Image Feature Fusion for Robust Object Recognition Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image-based object recognition systems face challenges in accurately recognizing objects due to variations in lighting, orientation, and other factors, leading to inconsistent feature descriptors across different images of the same object captured under varying conditions.
Innovation Solution
Generate synthetically altered images of an object to replicate expected capture conditions and combine features from both captured and synthetically generated images using a common coordinate system, enhancing the robustness of the object model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple images of the same object captured under different conditions are stored in the recognition database, then the recognition accuracy under varying conditions is improved, but the database size and processing complexity increase
Solution Approach 1:
The patent creates synthetic copies of the original image by applying geometric transformations (rotation, scaling, flipping) to generate multiple transformed images. These synthetic copies simulate different capture conditions without requiring actual multiple photographs, thereby populating the recognition database with varied representations while avoiding the complexity of storing and processing numerous real captured images.
2Reliability
If feature descriptors are extracted from multiple images captured under different conditions, then the robustness of the object model is improved, but the computational time and resources increase
Solution Approach 1:
The patent performs geometric transformations and feature extraction on synthetic images during the database population phase (pre-processing), rather than during actual recognition operations. By pre-computing descriptors from transformed images and storing them in the recognition database, the system prepares robust object models in advance, eliminating the need for time-consuming real-time processing during query operations.
3Productivity
If the recognition system uses invariant feature descriptors, then the recognition speed is improved, but the accuracy under varying lighting and orientation conditions deteriorates
Solution Approach 1:
The patent applies geometric transformations (rotation, scaling, flipping) to the original image to generate synthetic images with varied parameters. By training the recognition system on these transformed images with different geometric parameters, the system learns to recognize objects under varying conditions while maintaining the efficiency of descriptor-based matching.
Data Source
AI summary
Methods, systems, and articles of manufacture to improve image recognition searching are disclosed. In some embodiments, a first document image of a known object is used to generate one or more other document images of the same object by applying one or more techniques for synthetically generating images. The synthetically generated images correspond to different variations in conditions under which a potential query image might be captured. Extracted features from an initial image of a known object and features extracted from the one or more synthetically generated images are stored, along with their locations, as part of a common model of the known object. In other embodiments, image recognition search effectiveness is improved by transforming the location of features of multiple images of a same known object into a common coordinate system. This can enhance the accuracy of certain aspects of existing image search/recognition techniques including, for example, geometric verification.