A method for optimizing video post - editing and video synthesis

Abstract

The present disclosure relates to the field of film and television media, and particularly to a method for optimizing video post-editing and video synthesis. Specifically, the present invention includes: S1, splitting the original video file into video key frames; S2, extracting dynamic video features from the split video key frames; S3, synthesizing the extracted dynamic video segments; wherein in the video synthesis, automatic splicing optimization of the dynamic video segments is established. The method for optimizing video post-editing and video synthesis of the present invention establishes the extraction of dynamic video features. By establishing the splitting of video features and the auxiliary judgment with audio features in the extraction of dynamic video features, targeted video frame collection and summarization are carried out under specified features, and the features to be extracted are optimized by establishing automatic video transition and pixel point adjustment at the video splicing point, so as to better improve the video editing efficiency of short videos and the back-end of self-media, and at the same time improve the quality of automatic video editing and synthesis.

Description

Technical Field

[0001] The present disclosure relates to the field of film and television media, and particularly to a method for optimizing video post - editing and video synthesis. Background Art

[0002] At present, with the development of the film and television industry and the rise of self - media, video editing has gradually become an important technology serving the film production work. However, the traditional video editing process usually mainly adopts manual editing supplemented by software editing. For films with a large volume or a large number, the editing efficiency is relatively low during the synchronous editing process. At the same time, the editing quality of the final finished films is also uneven. To address this problem, automated video editing technology has emerged. However, existing automatic editing technologies still require manual intervention and selection of features for specific video features. After selecting the features to be edited, they cannot automatically extract and select the feature videos, and fundamentally do not further realize the automated editing process of videos.

[0003] Patent CN202110001261 provides a video editing method, a video editing device, and a storage medium. In this patent, by presetting a video editing sequence and automatically integrating the preset video editing sequence, fast video editing is achieved. However, the fast video editing involved in this patent is not automatic video editing. It requires manual editing planning and completing the video editing process according to the editing plan.

[0004] Patent CN201710112722 provides a video editing method and device based on deep learning. In this patent, the speech features in the video are converted into text data, and deep learning algorithms are used to train and match the combined text data and video image features. However, this patent does not involve the problem of how to uniformly learn dynamic video features during the video editing process. At the same time, due to the fact that a single feature in the video in a dynamic environment will have different motion forms, the resulting large amount of feature data will lead to a long training time and low video editing efficiency.

[0005] Therefore, in view of the problems existing in the existing automatic video editing methods, the present invention provides a method for optimizing video post - editing and video synthesis. Summary of the Invention

[0006] In view of the above - mentioned problems, the present invention provides a method for optimizing video post - editing and video synthesis, specifically including: S1, splitting the original video file into video key frames; S2, establishing dynamic video feature extraction for the split video key frames; S3, synthesizing the extracted dynamic video segments; wherein in the video synthesis, automatic splicing optimization of dynamic video segments is established.

[0007] Preferably, in the extraction of dynamic video features, the split single video key frames are divided into audio features and video features in the same time series.

[0008] Preferably, static video images are extracted from multiple video key frames, and features to be extracted are customized in the static video images.

[0009] Specifically, the features to be extracted in the static video images are first obtained through video analysis before video editing, and the extraction features obtained through video analysis are listed for the personnel to be edited to select.

[0010] Preferably, for the features to be extracted, feature points and feature line segments are split and extracted under multiple objectives, and the split features to be extracted are transmitted to the database for summary storage.

[0011] Preferably, in the extraction of dynamic video features, the features to be extracted in the video key frames are locked as specified features, and through training on the specified features, feature video frames containing the specified features in the complete video are automatically edited.

[0012] Preferably, in the automatic editing, on the basis of the feature training, the feature video frames are converted into pixel point matrices, and by calculating the difference between the pixel points of the latter pixel point matrix and the former pixel point matrix, it is used to judge whether there are specified features in two adjacent video key frames.

[0013] Specifically, the calculation of the pixel point matrix is used to identify the final frames of the features to be extracted involved in the dynamic transformation in the dynamic video.

[0014] Preferably, the feature video frames are combined with audio features to judge the appearance probability of the specified features in the feature video frames, and through the custom setting of the appearance probability values, the feature video frames are automatically screened.

[0015] Specifically, the appearance probability values are used to judge the number of specified features contained in the intercepted feature video frames, and whether the feature video frames are the feature video frames required by the editor is judged by the number of specified features, so as to further obtain the dynamic video required by the editor and improve the quality of video editing.

[0016] Preferably, in the automatic screening of the feature video frames, video key frames with continuous time series are selected to integrate the feature video frames to form video segments with specified features, and the dynamic video segments are automatically spliced and optimized.

[0017] Preferably, in the video clip, if the difference between pixel matrices exceeds the threshold under a specified feature, then the specified feature does not exist in the subsequent pixel matrix, and the previous pixel matrix is the final frame corresponding to the video clip.

[0018] Preferably, for the optimization of automatic splicing of dynamic video clips, a method based on video automatic transition and pixel optimization adjustment is established.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0020] The video post-editing and video synthesis optimization method of the present invention establishes dynamic video feature extraction. In the dynamic video feature extraction, the splitting of video features and the auxiliary judgment with audio features are established to collect and summarize targeted video frames under specified features, and the features to be extracted are optimized by establishing video automatic transition and pixel adjustment at the video splicing point, so as to better improve the video editing efficiency of short videos and the back-end of self-media, and at the same time improve the quality of automatic video editing and synthesis. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flowchart of a video post-editing and video synthesis optimization method. DETAILED DESCRIPTION OF THE INVENTION

[0022] Example 1:

[0023] In this embodiment, a video post-editing and video synthesis optimization method described herein, as Figure 1 shown, the present invention specifically includes splitting the original video file into video key frames, establishing dynamic video feature extraction for the split video key frames, and performing video synthesis on the extracted dynamic video clips; wherein dynamic video clip automatic splicing optimization is established in the video synthesis.

[0024] Specifically, in the video post-editing and video synthesis optimization method, after confirming the features to be extracted, the summary and integration of the video key frames involved in the features to be extracted can be automatically performed. Since there may be problems such as transitional jamming, low transition smoothness, and inconsistent playback time axes due to different video lengths during the integration of different video clips, resulting in black screens during playback. To further overcome this problem, the present invention establishes dynamic video clip automatic splicing optimization on the basis of video synthesis.

[0025] In the dynamic video feature extraction, the split single video key frame is divided into audio features and video features in the same time sequence.

[0026] Specifically, the audio features are used to distinguish the sound subjects in dynamic videos. The sound subjects are used as an auxiliary recognition calculation method for the appearance probability of specified features during the extraction process of dynamic video features, to determine whether the selected feature to be extracted in a single video key frame is the main appearance subject, and through custom options, video key frames with non-main appearance subjects can be excluded, so as to retain a video key frame with the feature to be extracted as the main subject and a higher appearance rate of the feature to be extracted.

[0027] Extract static video images from multiple video key frames, and customize the features to be extracted in the static video images.

[0028] For the features to be extracted, perform the splitting and extraction of feature points and feature line segments under multiple objectives, and transmit the split features to be extracted to the database for summary storage.

[0029] Specifically, for the splitting and extraction of feature points and feature line segments under multiple objectives, the feature to be extracted is split in terms of contour. When performing feature matching based on video key frames, clustering processing will be performed on the split features, so that when the feature to be extracted undergoes dynamic changes in multiple video key frames, the specified feature can be recognized based on the feature points and feature line segments under multiple objectives.

[0030] The automatic splicing and optimization of dynamic video segments establish a method based on video automatic transition and pixel point optimization adjustment.

[0031] Specifically, the automatic splicing and optimization of dynamic video segments are used to further fuse the combined part of two synthesized video segments, and on the basis of not adding transition effects, further achieve smooth and natural transition of video segments. Among them, for the method of video automatic transition and pixel point optimization adjustment, an interpolation process based on brightness values is performed on the established pixel matrix between video segments, so as to adjust the illumination difference at the connection of video segments to achieve a soft transition at the connection of video segments.

[0032] In summary, the specific process of the video post-editing and video synthesis optimization method described in the present invention is as follows: First, the original video file is split based on video key frames, the split video key frames are analyzed, static pictures are extracted to list the features to be extracted, the editor selects the features to be extracted, after the selection is completed, the feature points and feature line segments of the features to be extracted under multiple objectives are split, and the obtained feature points and feature line segments are stored in the database. On this basis, the features to be extracted in the video key frames are locked as specified features, and through the training of the specified features, the feature video frames containing the specified features in the complete video are automatically edited. On the basis of the feature training, the feature video frames are converted into pixel point matrices, and by calculating the difference between the pixel points of the latter pixel point matrix and the former pixel point matrix, it is used to judge whether there are specified features in two adjacent video key frames. At the same time, the audio features are imported, and by judging the appearance probability of the specified features in the feature video frames, through the custom setting of the appearance probability value, the feature video frames are automatically screened. By selecting the video key frames with continuous time series, the feature video frames are integrated to form a video segment with specified features, and the dynamic video segment is automatically spliced and optimized.

Claims

1. A method for optimizing video post - editing and video synthesis, characterized in that, Specifically, it includes: S1, splitting the original video file into video key frames; S2, extracting dynamic video features from the split video key frames; S3, synthesizing the extracted dynamic video segments; among which, automatic splicing optimization of dynamic video segments is established in the video synthesis; In the extraction of the dynamic video features, the split single video key frame is divided into audio features and video features in the same time sequence; For the video features, static video images are extracted from multiple video key frames, and features to be extracted are customized in the static video images; For the features to be extracted, splitting extraction of feature points and feature line segments under multiple objectives is established, and the split features to be extracted are transmitted to the database for summary storage; In the extraction of the dynamic video features, the features to be extracted in the video key frames are locked as specified features, and through training the specified features, feature video frames containing the specified features in the complete video are automatically clipped; For the automatic clipping, on the basis of the feature training, the feature video frames are converted into pixel point matrices, and by calculating the difference between the pixel points of the latter pixel point matrix and the former pixel point matrix, it is used to judge whether there are specified features in two adjacent video key frames; The feature video frames are combined with the audio features to judge the appearance probability of the specified features in the feature video frames, and through the custom setting of the appearance probability values, automatic screening of the feature video frames is carried out; For the automatic screening of the feature video frames, by selecting video key frames with continuous time sequences, the feature video frames are integrated to form a video segment with specified features, and automatic splicing optimization of dynamic video segments is carried out on the video segment; 2. The video post - editing and video synthesis optimization method according to claim 1, wherein In the video segment, if the difference of the pixel point matrices exceeds the threshold under the specified features, there are no specified features in the latter pixel point matrix, and the former pixel point matrix is the final frame corresponding to the video segment; 3. A method for optimizing video post - editing and video synthesis according to claim 1, characterized in that, For the automatic splicing optimization of dynamic video segments, a method based on video automatic transition and pixel point optimization adjustment is established.

Images