Thumbnail Extraction from Video Using Scene and Face Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face difficulties in finding specific scenes within moving pictures due to the large amount of data and the time required to review them, making it challenging to locate desired moments.
Innovation Solution
An electronic apparatus and method for extracting and displaying thumbnail images from moving-picture data based on scene shift, face detection, and registered face recognition, prioritizing registered faces, followed by face detection and scene shifts, to facilitate easy scene selection and reproduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a moving picture is recorded to capture a scene, then the user can vividly photograph the desired scene, but the user has difficulty in finding a desired scene in the moving picture due to the large amount of data and playback time required
Solution Approach 1:
The patent extracts key frames from the moving picture based on multiple criteria including scene shifts, face detection, and registered face recognition. This extraction process isolates the most important moments from the entire video, allowing users to quickly locate desired scenes without watching the full playback, thus resolving the contradiction between complete scene capture and efficient scene retrieval
Solution Approach 2:
The system performs preliminary analysis of the moving picture during or after recording to identify and mark key frames before user review. By pre-processing the video to detect scene changes, faces, and registered individuals, the system prepares an organized thumbnail sequence that enables rapid navigation, eliminating the time loss associated with manual scene searching
2Measurement precision
If multiple thumbnail extraction criteria are used to identify key frames, then the accuracy of thumbnail selection is improved, but the complexity of the processing system increases
Solution Approach 1:
The patent divides the thumbnail extraction process into distinct functional modules: scene shift detection, face detection, and registered face recognition. Each module independently evaluates specific criteria and generates candidate key frames, which are then combined and prioritized. This segmentation allows the system to maintain high selection accuracy through multiple criteria while managing complexity through modular architecture
Solution Approach 2:
Different parts of the processing system apply different extraction criteria with varying weights based on local requirements. For example, registered face recognition is given higher priority than general face detection, which in turn is prioritized over scene shifts alone. This local quality approach ensures accurate thumbnail selection by applying the most appropriate criteria to specific content types while keeping the overall system manageable through differentiated processing
Data Source
AI summary
An electronic apparatus is provided. The electronic apparatus includes a processing unit configured to extract at least one frame selected from a plurality of frames of moving-picture data as a candidate thumbnail image, wherein the at least one frame corresponds to at least one frame selected according to a plurality of thumbnail extraction criteria, and to determine at least one candidate thumbnail image selected from a group comprising candidate thumbnail images obtained by the extracting, as at least one final thumbnail image, based on a thumbnail extraction criterion to which the candidate thumbnail images obtained by the extracting respectively correspond, and a display unit configured to display the at least one final thumbnail image.