AI Image Frame Selection for Automatic Best Shot
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image capturing technologies struggle to accurately identify and provide the most appropriate image frame from a series of frames in a video without user intervention, often resulting in unnecessary image storage and limited composition applicability.
Innovation Solution
An electronic apparatus utilizing an artificial intelligence model to analyze and identify the best image frame by determining the degree of matching between input frames and predetermined feature information, allowing for automatic selection and provision of the desired image frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If continuous photographing is performed for a specific time, then the user can select the most desired picture among photographed images, but a large number of unnecessary images remain in memory
Solution Approach 1:
The patent extracts only the essential and meaningful images from the continuous photographing sequence using AI-based analysis. The system identifies and extracts key moments (such as smiles, eye contacts, or specific expressions) while automatically discarding redundant frames, thereby reducing memory storage requirements while maintaining user satisfaction.
Solution Approach 2:
The system performs automatic image selection and optimization without requiring user intervention. The AI algorithm autonomously analyzes photographed images, determines the most desirable frames based on predefined criteria (such as facial expressions, composition, or moment significance), and presents the optimized set to the user, eliminating the need for manual selection of numerous images.
2Measurement precision
If photographing is performed based on predetermined composition, then the picture can be taken at a specific moment, but a tripod installation is necessary and utility range is limited
Solution Approach 1:
The patent transitions from static, predetermined composition rules to dynamic, adaptive composition detection. The AI system continuously analyzes incoming video frames in real-time, dynamically identifying moments that satisfy composition criteria (such as rule of thirds, golden ratio, or subject positioning) without requiring the camera to be fixed on a tripod. This enables handheld shooting while maintaining compositional accuracy.
Solution Approach 2:
The system changes the parameters of composition detection from fixed, pre-programmed rules to flexible, learnable parameters through machine learning. The AI model can be trained on diverse composition examples and adapt to different shooting scenarios, enabling versatile application across various subjects and situations while maintaining precise moment capture.
3Extent of automation
If AI system is used to select best image among plurality of images, then the most appropriate image can be provided automatically, but computation process requires optimization
Solution Approach 1:
The patent segments the image analysis process into multiple stages: first, a coarse filtering stage that quickly identifies frames containing potential key moments using simplified features; second, a detailed analysis stage that applies comprehensive AI evaluation only to the filtered subset of frames. This segmentation reduces the overall computational load while maintaining accurate automatic selection.
Solution Approach 2:
The system applies partial AI analysis to all frames (identifying basic features like presence of faces, expressions, or motion) and excessive/detailed analysis only to selected frames that show potential key moments. This approach balances automation extent with computation efficiency, providing automatic best image selection without requiring full-computation processing of every frame in the sequence.
Data Source
Figure 1~3
Figure 4
Figure 5a~6a
AI summary
An electronic apparatus is provided. The electronic apparatus includes: a camera, a processor, and a memory configured to store a network model trained to determine a degree of matching between an input image frame and predetermined feature information, wherein the processor is configured to: identify a representative image frame based on a degree of matching obtained by applying image frames, selected from among a plurality of image frames, to the trained network model, while the plurality of image frames are captured through the camera, identify a best image frame based on a degree of matching obtained by applying image frames within a specific section including the identified representative image frame, to the trained network model, from among the plurality of image frames, and provide the identified best image frame.