Material Classification Using Reduced Light Source Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing material classification methods using bidirectional reflectance distribution function (BRDF) require a large number of light sources and complex illumination configurations, leading to geometric and dimensional expansions in data measurements and analysis, making them cumbersome and inefficient.
Innovation Solution
The technique reduces the number of clustered light sources from a superset to a subset of fewer sources, such as two or three, by selecting optimal angular locations and feature vectors through mathematical clustering and training a classification engine, without significant loss in accuracy, using labeled training data and algorithms like K-means clustering and SVM.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large number of light sources (e.g., 150 LEDs in 25 clusters) are used for spectral BRDF-based material classification, then measurement precision and classification accuracy are improved, but device complexity and data processing burden increase significantly
Solution Approach 1:
The patent extracts and identifies a small subset of critical light source clusters (e.g., 2-3 clusters) from the complete set of 25 clusters that provide the most discriminative information for material classification. This extraction approach maintains classification accuracy while dramatically reducing system complexity and data processing requirements.
Solution Approach 2:
The patent changes the parameter of light source quantity from 150 LEDs across 25 clusters to a reduced set of 2-3 clusters with 6 LEDs each. This parameter change is achieved through mathematical clustering analysis that identifies the optimal subset of light sources based on their contribution to material differentiation.
2Measurement precision
If 25 clusters of 6 LEDs each are used for illumination, then spectral BRDF measurement precision is improved, but the geometric and dimensional expansion of data measurements and analysis increases
Solution Approach 1:
The patent extracts only the essential light source clusters needed for accurate material classification, reducing the data volume from 150 LED measurements to measurements from just 2-3 clusters. This extraction is guided by mathematical clustering that identifies which clusters provide the most valuable discriminatory information.
3Device complexity
If a reduced number of light source clusters (2-3 clusters) is used, then device complexity and data processing requirements are reduced, but measurement precision may be compromised
Solution Approach 1:
The patent performs preliminary mathematical clustering analysis on training data to identify which light source clusters provide the most discriminative information before actual material classification. This preliminary action ensures that the selected 2-3 clusters are optimally chosen to maintain high classification accuracy despite the reduced number of light sources.
Solution Approach 2:
The patent uses training data to create a model or copy of the optimal light source configuration through mathematical clustering. This model identifies the essential clusters that replicate the classification performance of the full 25-cluster system, allowing accurate classification with fewer physical light sources.
4Reliability
If multiple light sources at different angles are used for BRDF measurement, then material classification reliability is improved, but the difficulty of detecting and measuring increases
Solution Approach 1:
The patent extracts the essential viewing and illumination geometry from the complete BRDF measurement space by identifying 2-3 critical light source clusters. This extraction maintains reliable material classification by preserving the most discriminative angular relationships while eliminating redundant measurements.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach simplifies material classification by reducing the number of light sources and data complexity, maintaining accuracy and enabling efficient classification of unknown materials with fewer images and processing requirements.
Implementation Method 1
an object fabricated from an unknown material is illuminated with light, and light reflected therefrom is measured
Data Source
AI summary
Material classification of an object is provided. Parameters for classification are accessed. The parameters include a selection to select a subset of angles for classification, a selection to select a subset of spectral bands for classification, a selection to capture texture features, and a selection to compute image-level features. The object is illuminated and a feature vector is computed based on the parameters. The material from which the object is fabricated is classified using the feature vector.