Video Comparison Using Orientation-Grouped Movement Indicators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The proliferation of duplicate videos in video sharing platforms makes it difficult for users to find the desired content, as existing video copy detection techniques are inefficient in handling large databases and providing real-time responses, leading to increased network overhead and user dissatisfaction.
Innovation Solution
A method that detects local features in video frames, generates movement indicators, and groups them by orientation ranges to create time series values, which are then compared between videos to determine similarity, using computationally efficient and robust local feature descriptors like ORB and SIFT, and applying descriptor matching to find corresponding features across frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional video copy detection techniques are used to search large video databases, then duplicate videos can be detected, but the processing time is too long and real-time response cannot be provided
Solution Approach 1:
The patent segments video content into multiple shots and extracts features from each shot independently. Instead of processing entire videos at once, the system divides videos into smaller temporal units (shots) and extracts keyframe features from each, enabling parallel processing and reducing overall computation time while maintaining detection accuracy
Solution Approach 2:
The patent performs preliminary feature extraction by identifying keyframes and extracting descriptors from them before actual video comparison. Shot detection and keyframe selection are performed in advance, creating a compressed feature representation that can be quickly compared without processing all video frames in real-time
2Measurement precision
If more video features are extracted and stored to improve detection accuracy, then duplicate detection becomes more precise, but storage requirements and computational overhead increase
Solution Approach 1:
The patent extracts only the most discriminative features from video content by selecting keyframes based on shot boundaries and motion detection. Instead of storing or processing all video data, the system extracts essential visual descriptors (SIFT, SURF, or ORB features) from representative keyframes only, significantly reducing storage requirements while preserving detection capability
Solution Approach 2:
The patent applies different feature extraction strategies to different parts of video content. Local SIFT/SURF/ORB descriptors are extracted from salient regions within keyframes, and motion information is extracted from regions with significant pixel changes. This localized feature extraction focuses computational resources on discriminative regions rather than processing uniform areas
3Reliability
If comprehensive video analysis is performed to detect various types of duplicates (including edited versions), then detection robustness improves, but processing complexity increases
Solution Approach 1:
The patent implements a universal detection framework that handles multiple types of video duplicates using the same core methodology. The shot-based feature extraction and keyframe comparison approach works for exact copies, cropped versions, and edited videos without requiring separate detection algorithms for each case, reducing system complexity while maintaining broad detection capability
Solution Approach 2:
The patent uses dynamic shot detection that adapts to different video content types and editing styles. Motion detection thresholds and shot boundary identification are adjusted based on content characteristics, allowing the system to robustly detect duplicates across various video formats and editing operations without fixed rigid parameters
Data Source
Figure 1
Figure 2~3
Figure 4A~4D
AI summary
A method for comparing videos includes detecting local features in frames of a video and determining the positions of detected local features in a first frame compared to their respective positions in a succeeding second frame. For at least some of the detected local features, a movement indicator is generated representing the position of a local feature in the first frame relative to the position of that local feature in the second frame, the movement indicator including an orientation, to obtain a plurality of movement indicators. Movement indicators are grouped by orientation ranges. The number of grouped movement indicators in an orientation range are summed to obtain a value for that orientation range. A time series of values for at least one given orientation range is obtained by determining the positions of detected local features in a first frame compared to their respective positions in a succeeding second frame for a plurality of pairs of frames. For at least one orientation range, a first time series of values for a first video is compared with a second time series of values for a second video to determine the similarity of the first and second videos.