CN-121147364-B - Digital human grid creation method for ensuring sequence frame continuity

Abstract

The invention relates to the technical field of digital human modeling, and discloses a digital human grid creation method for ensuring sequence frame continuity. The method comprises the steps of sequence frame grid monitoring, real-time acquisition of digital human sequence frame grid data, real-time combing and screening of monitored grid continuity anomalies, real-time updating of screening results to a sequence frame continuity state diagram, continuous integration evaluation, calculation and analysis of anomaly influence of corresponding frame sequences according to real-time updating contents of the state diagram, further implementation of first dynamic management on grid creation of subsequent frame sequences, anomaly type expansion evaluation, analysis of influence of various anomalies on grid creation by combining with updating data of different anomaly types in the state diagram, and implementation of second dynamic management on the whole digital human grid creation process. The method can accurately process the grid continuity abnormality, ensure the sequence frame grid continuity and optimize the dynamic expression effect of the digital person.

Inventors

• WANG WEIDONG
• LI LIAN

Assignees

• 浙江博采传媒有限公司

Dates

Publication Date

20260505

Application Date

20250912

Claims (7)

1. 1. A digital human mesh creation method for ensuring sequence frame continuity, comprising: A sequence frame grid monitoring step, which is used for monitoring grid data of the digital human sequence frame in real time, carrying out targeted abnormal carding and screening on the monitored grid continuity abnormality, and updating the screening result to a sequence frame continuity state diagram in real time; a continuity integration evaluation step, which is used for carrying out integration calculation and data analysis of the abnormal influence of the corresponding frame sequence according to the real-time update of the sequence frame continuity state diagram, and carrying out first dynamic management on the grid creation of the subsequent frame sequence according to the analysis result; An anomaly type expansion evaluation step, which is used for carrying out integration calculation and data analysis of the influences of different anomaly types according to the real-time update of the different anomaly types in the sequence frame continuity state diagram, and carrying out second dynamic management on the creation process of the digital human grid according to the analysis result; The continuity integration evaluating step further includes: Marking the corresponding target frame sequence interval as a low-influence frame sequence interval or a high-influence frame sequence interval according to the influence degree mark with the value of the first preset value or the second preset value; When grid creation of a subsequent frame sequence corresponding to a target frame sequence interval is subjected to first dynamic management, a targeted strengthening creation scheme is implemented for subsequent sequence frames of other normal grids in a low-influence frame sequence interval, and a targeted optimizing creation scheme is implemented for sequence frames of other normal grids in a high-influence frame sequence interval; the anomaly type expansion evaluation step comprises the following steps: Obtaining all abnormal identifications of which the numerical values are not the preset initial values, corresponding abnormal identification numerical values and total occurrence numbers, and calculating and obtaining type influence values corresponding to different abnormal types through type abnormal influence rules; The type abnormal influence rule defines a type influence value as a preset reference value or an adjustment value, and calculates based on the total number of sequence frames of the normal grid corresponding to all frame sequence intervals and the total number of abnormal types with all the values not being preset initial values; when the creation process of the digital personal grid is subjected to second dynamic management, carrying out data analysis on type influence values corresponding to different abnormal types; the anomaly type expansion evaluation step further comprises the following steps: If the type influence value is an adjustment value, the existing creation scheme corresponding to the type of the abnormality is adjusted to implement targeted creation optimization management.
2. 2. The digital human mesh creation method of ensuring sequential frame continuity of claim 1, wherein the sequential frame mesh monitoring step comprises: acquiring an abnormal type corresponding to the sequence frame in the abnormal state, and performing digital processing on the acquired abnormal type to acquire a corresponding abnormal identifier; Analyzing the abnormal identifier when targeted abnormal carding and screening are carried out on the monitored grid continuity abnormality according to the abnormal identifier; If the sample abnormal type which is the same as the abnormal type does not exist in the abnormal type reference table, setting the obtained abnormal identifier corresponding to the abnormal type as a preset initial value; If the sample abnormal type which is the same as the abnormal type exists in the abnormal type reference table, the abnormal identification corresponding to the sample abnormal type which is the same as the abnormal type is associated with the obtained abnormal type.
3. 3. The digital human mesh creation method of ensuring sequential frame continuity of claim 2, wherein the sequential frame mesh monitoring step further comprises: Acquiring a frame sequence interval corresponding to a sequence frame in an abnormal state according to an abnormal identifier with a value which is not a preset initial value and adding one to the total number of the abnormal frames; The method comprises the steps of obtaining a frame sequence interval, grid coordinates, an anomaly identification and an anomaly occurrence time point corresponding to a sequence frame in an anomaly state, and carrying out sequencing combination to obtain a sequence frame continuity monitoring sequence corresponding to the sequence frame in the anomaly state; And supplementing the frame sequence interval in the sequence frame continuity monitoring sequence to the corresponding same sample frame sequence interval in the sequence frame continuity state diagram to update in real time.
4. 4. A digital personal mesh creation method for ensuring sequential frame continuity as recited in claim 3, wherein the continuity integration evaluating step comprises: Generating an integration instruction when data updating exists in a frame sequence interval of the sequence frame continuity state diagram, marking the updated frame sequence interval as a target frame sequence interval according to the integration instruction, and counting the total number of sequence frames of a normal grid in the target frame sequence interval and the total number of sequence frames with abnormal continuity; Carrying out data integration calculation through a frame sequence identification segmentation rule and outputting an influence state identifier corresponding to a target frame sequence interval, wherein the influence state identifier comprises a preset normal value or a numerical value of a non-preset normal value and respectively represents that the overall state of the frame sequence corresponding to the target frame sequence interval is normal or the overall state of the frame sequence is abnormal; And analyzing the calculated and output influence state identification to determine an abnormal influence state corresponding to the target frame sequence interval.
5. 5. The digital personal grid creation method for ensuring sequential frame continuity as recited in claim 4, wherein the continuity integration evaluating step further comprises: if the value affecting the state identification is a preset normal value, the overall state of the frame sequence related to the target frame sequence interval is normal; If the value of the influence state identifier is not a preset normal value, an anomaly tracing instruction is generated, data integration calculation is carried out through an influence degree identification segmentation rule, and an influence degree identifier corresponding to the target frame sequence interval is output, wherein the influence degree identifier comprises a first preset value or a second preset value, and the anomaly influence degree corresponding to the target frame sequence interval is respectively indicated to be slightly abnormal in the overall state of the frame sequence or severely abnormal in the overall state of the frame sequence.
6. 6. The digital personal mesh creation method for ensuring sequential frame continuity of claim 1, further comprising: a sequence frame data storage step for storing the monitored grid data in a time sequence database and ordering according to the sequence of the frames; a sequential frame query step for retrieving corresponding grid data from a sequential database according to a query request of a sequential frame continuity state diagram; And a sequential frame analysis step for performing cleaning and missing value processing on the stored grid data, and performing statistical analysis and continuity detection analysis on the retrieved grid data.
7. 7. The digital human mesh creation method of ensuring sequential frame continuity of claim 6, wherein the sequential frame analysis step comprises: judging whether to collect additional grid data based on the output of the sequence frame continuity state diagram, and executing a grid data collection step when judging to collect the additional grid data; Determining optimal continuity parameters of the digital personal grid based on the acquired additional grid data; the mesh creation process of the subsequent frame sequence is adjusted based on the optimal continuity parameter.

Description

Digital human grid creation method for ensuring sequence frame continuity Technical Field The invention relates to the technical field of digital human modeling, in particular to a digital human grid creation method for ensuring sequence frame continuity. Background With rapid penetration of digital technologies in the fields of film and television production, virtual interaction, game development, etc., digital man-made technology has become a hotspot for current technical development and industrial application. In the development process of the digital person, the creation of the sequence frame grid is an important link for constructing dynamic expression of the digital person, and the quality of the sequence frame grid is directly related to the smoothness of the action presentation of the digital person and the authenticity of the visual effect. Currently, the creation of a digital human sequence frame grid in the industry depends on a preset grid generation algorithm or a manual intervention adjustment mode, but in an actual application scene, the modes face a plurality of technical bottlenecks. In a scene with high real-time requirements, such as virtual live broadcast, real-time interactive virtual assistant, etc., a digital person needs to quickly generate a sequence frame grid according to an external instruction or environmental change. However, in the prior art, the grid data of the sequence frames are difficult to monitor in real time, when the continuity anomalies such as grid vertex offset, patch connection fracture and the like occur, the continuity anomalies are often not timely perceived, so that obvious visual faults appear in the sequence frames generated subsequently, and the dynamic expression effect of digital people is affected. Even if some techniques can realize the monitoring of the sequential frame grid, the capability of targeted combing and screening of anomalies is lacking. The types of anomalies in the sequence frames are various, the influence of partial anomalies on the overall continuity is small, and the continuity of the sequence frames is seriously damaged by the partial anomalies. In the prior art, all the anomalies are treated equally, and the anomalies cannot be distinguished according to the actual influence degree of the anomalies, so that not only is the redundancy of data processing increased, but also key problems are possibly ignored due to excessive attention to tiny anomalies, and the anomaly processing efficiency is low. In the sequential frame grid creation process, the prior art lacks a system evaluation mechanism for anomaly impact. The method has the advantages that the grid creation of the subsequent frame sequence can not be dynamically adjusted according to the monitored abnormal conditions, the grid can be generated only according to a fixed algorithm flow, the abnormal influence is continuously diffused, the discontinuity of the sequence frames is further aggravated, the respective influence range and degree of different types of anomalies can not be analyzed, the differential processing strategy is difficult to formulate for different types of anomalies, the adaptability and flexibility of the grid creation process are insufficient, and the high-quality generation requirements of the digital human sequence frame grid under different scenes can not be met. Disclosure of Invention The invention aims to provide a digital human grid creation method for ensuring the continuity of sequence frames so as to solve the problems in the background technology. To achieve the above object, the present invention provides a digital personal mesh creation method for ensuring sequence frame continuity, the method comprising: A sequence frame grid monitoring step, which is used for monitoring grid data of the digital human sequence frame in real time, carrying out targeted abnormal carding and screening on the monitored grid continuity abnormality, and updating the screening result to a sequence frame continuity state diagram in real time; a continuity integration evaluation step, which is used for carrying out integration calculation and data analysis of the abnormal influence of the corresponding frame sequence according to the real-time update of the sequence frame continuity state diagram, and carrying out first dynamic management on the grid creation of the subsequent frame sequence according to the analysis result; and the anomaly type expansion evaluation step is used for carrying out integration calculation and data analysis of the influences of different anomaly types according to the real-time update of the different anomaly types in the sequence frame continuity state diagram, and carrying out second dynamic management on the creation process of the digital human grid according to the analysis result. Preferably, the step of monitoring the sequence frame grid includes: acquiring an abnormal type corresponding to the sequence frame in the abnormal state, and