CN-122023675-A - Environment visual simulation method based on wave collapse mechanism

CN122023675ACN 122023675 ACN122023675 ACN 122023675ACN-122023675-A

Abstract

The invention relates to the technical field of environment visualization and discloses an environment visualization simulation method based on a wave collapse mechanism, which comprises the steps of firstly, obtaining a three-dimensional visual scene unit in a current visual field range, and initializing the three-dimensional visual scene unit to generate modal probability distribution; the visual information density characteristics, visual saliency factors, and uncertainty indicators are then calculated and line-of-sight associated sensitivity coefficients are found based on logarithmic space regression to determine a visual convergence hysteresis threshold and a frame-level state establishment quota. And dynamically sequencing the importance of the scene units through a shaping sequencing function, performing progressive state convergence and semantic association propagation on the key units, and finally outputting an environment visualization result conforming to the target visual quality. The invention can obviously reduce the calculation burden of rendering and generating, improves the real-time generation efficiency and visual stability of complex scenes, and is suitable for the fields of virtual simulation, game rendering, digital twin, intelligent visualization systems and the like.

Inventors

ZHU ZHENYU
SUN JING
LIU WEIWEI

Assignees

南京宇天智云仿真技术有限公司

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (9)

1. An environment visualization simulation method based on a wave collapse mechanism is characterized by comprising the following steps: Acquiring three-dimensional visual field units in a current visual field range, and initializing the generation modal probability distribution of each three-dimensional visual field unit; Calculating visual information density characteristics and visual saliency factors of the three-dimensional visual scene unit, and calculating a sight-line association sensitivity coefficient and a visual convergence hysteresis threshold value based on a mapping relation between the visual saliency factors and a generated uncertainty index derived from the generated modal probability distribution; Setting a frame-level state establishment quota of a current simulation frame according to the vision convergence hysteresis threshold, constructing a shaping and sorting function for preferentially processing high vision significance factors by utilizing the vision correlation sensitivity coefficient, and executing progressive state convergence calculation and semantic correlation propagation on the three-dimensional visible scene units in the frame-level state establishment quota according to the determined sequence of the shaping and sorting function to output a visualized environment scene.
2. The method for environmental visualization simulation based on wave collapse mechanism according to claim 1, wherein obtaining a three-dimensional visual scene unit in a current view range comprises: The method comprises the steps of traversing hierarchical nodes based on a three-dimensional tile data structure, calculating screen space errors of the hierarchical nodes according to virtual viewpoint positions, recursively refining the hierarchical nodes with the screen space errors exceeding a preset maximum allowable error threshold, taking the hierarchical nodes with the screen space errors not exceeding the preset maximum allowable error threshold as nodes to be rendered at the current moment, establishing index mapping between the nodes to be rendered and a bottom semantic object, and taking a semantic object set obtained by mapping as the three-dimensional visual scene unit.
3. The method of claim 1, wherein initializing the generated modal probability distribution of each of the three-dimensional visual scene units comprises: The method comprises the steps of obtaining sample statistical frequency or priori frequency weight of each semantic mode to be selected in a wave collapse mode library, carrying out addition normalization processing on the priori frequency weight of all the semantic modes to be selected, calculating to obtain initial probability values of each semantic mode to be selected, and taking a vector set formed by the initial probability values as generated modal probability distribution of each three-dimensional visual scene unit at initial time.
4. The method for simulating environmental visualization based on wave collapse mechanism according to claim 1, wherein calculating the visual information density characteristics and visual saliency factors of the three-dimensional visual scene unit comprises: Collecting rendering characteristic data in a screen projection area of the three-dimensional visual field scene unit, constructing a characteristic distribution histogram, calculating shannon entropy of the characteristic distribution histogram, taking the calculated shannon entropy value as the visual information density characteristic, calculating Euclidean distance between a projection center coordinate of the three-dimensional visual field scene unit in a screen space and a screen geometric center coordinate, calculating a view center cone weight which is in center attenuation distribution based on the Euclidean distance, weighting and combining the visual information density characteristic and the view center cone weight by using an exponential function, normalizing the weighted result of all units in the current view range, and taking the normalized value as the visual significance factor.
5. The method for simulating environment visualization based on wave collapse mechanism according to claim 1, wherein the generated uncertainty indicator derived from the generated modal probability distribution comprises: Traversing probability components of each semantic mode to be selected in the generated modal probability distribution, calculating the product of each probability component and the corresponding natural logarithm value, accumulating and summing all product results, and taking the negative value of the accumulated and summed result as the generated uncertainty index of the three-dimensional visual scene unit.
6. The method of claim 4, wherein the calculating the line-of-sight associated sensitivity coefficient and the vision convergence hysteresis threshold based on the mapping relationship between the vision saliency factor and the uncertainty index comprises: and respectively calculating natural logarithmic values of the generated uncertainty index and the visual saliency factor of each three-dimensional visual scene unit, carrying out regression fitting on a logarithmic linear relation between the generated uncertainty index and the natural logarithmic value of the visual saliency factor by using a least square method, taking a regression slope obtained by fitting as the sight-line associated sensitivity coefficient, taking a numerical value II as a base number, taking the inverse number of the sum of the sight-line associated sensitivity coefficient and the numerical value I as an index, carrying out power operation, and taking a power operation result as the visual convergence hysteresis threshold.
7. The method of claim 1, wherein setting a frame-level state establishment quota for a current simulation frame based on the vision convergence hysteresis threshold comprises: The method comprises the steps of calculating the total number of three-dimensional visible scene units in a current visual field range, calculating the product of a vision convergence hysteresis threshold and the total number, carrying out upward rounding operation on the product, and taking the rounded integer value as the frame-level state of the current simulation frame completion state convergence to establish a quota.
8. The method of claim 1, wherein constructing a shaping ranking function that prioritizes high visual saliency factors using the line-of-sight associated sensitivity coefficients comprises: the method comprises the steps of calculating the reciprocal of the sum of the sight-associated sensitivity coefficient and a numerical value one, taking the reciprocal as a power exponent, adding the visual saliency factor of the three-dimensional visual scene unit and a preset non-zero correction term as a base, performing power operation, calculating the ratio of the uncertainty index generated by the three-dimensional visual scene unit to the power operation result, taking the ratio as a shaping ordering function value, and arranging all three-dimensional visual scene units in the current visual field range according to the order of the shaping ordering function value from small to large to generate a processing sequence for preferentially processing the high visual saliency factor.
9. The method of claim 1, wherein progressive state convergence computation and semantic association propagation are performed on the three-dimensional visual scene units within the frame-level state establishment quota in the order determined by the shaping ordering function, and a visualized environmental scene is output, comprising: Establishing each three-dimensional visible scene unit in the quota of the frame-level state, carrying out power sharpening and renormalization processing on the corresponding generated modal probability distribution by utilizing temperature parameters attenuated along with the processing progress to realize gradual state convergence, inquiring a predefined semantic compatibility matrix based on the urban semantic space topological relation, calculating a compatibility propagation message containing a bottom protection factor, and updating the generated modal probability distribution of the adjacent units by utilizing the compatibility propagation message, wherein the bottom protection factor is used for ensuring that the probability value is constantly greater than zero, selecting the semantic mode with the largest probability in each three-dimensional visible scene unit for geometric instantiation rendering, and generating a visualized environment scene.

Description

Environment visual simulation method based on wave collapse mechanism Technical Field The invention relates to the technical field of environment visualization, in particular to an environment visualization simulation method based on a wave collapse mechanism. Background With the rapid development of technologies such as virtual simulation, digital twin, intelligent visualization and the like, the demands for real-time generation and dynamic presentation of a large-scale three-dimensional environment are continuously increased. Particularly in application scenes such as city simulation, interactive virtual space, intelligent driving simulation and the like, the system needs to continuously generate a three-dimensional visual environment with complex structure, rich semantics and stable vision under the condition of limited calculation force. Therefore, more and more rendering and generating frames introduce mechanisms such as probability models, sparse representation, semantic inference and the like so as to improve the visual quality of the dynamic environment. However, in the face of high resolution views, large scene scale and real-time interaction requirements, conventional visualization methods still have difficulty balancing efficiency with visual quality. The existing three-dimensional environment generation method is used for performing rendering resource allocation by depending on fixed rules or static priorities, and real-time importance changes of all three-dimensional scene units in a view area are difficult to reflect. Meanwhile, when the traditional technology based on LOD (LevelofDetail) or local heuristic is used for processing a scene with high complexity, the problems of slow visual convergence, inconsistent local semantics, unbalanced resource allocation and the like are easy to occur. When the scene has the conditions of large texture density difference, uneven semantic distribution or frequent movement of sight, the existing method cannot effectively quantify the uncertainty, and the rendering sequence is difficult to dynamically regulate and control according to the visual saliency, so that the overall generation efficiency is reduced and the visual experience is unstable. Disclosure of Invention The invention provides an environment visualization simulation method based on a wave collapse mechanism, which solves the technical problems in the background technology. The invention provides an environment visualization simulation method based on a wave collapse mechanism, which comprises the following steps: Acquiring three-dimensional visual field units in a current visual field range, and initializing the generation modal probability distribution of each three-dimensional visual field unit; Calculating visual information density characteristics and visual saliency factors of the three-dimensional visual scene unit, and calculating a sight-line association sensitivity coefficient and a visual convergence hysteresis threshold value based on a mapping relation between the visual saliency factors and a generated uncertainty index derived from the generated modal probability distribution; Setting a frame-level state establishment quota of a current simulation frame according to the vision convergence hysteresis threshold, constructing a shaping and sorting function for preferentially processing high vision significance factors by utilizing the vision correlation sensitivity coefficient, and executing progressive state convergence calculation and semantic correlation propagation on the three-dimensional visible scene units in the frame-level state establishment quota according to the determined sequence of the shaping and sorting function to output a visualized environment scene. The visual field unit dynamic, controllable and progressive convergence and semantic consistency propagation method has the beneficial effects that by introducing the multi-mode indexes such as visual information density, visual saliency and uncertainty generation and combining the visual line association sensitivity coefficient and the shaping and sequencing mechanism, the three-dimensional visual field unit dynamic, controllable and progressive convergence and semantic consistency propagation are realized. The invention can reduce unnecessary calculation expenditure on the premise of ensuring visual quality, so that rendering resources are intensively distributed to the key region of the vision, thereby greatly improving the real-time generation efficiency and stability of complex scenes, and being particularly suitable for high-performance environments such as interactive visualization, virtual simulation, digital twinning and the like. Drawings FIG. 1 is a flow chart of an environmental visualization simulation method based on wave collapse mechanism of the present invention; Fig. 2 is a schematic diagram of a three-dimensional visual field unit within the current field of view of the present invention. Detailed