Key Frame Selection via Two-Stage Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for selecting key frames from moving images are inefficient due to the time-consuming process of obtaining multiple feature amounts, which prolongs the frame selection process, especially when dealing with a large number of frames.
Innovation Solution
An information processing apparatus that prioritizes the selection of frames by first obtaining a first feature amount, then selectively obtaining a second feature amount only for frames with longer processing times, allowing for efficient key frame selection by focusing on frames with higher evaluation scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple feature amounts are obtained for each frame to improve selection accuracy, then frame selection accuracy is improved, but processing time increases proportionally
Solution Approach 1:
The patent segments the feature amount calculation process into two distinct stages: a first stage that calculates a first feature amount for all frames using a first calculation expression, and a second stage that calculates a second feature amount for only selected frames using a second calculation expression. This segmentation allows the system to efficiently screen all frames initially, then apply more computationally intensive analysis only to promising candidates, thereby maintaining accuracy while reducing overall processing time.
Solution Approach 2:
The patent applies partial action by calculating the second feature amount (which requires longer processing time) not for all frames but only for frames that meet certain criteria based on the first feature amount. This selective approach means that the more time-consuming calculation is performed partially rather than excessively on every frame, thus reducing total processing time while still ensuring that the most important frames are evaluated with high accuracy.
2Measurement precision
If all types of feature amounts are obtained for every frame, then selection accuracy is improved, but the time required for frame selection increases significantly
Solution Approach 1:
The patent performs preliminary action by first calculating the first feature amount for all frames before proceeding to calculate the second feature amount. This preliminary calculation serves as a screening step that identifies which frames are worth further analysis, allowing the system to prepare in advance and avoid performing the more time-consuming second feature amount calculation on frames that would not be selected anyway, thus improving overall productivity.
Solution Approach 2:
The patent applies local quality by using different calculation expressions (first and second feature amounts) for different frames based on their characteristics. Frames that show promise based on the first feature amount receive the more intensive second feature amount calculation, while frames that do not meet criteria receive only the lighter first feature amount calculation. This localized application of computational resources improves efficiency without sacrificing the accuracy needed for frames that will be selected.
Data Source
AI summary
An information processing apparatus includes at least one processor causing the information processing apparatus to act as a first obtainment unit configured to execute processing for obtaining a first feature amount for each of a plurality of frames, a specification unit configured to specify a priority order of frames for obtaining a second feature amount different from the first feature amount based on the first feature amount obtained by the first obtainment unit, a second obtainment unit configured to execute processing for obtaining the second feature amount from a frame in accordance with the priority order, and a selection unit configured to select, based on the second feature amount obtained by the second obtainment unit, an image processing target frame. The number of frames from which the second feature amount is obtained is fewer than the number of the plurality of frames from which the first feature amount is obtained.