Image Classification Using Reference Distance for Event Grouping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image-classifying methods fail to accurately group images corresponding to user-recognized events, such as trips or home parties, due to inconsistencies between user perception and automated classification based on time and location information.
Innovation Solution
An image-classifying apparatus that calculates a reference distance between the image capture location and a user-defined reference position, using this distance to classify images into groups that align with the user's cognitive understanding of events, with optional consideration of time differences and distance thresholds for more detailed classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If images are classified based on time information only, then images captured on the same day are grouped together, but images captured on different days for the same event are divided into separate groups
Solution Approach 1:
The patent segments the classification criteria into multiple dimensions: time information, position information, and movement distance. By dividing the classification task into these segments, the system can evaluate images based on comprehensive criteria rather than relying solely on time, thus preventing the fragmentation of single events across multiple groups while maintaining efficient automated processing.
2Productivity
If images are classified based on position information only, then images from the same location are grouped together, but images from different locations for the same event are divided into separate groups
Solution Approach 1:
The classification system is segmented into multiple evaluation dimensions including position information, time information, and movement distance. This segmentation allows the system to consider images from different locations as part of the same event when the movement distance and time interval suggest continuous activity, thereby improving classification accuracy while maintaining automated efficiency.
3Device complexity
If strict time and position boundaries are used for classification, then classification processing is simple and fast, but user-perceived events are incorrectly split into multiple groups
Solution Approach 1:
The patent introduces dynamic evaluation criteria that adapt to user behavior patterns. Instead of using fixed time and position boundaries, the system dynamically calculates movement distances and evaluates time intervals in context, allowing classification boundaries to flex according to the actual event characteristics. This dynamic approach maintains processing simplicity while significantly improving classification accuracy.
4Reliability
If flexible classification criteria incorporating movement distance are used, then user-perceived events are accurately grouped, but classification processing becomes more complex
Solution Approach 1:
The patent changes the classification parameters from simple time and position thresholds to include movement distance calculations. By introducing this new parameter and establishing corresponding evaluation standards, the system achieves more accurate classification that reflects user perception of events, while the added complexity is managed through systematic parameter evaluation.
Data Source
AI summary
An image information-inputting unit inputs image information including position information indicating a position where an image was captured. A reference distance-calculating unit calculates a reference distance from a predetermined reference position utilizing the image information. An image information-classifying unit classifies the image information based on the reference distance.