Moving Picture Common Section Detection via Feature Vector Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current moving picture search technologies are inefficient due to the high computational power required for direct comparison of large-size moving pictures, often resulting in repetitive or similar results, and lack effective methods to detect commonality or similarity between images for copyright infringement detection.
Innovation Solution
A method that generates feature vectors from moving pictures by analyzing color distribution, first-order, and second-order differentials of sub-frames, allowing for efficient comparison and detection of common sections between moving pictures, which can be used to cluster similar content and identify copyright infringement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If direct comparison of moving pictures in binaries is performed to determine commonality or similarity, then measurement precision is improved, but use of energy and calculating power increase excessively
Solution Approach 1:
The patent divides the moving picture into multiple frames and extracts feature vectors from each frame independently. This segmentation allows comparison of essential visual features without processing the entire binary data stream, significantly reducing computational power while maintaining detection accuracy.
Solution Approach 2:
The patent extracts key feature vectors (such as color distribution, edge information, and motion features) from moving pictures to create a condensed representation. This extraction process removes unnecessary data while preserving the essential characteristics needed for similarity detection, reducing both computational load and data size.
2Measurement precision
If large-size moving pictures are compared using relatively small size comparison criterion, then measurement precision is improved, but loss of time increases as it is time-consuming work
Solution Approach 1:
The patent performs preliminary extraction of feature vectors from moving pictures before the actual comparison process. This preliminary action prepares condensed representations that can be compared quickly, avoiding the need to process large-size original files during the time-critical comparison stage.
Solution Approach 2:
The patent transforms the moving picture data from its original large-size format into feature vector representations with different parameters (dimensionality, data type, structure). This parameter change enables efficient comparison while maintaining the ability to detect similarity accurately.
3Productivity
If feature vectors of moving pictures are used for comparison, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent designs a feature extraction system that can handle multiple types of moving pictures and comparison tasks using the same core methodology. The feature vector extraction process is universal and can be applied to various video formats and comparison scenarios, reducing the need for multiple specialized systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of processing a moving picture comprises generating a frame feature vector for each of a first moving picture and a second moving picture; and detecting a common section of the first moving picture and the second moving picture by comparing the frame feature vector of the first moving picture with the frame feature vector of the second moving picture. The detecting of the common section comprises comparing moving picture segments by generating commonality evaluation value of a first moving picture segment and a second moving picture segment by comparing feature vectors of p frames of the first moving picture segment of the first moving picture with feature vectors, corresponding arrangement to an arrangement of the p frames of the first moving picture segment, of p frames of the second moving picture segment of the second moving picture, p being a natural number of greater than or equal to 1. The first moving picture segment has a start time of tl after the first moving picture starts and the second moving picture segment has a start time of t2 after the second moving picture starts. The comparing of moving picture segments is repeated for tl and t2, tl being a start time of the first moving picture segment that is equal to or greater than 0 and is smaller than a length of the first moving picture, t2 being a start time of the second moving picture segment that is equal to or greater than 0 and is smaller than a length of the second moving picture. The detecting of common section comprises detecting a common section start-end point by detecting a start time and an end time of a common section in the first moving picture and the second moving picture, respectively, by comparing the feature vector of the first moving picture with the feature vector of the second moving picture by applying a greater number of frames per second than the number of frames per second of the p frames in the first moving picture segment and the second moving picture segment, in case the commonality evaluation value indicates that there is commonality in the first moving picture segment and the second moving picture segment.