Digital Signature Object Recognition in Augmented Reality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Augmented Reality (AR) approaches that rely on 3D mapping and localization are inefficient and require significant processing power, making them difficult to scale for real-time applications.
Innovation Solution
A method and apparatus that utilize digital signatures to efficiently identify objects of interest within images by training a signature encoding module to generate compact digital signatures representing objects and their context, which can be used by a query processing module to recognize these objects in other images, allowing for efficient object recognition and highlighting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 3D mapping and localization approaches are used for object identification in AR applications, then measurement precision and object recognition accuracy are improved, but processing power requirements and device complexity increase significantly
Solution Approach 1:
The patent extracts only the essential visual features needed for object identification rather than performing complete 3D mapping. By taking out only the necessary image patches and visual descriptors required for recognition, the system achieves accurate object identification without the computational burden of full 3D reconstruction and localization.
Solution Approach 2:
The patent segments the image processing task into distinct components: extracting image patches containing objects of interest, generating visual descriptors for these patches, and matching them against a database. This segmentation allows each component to be optimized independently, reducing overall processing complexity while maintaining recognition accuracy.
2Measurement precision
If exhaustive back end processing is performed for 3D localization, then object identification accuracy is improved, but processing time and productivity decrease
Solution Approach 1:
The patent performs preliminary actions by pre-extracting and storing visual descriptors for objects of interest in a database before runtime. During actual AR processing, the system only needs to extract image patches and perform matching against pre-computed descriptors, rather than performing exhaustive 3D localization processing in real-time. This preliminary preparation significantly speeds up processing while maintaining identification accuracy.
3Loss of information
If large amounts of data are generated for 3D mapping, then comprehensive object information is captured, but data storage and transmission requirements increase
Solution Approach 1:
The patent extracts only the essential visual features and image patches needed for object identification rather than storing complete 3D point cloud data. By taking out only the necessary visual descriptors and representative image patches, the system maintains comprehensive object information for recognition purposes while dramatically reducing data volume for storage and transmission.
Data Source
AI summary
A method, apparatus and computer program product train a signature encoding module and a query processing module. In a method, at least one of the signature encoding module and the query processing module is trained by providing the signature encoding module with a reference image containing a particular object of interest that is marked therewithin. The method generates a digital signature of the object of interest and at least some context associated therewith and provides the query processing module with a query image and the digital signature representing the object of interest and at least some of the context associated therewith. The method additionally identifies the object of interest within the query image based upon the digital signature and modifies at least one of the signature encoding module or the query processing module based upon a qualitative or quantitative difference between the objects of interest identified within the query image and marked in the reference image.