluctuation interval information, first, the standard deviation value of each node in its 10-frame ternary data sequence is calculated, the standard deviation σ is calculated using the normalized Euclidean distance value of the node in 10 frames as the input data, and the range R of each node is calculated respectively, defined as the difference between the maximum and minimum values of the spatial displacement of the node and normalized, and then the stability threshold θ is set to 0.10, which is derived from the XR system in the training task for the average standard deviation fluctuation of each frame node translation of 0.086, taking 1.15 times of the upper limit of the system steady state range for setting, therefore θ = 0.086 × 1.15 = 0.099 ≈ 0.10, which changes with the stability of the system frame background jitter, and the value is stable when the system frame rate remains 30 frames. Then, the structural instability ΔS of each node is calculated, which uses the following formula:

[0078]

[0079] wherein, ΔS represents the structural instability, σ represents the normalized value of the standard deviation, θ is the structural stability threshold, R is the normalized value of the range, n is the number of frames, and γ represents the normalized value of the number of adjacent conflicts around the node, which is derived from the proportion of the change in the number of node adjacencies in the same frame, γ takes the value range of 0, 1, and the change in the number of node adjacencies is more than 2, which is set as a conflict, in the example, the number of node adjacencies of node N3 fluctuates by ±3 in the frame sequence, so Let σ = 0.15, θ = 0.10, R = 0.25, n = 10, and γ = 0.08, and bring them into the formula:

[0080]

[0081] The logic of the formula is as follows: the first term is the relative difference between the standard deviation and the threshold value, which is used to measure whether the fluctuation intensity of the node deviates from the steady state interval, the second term is the amplification term of spatial amplitude change to stability, combined with the range and adjacent disturbance factor, reflecting the structural mutation probability. The results show that the state fluctuation degree of the node is high, and it should be classified as a structural boundary node. The structural instability value AS is 0.5061, when it is greater than the stability threshold value 0.40, it is marked as a spatial boundary identification node, and in this way, the AS values of all nodes in 10 frames are compared, and the node set with AS>0.40 is extracted as the boundary node, and the spatial boundary node identification result is obtained;

[0082] Based on the spatial boundary node identification result, the spatial point cloud set is regionally segmented according to the position index of the structural boundary identification node, the independent structural block boundary line group is demarcated, and the demarcated spatial structure block is mapped and bound with the interactive event response area set in the XR virtual scene, the pairing relationship between each structure block and the response area is obtained, and the virtual-real structure mapping block set is established;

[0083] Based on the spatial boundary node identification result, the spatial structure is re-divided according to the node index of all point cloud sets in 10 frames, the boundary node is taken as the division reference, the boundary line is formed by extending from the boundary node, the point-line-face structure connection mode is used for regional demarcation, the structure mapping in 10 frames is compressed into 4 spatial blocks, and the point set in each block is renumbered to form a structural block boundary line group. According to the interactive event response area set in the XR virtual scene, the interactive trigger event number and the point cloud block number are matched, the events E1 to E4 are bound to the four structure blocks respectively, and the trigger point coordinate mapping is recorded, a pairing relationship dictionary is constructed, such as event E1 mapping structure block B1, event E2 mapping structure block B2, and finally a virtual-real structure mapping block set is established, as shown in Table 1:

[0084] Table 1 Virtual-real structure mapping relationship table

[0085] As shown in Table 1, the interactive event number and the structure block number correspond one by one, and the point coordinate is used for spatial mapping matching, which is the basis data for subsequent

[0086] Event number Mapping structure block number Trigger point coordinate x Trigger point coordinate y Trigger point coordinate z E1 B1 2.13 4.76 1.04 E2 B2 5.21 3.42 0.88 E3 B3 3.67 6.13 1.75 E4 B4 1.92 2.56 0.61

[0087] Path comparison of virtual and real response time, spatial position comparison basis data.

[0088] Please refer to Figure 3 The acquisition steps of the virtual-real path offset component node set are as follows:

[0089] Based on the virtual-real structure mapping block set, the response trigger number, path segment stay duration and interactive action logic sequence number of the component in the interactive task chain in the virtual scene are extracted, the trigger parameter sequence set corresponding to the task chain is constructed, the interactive action number is indexed and sorted according to the task node order, the distribution interval value of the number sequence in the chain is calculated, and the task path response sequence distribution information is obtained;

[0090] Based on the virtual-real structure mapping block set, the response trigger number, path segment stay duration and interactive action logic sequence number of the component in the interactive task chain in the virtual scene are extracted, the trigger parameter sequence set corresponding to the task chain is constructed, the interactive action number is indexed and sorted according to the task node order, the distribution interval value of the number sequence in the chain is calculated, and the task path response sequence distribution information is obtained.

[0091] The task path response sequence distribution information is called, the component execution order number and response segment number are obtained, and the number difference value calculation and sequence offset position comparison are performed with the corresponding virtual path sequence, the path segment number dislocation, the reverse or missing position of the response action sequence are screened and given a logical offset marker, a marker grouping set is established and the corresponding number paragraph index position is recorded, and the path number logical offset record information is obtained.

[0092] The task path response sequence distribution information is called to obtain the component execution order number and the response segment number in the real system. In the comparison process, the path segment number sequence of the component in the XR virtual scene and the component task execution log collected by the real system are used to extract the execution order number and the operation action segment number. The component number is used as the index of the execution sequence comparison operation. First, the two sequences are processed to be consistent in length. If the virtual sequence is longer than the real sequence, the redundant number segment is filled with "no action mark". If the real sequence is longer, the last redundant part is discarded. In the comparison of component number C1, if the virtual path segment number sequence is 1, 2, 3, 5, and the real path segment number is 1, 2, 4, 5, there is a path misplacement at the number 3 and the number 4, and the execution number difference is |3-4|=1. When the difference value is greater than 0, it can be considered as a slight misplacement. When the difference value exceeds 2, it is marked as a serious misplacement. The misplacement judgment reference value is defined as 2. This value is set as a redundant boundary based on the maximum interval 3 of the 95% path consistency identification experiment of the XR system. If the action order mark changes from A-B-C to A-C-B, it is recorded as "reverse order". The missing path segment is identified by the component that exists in the virtual path but has no corresponding number segment in the real record. It is uniformly marked as "missing path segment". All abnormal segment numbers are uniformly summarized to establish a number index set, and the logical tags of the corresponding components and path segments are recorded. The offset mark matrix of each component is constructed, and the structure is: component number, offset type, segment number, comparison difference. Finally, all component offset paragraphs are integrated and stored as object dictionaries to obtain path number logical offset record information.

[0093] According to the path number logical offset record information, the component number corresponding to the inner powder number paragraph is extracted, and the structure node coordinate point mapped in the three-dimensional space is located. The node point set is summarized and the offset corresponding index set is established. The component node coordinate set with offset relationship under the task response chain is integrated to establish the virtual-real path offset component node set.

[0094] According to the path number logical offset record information, the component number pointed to in each offset record is extracted, the spatial node index of the component number in the virtual three-dimensional space structure is matched, the node coordinate index table is queried, the offset paragraph corresponding component is mapped to the three-dimensional space point set, the three-dimensional point coordinate array is constructed, the array structure is component number x (x, y, z), and in the example, the node coordinates corresponding to the number C1 are (1.24, 3.18, 0.86), and the number C3 is (2.01, 2.79, 1.10). After merging all the spatial node coordinates of the components, the coordinate mapping instruction is executed to establish the mapping table between the task response chain and the structure space, and the mapping table field includes: component number, task path number, offset type, coordinate vector, and finally the direct mapping relationship between the structure node and the offset path paragraph is formed. Through component number sorting, the offset tracking and interference source node group extraction across the path can be realized. The integrated result set is the positioning set of the components with path offset behavior in the three-dimensional space in the XR system, and the virtual and real path offset component node set is established.

[0095] Please refer to Figure 4 The structure evaluation interference node grouping table acquisition step is specifically:

[0096] The virtual and real path offset component node set is called, the CPU occupancy rate, interaction time length and frame rate fluctuation rate data of the corresponding component in the XR virtual scene task library are obtained according to the component number index, the three parameter corresponding sequence set is constructed according to the component number, and the product calculation item between the CPU occupancy rate and the frame rate fluctuation rate is established to obtain the load fluctuation value sequence;

[0097] The component number index in the virtual-real path offset component node set is called to obtain the performance resource data of the corresponding component in the XR virtual scene task library according to the number. The data source covers the execution record file of each frame and the system running monitoring module log. The CPU occupancy rate of each component in the current task period is collected, which is expressed by percentage. The sampling period is 0.1 second, and 100 data points are collected in 10 seconds, and the average value is taken as the CPU usage rate of the component. For example, the average occupancy rate of component C1 in 10 seconds is 23.6%, and the average occupancy rate of component C2 is 17.9%. At the same time, the interaction time of the component from the task trigger to the completion is collected, which is expressed by seconds. The start and end timestamps of the component response event are recorded for difference calculation. For example, the trigger time of component C1 is 1.2 seconds, and the completion time is 4.8 seconds, so the interaction time is 4.8-1.2=3.6 seconds. For frame rate fluctuation rate, the system frame rate change data of the component during the trigger period is extracted. The standard frame rate is set to 30FPS. The fluctuation range and fluctuation frequency are recorded, and the fluctuation amplitude change (more than ±3FPS) is taken as the fluctuation point. The absolute value of the frame rate value of the component in the continuous frame is processed. The fluctuation frequency and amplitude mean value are calculated every 5 frames as a window. The calculation method is to divide the absolute value of the frame rate difference between the current frame and the previous frame by the standard frame rate. For example, the frame sequence of component C1 is 28, 31, 26, 30, and 29. The fluctuation rate sequence is [|28-30|, |31-30|, |26-30|, |30-30|, |29-30|] / 30 = [0.066, 0.033, 0.133, 0, 0.033]. Then the product of the CPU occupancy rate and the frame rate fluctuation rate is calculated. The following sequence is constructed for each component: i × fluctuation rate i For example, the mean value of component C1 is 0.236x0.133=0.0314. After generating a complete sequence for each component, the system load fluctuation matrix is obtained, and the load fluctuation value sequence is obtained.

[0098] According to the load fluctuation value sequence, the average change amplitude value of each component corresponding value is calculated according to the component number, and is set as the judgment reference. The change amplitude of each component corresponding load fluctuation value is compared with the judgment reference, and the component number set whose change amplitude exceeds twice the judgment reference is selected, and the abnormal fluctuation node number set is obtained.

[0099] Based on the load fluctuation value sequence, the fluctuation values ​​of all components are first arranged by component number as index. The value range change of each component is statistically analyzed, and the average change amplitude of its fluctuation value sequence is calculated. The change amplitude is defined as the sum of the absolute values ​​of the differences between the load products of adjacent frames divided by the frame number minus 1. For example, if the fluctuation sequence of component C1 in the frame window is 0.031, 0.027, 0.043, then its average change amplitude is (|0.027-0.031|+|0.043-0.027|) / 2=0.010. After listing the average fluctuation amplitudes of all components, the overall average amplitude value is calculated as the judgment benchmark. This judgment benchmark is set as the arithmetic mean of the average change amplitudes of all components. For example, component C1... C1 to C4 are 0.010, 0.012, 0.008, and 0.015 respectively. The baseline value is (0.010+0.012+0.008+0.015) / 4 = 0.01125. The identification threshold is set to twice the judgment baseline value, i.e., 0.01125×2 = 0.0225. When the single fluctuation amplitude of a component exceeds this threshold, it is identified as an abnormal fluctuation component. This multiple is set based on the maximum response point of detection sensitivity in the experiment of the system's false trigger threshold when resisting abnormal fluctuations. Finally, the set of component numbers that meet the conditions is selected. For example, if the change amplitude of component C3 is 0.029, which is greater than 0.0225, then C3 is marked as abnormal, and the set of abnormal fluctuation node numbers is obtained.

[0100] Based on the abnormal fluctuation node number set, the number of times the corresponding component node triggers interaction in the XR scene is extracted and the interaction frequency is calculated. The proportion of frame rate fluctuation times to the total number of frames is calculated, and the ratio between the proportion value and the interaction frequency is established to determine the response interference intensity level of the component in the XR task and establish a structural evaluation interference node grouping table.

[0101] Based on the abnormal fluctuation node number set, the interaction trigger times of the marked components during the task are extracted, which is recorded in the action trigger record of the components during the task process, and the unit is times, and the total frame number of the task is extracted, the interaction frequency corresponding to the component is calculated, the frequency is the trigger times divided by the total task duration, for example, the component C3 triggers the operation 7 times in 10 seconds, the interaction frequency is 0.7 times / s, and then the frame rate fluctuation times ratio value is calculated, assuming that the frame rate fluctuation times in 10 seconds of the task is 21 times, and the total frame number is 300 frames, then the frame rate fluctuation ratio is 21 / 300=0.07, the ratio of the two is calculated to construct the interference coefficient, the interference coefficient is defined as the ratio of the frame rate fluctuation ratio and the interaction frequency, that is, 0.07 / 0.7=0.1, all components construct an interference coefficient matrix, set the interference judgment threshold to 0.09, according to the XR system stability benchmark, the threshold is set to be greater than 18 times of the task fluctuation times and less than 1.0 times / s of the interaction frequency as the serious interference node standard, finally the component nodes with the interference coefficient greater than 0.09 are included in the marking, a judgment table with fields of component number, trigger times, frame fluctuation times, interference coefficient and marking state is established, a response strength grading marking model is constructed, and the result is output as a structure evaluation interference node grouping table.

[0102] Please refer to Figure 5 The acquisition steps of the same dimension index counter-correction coefficient set are specifically as follows:

[0103] According to the structure evaluation interference node grouping table, the task dimension label corresponding to each node is extracted, the dimension categories are divided according to the path navigation class, the interaction accuracy class and the component trigger frequency class, the index set of the node number and the task dimension type is constructed, and the task dimension classification index set is established;

[0104] According to the structure evaluation interference node grouping table, the task dimension label corresponding to each node is extracted, first, the interference node number index and the task dimension field combination item are read in each row, the field identification of the path navigation class, the interaction accuracy class and the component trigger frequency class are separated, the dimension label coding field of each interference node record item is extracted, whether the field value corresponds to the path, the accuracy and the frequency is judged, the task dimension is mapped to the dimension category, and the node number index sequence in the three categories is constructed, the node number of the path navigation class is stored as an array number A, the node number of the interaction accuracy class is stored as an array B, and the node number of the component trigger frequency class is stored as an array C, for example, the node ID numbers 10, 12 and 14 are marked as the path navigation class, and their codes are mapped to T_N1, T_N2 and T_N3 respectively, finally, the index set between the node number and the task dimension type is constructed as {T_N1→path navigation class, T_N2→path navigation class, …}, the field value mapping operation is used to classify the task dimension label field information into the dimension category field set, and the task dimension classification index set is established.

[0105] The task dimension classification index set is called to retrieve the evaluation items involved by the corresponding interference nodes according to the index values under each task dimension, and the index value pairs with opposite deviating directions are extracted, the included angle of the deviating direction vectors and the deviating amplitude values are calculated respectively, the deviation contrast score between the index pairs is obtained by operation, the index pair with the maximum score is selected, and the corresponding adjustment factor is constructed to obtain the countermeasure correction evaluation result.

[0106] The formula for calculating the deviation contrast score between the index pairs is specifically:

[0107]

[0108] Wherein, F is the deviation contrast score between the index pairs, used to determine the deviation contrast degree of the two index pairs in the dimension, V1 is the normalized value of the first evaluation index, representing the difference between its ratio to the reference index value in the current period, V2 is the normalized value of the second evaluation index, representing the difference between its ratio to the reference index value in the current period, D1 is the deviation amplitude normalized value of the first evaluation index, equal to the difference between the current period value and the average value of the past period of the index divided by the average value, D2 is the deviation amplitude normalized value of the second evaluation index, which is calculated in the same way as D1, and a is the normalized value of the association frequency of the node in the current task dimension, equal to the number of task events participated by the node divided by the total number of task events.

[0109] After calling the index values of each task dimension in the task dimension classification index set and extracting the node number group contained in the path navigation class, the record values of the corresponding evaluation items in the index database are retrieved, and the multiple index data corresponding to the node number index are selected, and the data pairs with opposite direction deviation characteristics in the index are selected as the participating items, for example, the direction of index A is 30°, and the direction of index B is -35°, which has obvious reverse direction, the included angle is 65°, and the deviation amplitude values are 0.16 and 0.22 respectively, and the formula is:

[0110]

[0111] The deviation contrast score F of the two index pairs is obtained, the score of all index pairs with reverse characteristics in the dimension is calculated, the maximum value of the score is selected and the corresponding index number is recorded, the maximum value reflects the strongest deviation between the index pairs, that is, the combination with the highest difference, which is used as the basis for constructing the adjustment factor to obtain the countermeasure correction evaluation result. In the formula, Part of it is used to describe the degree of numerical deviation between the indexes, V1 and V2 are the normalized difference values between the two indexes and their respective reference ratios, and the denominator is The influence of the magnitude of the fluctuation of the index itself on the calculation result is regulated, the item ln(1+α) enhances the consideration weight of the index on the frequency of the participating event, for example, assuming that V1=0.18, V2=0.29, D1=0.12, D2=0.25, and α=0.3, the formula is:

[0112]

[0113] The score 0.897 indicates that there is a significant difference between the indexes, which is suitable for the construction of the adjustment factor.

[0114] Based on the evaluation result of the countermeasure correction, the offset contrast score is used as a coefficient to act on the weight value of the original index corresponding to the interference node, the updated weight is superimposed and updated according to the dimension division mode of the participating index, the updated adjustment coefficient is integrated and associated with the original index system number, and the countermeasure correction coefficient set of the indexes in the same dimension is established;

[0115] Based on the evaluation result of the countermeasure correction, the offset contrast score F is used as an adjustment factor, which is applied to the adjustment of the original index weight value, and the weight update operation is performed on the path navigation type, interaction accuracy type and component trigger frequency type indexes according to the dimension classification, for example, the original weight of index A is 0.18, and the corresponding score is 0.897, then the adjusted weight is 0.18*0.897=0.1615, which is used to update the normalized total weight value of the index in the dimension, and the original index number field is retained, the original number, old weight, score value and new weight are recorded in the update content, and the same operation is performed on all indexes in each dimension class, then the updated weight values in the three types of dimensions and the original index number are mapped to form a mapping table, and the countermeasure correction coefficient set of the indexes in the same dimension is established.

[0116] Table 2 shows an example of the index weight after countermeasure correction:

[0117] Index number Belonging dimension Original weight Offset score F Corrected weight A101 Path navigation class 0.18 0.897 0.1615 A203 Interaction precision class 0.21 0.923 0.1938 A309 Component trigger frequency class 0.15 1.014 0.1521

[0118] As shown in Table 2, the modified index weight can be obtained by multiplying the offset contrast score and the original weight value, so as to update the weight structure according to the dimension. The score value is obtained from the calculation result of the formula in paragraph 2.

[0119] Please refer to Figure 6 The acquisition steps of the virtual-real evaluation matching level label set are as follows:

[0120] The countermeasure correction coefficient set of the indexes in the same dimension is called, the original weight values corresponding to the three types of evaluation indexes of component operation logic, path consistency expression and interaction timing period are extracted, the product calculation is performed on the original weight values and the correction coefficients according to the dimension to which the index belongs, the proportion update of the index in the original dimension is performed, and the evaluation index updated weight information is generated;

[0121] The same dimension index pair is called to modify the set of correction coefficients, the original weight values corresponding to the three types of evaluation indexes of component operation logic, path consistency expression, and interactive timing cycle are extracted, the three types of index numbers and their initial weights are extracted according to the dimension classification structure, for example, the weight of the operation logic type index component starting action delay time is 0.20, the weight of the path consistency type index path minimum angle difference is 0.18, and the weight of the interactive timing cycle type index average response interval time is 0.22, and the corresponding numbered correction coefficients are further called, such as the delay time index correction coefficient is 0.913, the angle difference correction coefficient is 0.862, and the interval time correction coefficient is 0.937. The original weight of each index and its correction coefficient are operated in a multiplication relationship to calculate the modified weight of each item, for example, 0.20*0.913=0.1826, 0.18*0.862=0.1552, 0.22*0.937=0.2061. The modified weight values of the three types of indexes are recorded respectively, and a structured data row list is established according to the index number and the dimension type index, as shown in Table 3. The proportional updating operation of all evaluation indexes under the original dimension is completed, and the evaluation index updating weight information is generated.

[0122] Table 3 Evaluation index updating weight table

[0123] Index number Dimension type Original weight Correction coefficient Corrected weight M101 Component operation logic 0.20 0.913 0.1826 M205 Path consistency expression 0.18 0.862 0.1552 M307 Interaction timing period 0.22 0.937 0.2061

[0124] As shown in Table 3, the modified weight is calculated according to the multiplication relationship between the original value and the correction coefficient. The numerical value is traceable and falls within the weight interval [0, 1].

[0125] According to the evaluation index updating weight information, the component operation value, the path matching score value and the interactive response cycle value in the virtual scene are obtained, the corresponding real operation reference value sequence is obtained, the numerical difference of the same type index is calculated and divided by the reference value mean value, the evaluation item deviation amplitude is obtained, the index deviation amplitude is summarized and matched with the corresponding operation link, and the virtual-real index deviation value group is obtained.

[0126] According to the evaluation index, the weight information is updated, the operation value, the path matching score value and the interaction response cycle value of each component in the XR virtual scene are obtained, the index values of the components M101, M205 and M307 in the current XR task are extracted by calling the task number and component number mapping relationship, which are respectively the operation value 3.8 seconds, the path matching score 0.87 and the response cycle value 2.6 seconds, and the corresponding real operation reference value sequence is searched, for example, the reference value of M101 is 3.2 seconds, the reference value of M205 is 0.95, and the reference value of M307 is 2.3 seconds, the numerical difference between each pair of indexes is calculated, and the deviation amplitude is calculated by dividing the difference by the average of the corresponding reference value, which is respectively (3.8-3.2) / ((3.8+3.2) / 2)=0.6 / 3.5=0.1714, (0.87-0.95) / 0.91=-0.0879, (2.6-2.3) / 2.45=0.1224, the above deviation amplitude values are classified into each evaluation index, and are indexed and classified according to the operation link to which the index and the component belong in the task, three groups of deviation amplitude indexes are constructed: component operation value deviation amplitude 0.1714, path matching score deviation amplitude-0.0879, and interaction cycle value deviation amplitude 0.1224, and finally one group of structure data is merged according to the link dimension to obtain the virtual-real index deviation value group.

[0127] Based on the virtual-real index deviation value group, the deviation amount of each evaluation link is compared with the matching threshold range, the operation scene type information is called and the index matching state is extracted according to the scene grouping, and each evaluation state level label is output according to the grade division rule to establish the virtual-real evaluation matching level label set.

[0128] Based on the virtual-real index deviation value group, the deviation amount of each evaluation link is compared with the matching threshold range, the operation scene type information is called and the index matching state is extracted according to the scene grouping, and each evaluation state level label is output according to the grade division rule to establish the virtual-real evaluation matching level label set.

[0129] Please refer to Figure 7 An XR virtual scene evaluation system based on virtual-real fusion data analysis is used to execute the above-mentioned XR virtual scene evaluation method based on virtual-real fusion data analysis, and the system comprises:

[0130] The structure mapping processing module obtains multiple frames of three-dimensional structure point cloud data of the interactive construction in the XR virtual scene, performs comparison based on the inter-node structure stability value, identifies the boundary marker node, associates and maps the structure block with the interactive event response area in the XR virtual scene, and obtains a virtual-real structure mapping block set;

[0131] The offset component identification module performs number sequence difference value calculation and logical offset position comparison based on the virtual-real structure mapping block set, logically marks and aggregates the position of the reverse or missing order of the path segment number occurrence sequence and the response action sequence to form an offset marker record table, and obtains a virtual-real path offset component node set;

[0132] The interference node analysis module calls the virtual-real path offset component node set, calculates the product of the frame rate fluctuation rate and the CPU occupation rate, establishes the ratio between the interaction frequency of the interference component appearing in the XR scene and the abnormal frame response fluctuation, and generates a structure evaluation interference node grouping table;

[0133] The correction coefficient evaluation module retrieves the evaluation items associated with the interference node in the same dimension based on the structure evaluation interference node grouping table, compares the deviation direction and the amplitude value, extracts a pair of index values with the largest difference to construct an antagonistic correction factor, and obtains a same-dimension index antagonistic correction coefficient set;

[0134] The evaluation matching analysis module calls the same-dimension index antagonistic correction coefficient set, updates the weight of the component operation logic, path consistency expression, and interaction timing cycle index according to the corresponding correction coefficient, performs comparison between the virtual scene evaluation result in the XR system and the real operation reference value, calculates the evaluation deviation amount of each link in the virtual-real scene, and outputs a discriminable matching level label according to the operation scene, and establishes a virtual-real evaluation matching level label set.

[0135] It should be understood that the term "and / or" herein is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects, but can also represent an "and / or" relationship, which can be understood according to the context before and after.

[0136] In the present application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of the items, including a single item or any combination of multiple items. For example, at least one of a, b, or c can mean a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0137] It should be understood that the size of the sequence number of the above-mentioned processes does not mean the order of execution in various embodiments of the present application. The execution order of the processes should be determined according to their functions and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0138] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0139] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-mentioned devices, apparatuses and units can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0140] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another device, or some features can be omitted or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed units can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0141] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. they can be located in one place or distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the present embodiment.

[0142] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit.

[0143] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the prior art that essentially contributes or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0144] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for evaluating XR virtual scenes based on virtual-real fusion data analysis, characterized in that, Includes the following steps: S1: Acquire multi-frame 3D structural point cloud data of interactive construction in the XR virtual scene, perform comparison based on the stability value between nodes, identify boundary marker nodes, and associate and map the structural blocks with the interactive event response areas in the XR virtual scene to obtain a set of virtual and real structure mapping blocks. S2: Based on the set of virtual and real structure mapping blocks, perform number sequence difference calculation and logical offset position comparison, logically mark the positions where the path segment numbers are out of order, the response action order is reversed or missing, and summarize them to form an offset mark record table to obtain the set of virtual and real path offset component nodes. S3: Call the set of virtual and real path offset component nodes, calculate the product of frame rate fluctuation rate and CPU utilization rate, establish the ratio between the interaction frequency of the interfering component in the XR scene and the abnormal frame response fluctuation, and generate a structural evaluation interference node grouping table. S4: Based on the structure evaluation interference node grouping table, retrieve the evaluation items associated with the interference nodes in the same dimension, compare the deviation direction and magnitude values, extract the pair of index values ​​with the largest difference to construct the adversarial correction factor, and obtain the set of adversarial correction coefficients for the same dimension indexes.

2. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 1, characterized in that, The virtual-real structure mapping block set includes spatial boundary identification markers, interactive area association tags, three-dimensional structure segmentation, and virtual response positioning indexes. The virtual-real path offset component node set includes path segment misalignment numbers, response order anomaly tags, action logic offset node numbers, and spatial coordinate mapping information. The structure evaluation interference node grouping table includes frame rate response mutation nodes, processor resource usage markers, interaction frequency fluctuation node numbers, and interference behavior frequency factors. The same-dimensional indicator adversarial correction coefficient set includes path-type indicator weighting factors, interaction-type indicator suppression factors, action execution offset weights, and indicator direction contrast coefficients.

3. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 2, characterized in that, The specific steps for obtaining the virtual-real structure mapping block set are as follows: Acquire multi-frame 3D structural point cloud data of interactive construction in XR virtual scene, extract the spatial displacement value, connection vector angle value and relative position offset between each structural spatial node and its neighboring nodes in consecutive frames, construct the three-dimensional data feature set of each node, calculate the value range change range of structural spatial nodes between differentiated frames, and obtain the node stable fluctuation range information. Based on the node stability fluctuation range information, the standard deviation and range of each structural spatial node in the time series are calculated. The standard deviation and range are used to construct a stability comparison term. The structural stability boundary value is used as a threshold reference to calculate the structural instability of each node. Nodes with structural instability greater than the stability division threshold are identified as spatial boundary marker nodes, and the spatial boundary node identification results are obtained. Based on the spatial boundary node identification results, the spatial point cloud set is divided into regions according to the position index of the structural boundary identifier nodes, independent structural block boundary line groups are delineated, and the delineated spatial structural blocks are mapped and bound to the interactive event response areas set in the XR virtual scene. The pairing relationship between each structural block and the response area is obtained, and a set of virtual and real structure mapping blocks is established.

4. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 3, characterized in that, The specific formula for obtaining the structural instability of each node is as follows: Where ΔS represents the structural instability, σ represents the normalized standard deviation of the node's stable fluctuation range, θ represents the structural stability threshold, R represents the normalized range of the node's fluctuation values ​​within consecutive frames, n represents the number of frames acquired, and γ represents the normalized number of adjacent structural conflicts.

5. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 4, characterized in that, The specific steps for obtaining the set of virtual and real path offset component nodes are as follows: Based on the virtual-real structure mapping block set, the response trigger number, path segment dwell time and interaction action logic sequence number of the components in the virtual scene in the interaction task chain are extracted, the trigger parameter sequence set corresponding to the task chain is constructed, the interaction action number is indexed and sorted according to the task node order, the distribution interval value of the number sequence in the chain is calculated, and the task path response sequence distribution information is obtained. Call the task path response sequence distribution information to obtain the component execution sequence number and response segment number, and perform number difference calculation and sequence offset position comparison with the corresponding virtual path sequence. Filter out positions with misaligned path segment numbers, reversed response action order, or missing positions and assign logical offset marks. Establish a set of marked groups and record the corresponding numbered segment index position to obtain path number logical offset record information. Based on the path number logical offset record information, extract the component number corresponding to the inner powder number segment and locate the coordinate points of the mapped structural node in three-dimensional space. Summarize the node point set and establish the offset corresponding index set. Integrate them into the component node coordinate set with offset relationship under the task response chain and establish the virtual and real path offset component node set.

6. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 5, characterized in that, The specific steps for obtaining the structural evaluation interference node grouping table are as follows: The virtual and real path offset component node set is called, and the CPU utilization, interaction time and frame rate fluctuation data of the corresponding component in the XR virtual scene task library are obtained according to the component number index. The three parameters corresponding sequence set are constructed according to the component number, and the product calculation term between CPU utilization and frame rate fluctuation is established to obtain the load fluctuation value sequence. Based on the load fluctuation value sequence, the average change amplitude value of the corresponding value of each component is calculated and set as the judgment benchmark according to the component number. The load fluctuation value change amplitude of each component is compared with the judgment benchmark, and the set of component numbers with change amplitude exceeding twice the judgment benchmark is filtered to obtain the abnormal fluctuation node number set. Based on the set of abnormal fluctuation node numbers, the number of times the corresponding component node triggers interaction in the XR scene is extracted and the interaction frequency is calculated. The proportion of frame rate fluctuations to the total number of frames is calculated, and the ratio between the proportion value and the interaction frequency is established. The response interference intensity level of the component in the XR task is judged, and a structural evaluation interference node grouping table is established.

7. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 6, characterized in that, The specific steps for obtaining the same-dimensional indicator anti-correction coefficient set are as follows: Based on the structure evaluation interference node grouping table, extract the task dimension label corresponding to each node, divide the dimension categories according to path navigation, interaction accuracy, and component trigger frequency, construct an index set of node number and task dimension type, and establish a task dimension classification index set. The task dimension classification index set is called, and the evaluation items involved in the corresponding interference node are retrieved according to the index value under each task dimension. The index value pairs with opposite deviation directions are extracted, and the angle between their deviation direction vectors and the deviation magnitude value are calculated respectively. The offset contrast score between the index pairs is obtained, the index pair with the largest score is selected, and the corresponding adjustment factor is constructed to obtain the adversarial correction evaluation result. The formula for obtaining the offset contrast score between the indicator pairs is as follows: Where F is the offset contrast score between indicator pairs, V1 is the normalized value of the first evaluation indicator, V2 is the normalized value of the second evaluation indicator, D1 is the normalized value of the offset magnitude of the first evaluation indicator, D2 is the normalized value of the offset magnitude of the second evaluation indicator, and α is the normalized value of the association frequency of the node in the current task dimension. Based on the aforementioned adversarial correction evaluation results, the offset contrast score is used as the weight value of the original index under the corresponding interference node. The participating indices are updated by superimposing their weights according to the dimensional division method. The updated adjustment coefficients are integrated and associated with the original index system number to establish a set of adversarial correction coefficients for indices of the same dimension.

8. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 7, characterized in that, The method further includes the following steps: S5: Call the same-dimensional indicator anti-correction coefficient set, update the component operation logic, path consistency expression, and interaction timing cycle indicators according to the corresponding correction coefficients, perform a comparison between the virtual scene evaluation results and the real operation reference values ​​in the XR system, calculate the evaluation deviation of each link in the virtual and real scenes, and aggregate and output discernible matching level labels according to the operation scene to establish a virtual and real evaluation matching level label set. The virtual-real assessment matching level label set specifically includes operation consistency level label, path reconstruction matching level, behavior synchronization assessment label, and structure alignment effectiveness label.

9. The XR virtual scene evaluation method based on virtual-real fusion data analysis according to claim 8, characterized in that, The specific steps for obtaining the virtual-real assessment matching level label set are as follows: The same-dimensional indicator set of the anti-correction coefficient is called to extract the original weight values ​​corresponding to the three types of evaluation indicators: component operation logic, path consistency expression, and interaction time sequence cycle. The original weight values ​​and correction coefficients are multiplied according to the indicator belonging dimension to perform the product calculation, and the indicators are updated proportionally under the original dimension to generate the evaluation indicator update weight information. Update weight information according to the evaluation indicators, obtain component operation value, path matching score value and interaction response cycle value in virtual scene, obtain corresponding real operation reference value sequence, calculate the numerical difference of each pair of similar indicators and divide by the average of reference values ​​to obtain the deviation of evaluation item, summarize the deviation of indicators and match the corresponding operation links to obtain virtual and real indicator deviation value group. Based on the virtual and real indicator deviation value group, the deviation of each evaluation step is compared with the matching threshold range, the operation scenario type information is called and the indicator matching status is extracted by scenario group, and the evaluation status level label of each group is output according to the level classification rule to establish a virtual and real evaluation matching level label set.

10. An XR virtual scene evaluation system based on virtual-real fusion data analysis, characterized in that, The system is used to implement the XR virtual scene evaluation method based on virtual-real fusion data analysis as described in any one of claims 1-9, and the system comprises: The structure mapping processing module acquires multi-frame 3D structure point cloud data interactively constructed in the XR virtual scene, performs comparison based on the stability value constructed between nodes, identifies boundary marker nodes, and associates and maps the structure blocks with the interactive event response areas in the XR virtual scene to obtain a set of virtual and real structure mapping blocks. The offset component identification module performs number sequence difference calculation and logical offset position comparison based on the virtual and real structure mapping block set. It logically marks the positions where the path segment numbers are out of order, the response action order is reversed, or the position is missing, and summarizes them to form an offset mark record table, thus obtaining the virtual and real path offset component node set. The interference node analysis module calls the set of virtual and real path offset component nodes, calculates the product of frame rate fluctuation rate and CPU utilization rate, establishes the ratio between the interaction frequency of interference components appearing in the XR scene and the abnormal frame response fluctuation, and generates a structural evaluation interference node grouping table. The correction coefficient evaluation module evaluates the interference node grouping table based on the structure, retrieves the evaluation items associated with the interference node in the same dimension, compares the deviation direction and magnitude values, extracts the pair of index values ​​with the largest difference to construct the adversarial correction factor, and obtains the set of adversarial correction coefficients for the same dimension index. The evaluation and matching analysis module calls the same-dimensional indicator anti-correction coefficient set, updates the component operation logic, path consistency expression, and interaction timing cycle indicators with corresponding correction coefficients, performs a comparison between the virtual scene evaluation results and the real operation reference values ​​in the XR system, calculates the evaluation deviation of each link in the virtual and real scenes, and aggregates and outputs discernible matching level labels according to the operation scene to establish a virtual and real evaluation matching level label set.

Images