Progressive Pattern Matching in Mobile Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile applications that rely on pattern recognition, such as augmented reality and image recognition, face limitations due to high computational and communication burdens, which can hinder user experience and strain network resources.
Innovation Solution
A method where a communication device performs initial processing and local searches on an object, then progressively sends reduced sets of features to a server for identification, reducing computational and communication costs by transmitting smaller data sets until a match meeting a confidence factor is found.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full-resolution objects are transmitted to the server for pattern matching, then identification accuracy is improved, but communication costs and network burden increase
Solution Approach 1:
The patent segments the object data into progressive resolutions (full-resolution, mid-resolution, low-resolution). The communication device first transmits low-resolution segments to the server for preliminary matching. If no match is found, progressively higher resolution segments are transmitted. This segmentation strategy enables accurate identification while minimizing communication data volume by transmitting only the necessary resolution level.
2Measurement precision
If complete object data is processed locally on the communication device, then identification accuracy is improved, but computational power requirements and device complexity increase
Solution Approach 1:
The patent divides the computational workload into segments: the communication device performs local pattern matching using a local library first (low computational requirement), and only if no match is found does it transmit data to the server for more intensive processing. This segmentation reduces the computational power requirements and device complexity needed at the communication device while maintaining identification accuracy through progressive escalation to server-side processing when necessary.
3Speed
If a local object library is maintained on the communication device, then identification speed is improved, but device storage requirements and library management complexity increase
Solution Approach 1:
The patent applies partial action by maintaining only a local object library on the communication device rather than storing all possible objects. This partial library enables fast local identification for common objects, while less common objects are identified through server assistance. This approach achieves good identification speed for the majority of cases while keeping local storage requirements manageable.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods and apparatus for performing efficient pattern matching in a client-server network are described herein. A communication device acquires an object to be matched. At least one reduced set of features is extracted from the object to be matched and a determination as to whether the object to be matched matches one of a plurality of known objects in a local library stored on the communication device is generated. An indication representing a match of the object is presented if the determination indicates a local match exists in the local library. A match request including the at least one reduced set of features is transmitted for a remote matching determination when the determination indicates that no local match exists in the local library. In accordance with some aspects, a series of one or more match requests may be transmitted for a remote matching determination.