Likelihood Distribution Generation for Robust Image Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The reliability of image feature extraction varies significantly with capturing conditions, making it difficult to accurately determine whether objects in images are identical, especially under different illumination and orientation.
Innovation Solution
A processing system that generates likelihood distribution information by associating pixel or pixel block features with categories based on sample images captured under various conditions, allowing for accurate feature processing and comparison.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image features are extracted based on capturing conditions, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system pre-generates likelihood distribution information for multiple categories (e.g., person, vehicle, animal) before actual image processing. This preliminary preparation stores pre-computed feature probabilities for different object types, allowing the processing unit to quickly retrieve and compare distributions without performing complex calculations during real-time operation, thus improving measurement precision while controlling device complexity
Solution Approach 2:
The patent transforms the feature comparison problem from direct feature matching into a probability distribution comparison in a different dimensional space. Instead of comparing raw image features directly, the system compares likelihood distributions across multiple categories, adding a probabilistic dimension that simplifies the decision-making process and improves accuracy in determining whether detected objects are identical
2Reliability
If likelihood distribution information is generated for multiple categories, then reliability is improved, but loss of time increases
Solution Approach 1:
The system pre-computes and stores likelihood distribution information for multiple categories before actual object identification is needed. This preliminary action allows the processing unit to quickly retrieve pre-generated distributions and perform simple comparisons during real-time operation, significantly reducing processing time while maintaining high reliability through multi-category analysis
Solution Approach 2:
The system generates likelihood distributions for multiple categories (excessive action) rather than focusing on a single expected category. This approach may seem to increase processing overhead, but by pre-generating these distributions and storing them, the system actually reduces real-time processing time while improving reliability through broader category coverage and more robust identification
Data Source
AI summary
A processing system includes an input unit and a generation unit. The input unit receives an input of a plurality of sample images obtained by capturing an object associated with a category in different conditions with respect to a plurality of the categories. The generation unit generates likelihood distribution information in which each value representing a possible feature of each pixel or each pixel block, and each value representing a likelihood belonging to each of the categories are associated with each other, based on a feature of each pixel or each pixel block included in an area relating to each of the objects within the plurality of sample images associated with a category.