Visual Summarization of Time-Stamped Images Using Object Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face difficulties in efficiently viewing and extracting important information from large collections of images, such as surveillance footage, due to the time-consuming process of understanding relationships between images based on time and location.
Innovation Solution
The method provides visual summarization of time-stamped images by allowing users to annotate objects of interest, filter images based on these annotations, and generate timelines and tables of contents to highlight relevant information, thereby reducing clutter and enhancing understanding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If users review each image individually to understand relationships between images, then information extraction completeness is improved, but time consumption increases significantly
Solution Approach 1:
The patent segments the large set of images by filtering to create a focused subset containing only images with annotated objects of interest. This segmentation reduces the total number of images users need to review while preserving all relevant information about specific objects, thereby resolving the contradiction between complete information extraction and time consumption.
Solution Approach 2:
The system performs preliminary annotation of objects in images before user review. By pre-identifying and annotating objects of interest across the image set, the system prepares the data in advance, allowing users to quickly locate and review only relevant images without manually examining every image, thus reducing time consumption while maintaining information completeness.
2Reliability
If users annotate multiple objects across numerous images, then object tracking capability is improved, but system complexity increases
Solution Approach 1:
The patent creates a simplified copy or representation of the complex annotated image data by generating a filtered subset that captures essential object information. This copying approach maintains accurate object tracking capability while presenting a less complex, more manageable dataset to users, resolving the contradiction between tracking reliability and system complexity.
3Loss of information
If all images are displayed to users, then information completeness is improved, but ease of operation deteriorates due to information overload
Solution Approach 1:
The patent extracts and removes irrelevant images from the complete set, keeping only those containing annotated objects of interest. This extraction process eliminates unnecessary information that would contribute to overload, while preserving all relevant information about target objects, thereby improving ease of operation without sacrificing information completeness.
Data Source
AI summary
Disclosed is an exemplary method of providing a visual summarization of a plurality of time-stamped annotated images. The method includes receiving the plurality of time-stamped annotated images. The plurality of time-stamped annotated images includes a plurality of object-based annotations. The method includes receiving an object of interest selection (OOI selection) corresponding to at least a first object-based annotation of the plurality of object-based annotations, thereby designating an object of interest (OOI). The method includes filtering the plurality of time-stamped annotated images, based at least in part on the OOI selection, to a subset of the plurality of time-stamped annotated images, thereby creating an object of interest image group (OOI image group). The method includes generating a visual timeline for concurrently displaying the OOI image group.