RGB-D Camera BRDF Parameter Retrieval via Neural Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for measuring material BRDFs are inefficient, requiring extensive setups, expensive equipment, and high time and hardware expenditure, especially for large databases of assets, and lack multispectral properties.
Innovation Solution
A method using an RGB-D camera to capture images, reconstruct 3D models, classify BRDFs with a deep neural network, and retrieve multi spectral BRDF parameters through iterative optimization, enabling efficient and accurate BRDF parameter retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If state-of-the-art processes are used to measure material BRDFs, then measurement accuracy is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent uses photogrammetry to create digital 3D copies of physical objects, extracting geometric and material properties from photographs taken from multiple viewpoints. This digital copying approach replaces complex physical measurement setups with computational methods, achieving accurate BRDF measurement without expensive specialized equipment
Solution Approach 2:
The patent replaces traditional mechanical/optical measurement systems with a computational approach based on photogrammetry and image processing. Instead of using specialized BRDF measurement devices, the system uses standard cameras and computational algorithms to extract material properties, substituting mechanical measurement with information processing
2Reliability
If extensive measurement iterations are performed for large databases of assets, then completeness of material data is improved, but time expenditure and hardware resources increase
Solution Approach 1:
The patent performs preliminary processing of photographs to extract 3D geometric information and material properties before actual BRDF measurement. By pre-processing the visual data and creating digital models in advance, the system reduces the time needed for subsequent measurement iterations when building large asset databases
Solution Approach 2:
The system uses the object's own visual appearance in photographs to automatically extract its material properties and BRDF characteristics. The method is self-contained, requiring no external measurement equipment or manual intervention, allowing rapid processing of large numbers of assets without additional hardware resources
3Measurement precision
If finely divided samples are stored for large databases, then retrieval accuracy is improved, but memory requirements increase
Solution Approach 1:
The patent extracts only the essential material properties and BRDF parameters from physical objects, storing these extracted characteristics rather than the complete original datasets. This extraction approach maintains retrieval accuracy by preserving key material identifiers while significantly reducing storage requirements compared to storing finely divided samples
Data Source
Figure 1~2
AI summary
In summary the invention relates to a computer implemented method and apparatus for retrieval of multi spectral bidirectional reflectance distribution function, BRDF, parameters by using red-green-blue-depth, RGBD, data. The method comprises the steps of: Capturing (S1) by an RGB-D camera at least one image of one or more objects in a scene, wherein the captured at least one image of the one or more objects comprises RGB-D data including color and geometry information of said objects; Reconstructing (S2) by a processing unit the captured at least one image of the one or more objects to one or more 3D reconstructions by using the RGB-D data; Classifying (S3) by a deep neural network the BRDF of a surface of the one or more objects based on the 3D reconstructions, wherein the deep neural network comprises an input layer, an output layer, and at least one hidden layer between the input layer and the output layer; and Retrieving (S4) the multi spectral BRDF parameters by approximating the classified BRDF by using an iterative optimization method. Due to the present invention, the multi spectral BRDF parameters of objects in specified environments can be retrieved.