Multi-Channel Video Search Using Similarity-Filtered Object Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing surveillance systems face challenges in efficiently searching multi-channel videos due to the large amount of data required for storing object features, leading to slow searches and the need for extensive databases.
Innovation Solution
A video search device and method that utilize object features and channel numbers as search criteria, along with time information, to create a minimized database by generating final object information with high similarity and temporary information with lower similarity, allowing for quick and easy video retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If object features are extracted and stored for each multi-channel video to enable feature-based search, then search capability is improved, but database size increases and search speed decreases
Solution Approach 1:
The patent segments object information into two categories: final object information (high similarity, stored permanently) and temporary object information (lower similarity, stored temporarily). This segmentation allows the system to maintain comprehensive search capability while reducing database size by not storing all extracted features indefinitely, thus improving search speed without sacrificing adaptability.
Solution Approach 2:
The patent changes the parameter of information retention by introducing a similarity threshold mechanism. Object information is retained based on its similarity score - high similarity objects become final object information while lower similarity objects become temporary. This parameter-based classification optimizes the balance between database size and search capability.
2Measurement precision
If all extracted object features are stored in the database to ensure comprehensive search coverage, then search accuracy is improved, but device complexity and storage requirements increase
Solution Approach 1:
The patent discards temporary object information that falls below the similarity threshold, reducing database complexity. However, the system recovers this information by storing it as temporary object information with a time limit, allowing potential retrieval if needed. This approach maintains search accuracy for important objects while reducing overall database size and complexity.
3Measurement precision
If feature extraction is performed for every channel to enable precise object search, then search precision is improved, but processing time and computational resources increase
Solution Approach 1:
The patent applies partial action by not storing all extracted feature information permanently. Instead, only final object information (above threshold) is stored long-term, while temporary object information (below threshold) is stored briefly or discarded. This partial storage approach maintains search precision for relevant objects while reducing processing time and computational resources required for comprehensive storage and retrieval.
Data Source
AI summary
Disclosed herein is a video search device comprising: at least one processor configured to implement an updating unit generating event object information based on feature information on an object detected in a video; and a storage configured to store comparison object information containing feature information on a reference object, wherein the processor is further configured to implement: an object information generating/deleting unit configured to create information on an object having a predetermined degree of similarity between the event object information and the comparison object information as final object information, and storing the final object information in the storage; and a data searching unit searching the final object information for data that satisfies a search criterion entered by a user.