Face Detection Based Time-Series Frame Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for reproducing or editing time-series frames of moving images are inefficient as they require browsing through all frames and selecting desired scenes, with fixed time intervals leading to an excessive number of representative frames, which can differ from user preferences.
Innovation Solution
An image processing apparatus that accepts user inputs to display a list of detected faces, allowing users to specify time-series sections and representative frames, enabling the display and reproduction of images based on face transitions, with options for creating new images from selected frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If fixed time intervals are used to calculate maximum evaluation values, then representative frames can be selected automatically, but the number of representative frames becomes excessive and may not match user preferences
Solution Approach 1:
The patent applies dynamics by making the time interval for calculating maximum evaluation values variable rather than fixed. The control section dynamically adjusts the time interval based on user input and face detection results, allowing the system to automatically select an appropriate interval that prevents excessive representative frames while maintaining automation. This resolves the contradiction by enabling adaptive automation that responds to actual content characteristics.
2Measurement precision
If short time intervals are used to strictly determine representative frames, then frame selection accuracy improves, but the number of representative frames becomes enormous causing inconvenience
Solution Approach 1:
The system dynamically adjusts the time interval based on face detection density and user preferences. When faces appear frequently, the interval increases to reduce the number of representative frames. When faces appear sparsely, the interval decreases to maintain selection accuracy. This dynamic adjustment resolves the contradiction between precision and ease of operation.
Solution Approach 2:
The patent changes the parameter of time interval from a fixed value to a variable that adapts based on face detection results and user input. By modifying this parameter dynamically, the system achieves both accurate representative frame selection and manageable quantities, resolving the contradiction between measurement precision and operational ease.
3Ease of operation
If all frames are browsed to select desired scenes, then complete user control is achieved, but the process is time-consuming and inefficient
Solution Approach 1:
The system performs automatic face detection and representative frame selection based on predetermined time intervals, eliminating the need for users to manually browse all frames. The automated system serves itself by selecting frames that contain detected faces, significantly reducing the time users need to spend while maintaining the ability to review and adjust selections. This resolves the contradiction between user control and time efficiency.
Solution Approach 2:
The system performs preliminary face detection and automatic representative frame selection before user review. By pre-processing the frame selection based on face detection, the system reduces the workload for users and eliminates the need to manually examine all frames, thereby resolving the contradiction between user control and time consumption.
4Adaptability or versatility
If conventional directional methods are used for frame selection, then all frames can be reviewed, but the method is not efficient as skill levels differ among individuals
Solution Approach 1:
The system automatically performs face detection and representative frame selection without requiring user expertise in frame browsing techniques. By making the system self-sufficient in identifying important frames through face detection, it eliminates the dependency on individual user skills while maintaining the ability to review results, thereby resolving the contradiction between adaptability and productivity.
Data Source
AI summary
An image processing apparatus may include an input accepting unit configured to accept an operation input about an input image; a time-series section generating unit configured to accept information specifying a time-series section where frames including a face of a specified person appear in the input image and information specifying a representative frame of the time-series section and to draw display of the time-series section and the representative frame on the basis of the information; and an image control unit configured to allow the time-series section generating unit to draw display of the time-series section in accordance with the operation input.