3D Vehicle Environment Modeling Using Entropy-Based Region Priority
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for creating three-dimensional virtual models of a motor vehicle environment are inefficient in poor visibility conditions, such as those caused by precipitation or fog, due to reduced image contrast and texture, which affects the accuracy of depth estimation and model generation.
Innovation Solution
Select regions with higher region entropy, i.e., those with more texture and contrast, as starting points for generating three-dimensional coordinates, using algorithms or triangulation calculations, and prioritize these regions for depth estimation, followed by calculating coordinates for less entropic regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If regions with low contrast and texture (low entropy regions) are used for three-dimensional coordinate calculation, then the processing can be simpler, but the accuracy of depth estimation and model generation deteriorates in poor visibility conditions
Solution Approach 1:
The patent applies local quality by differentiating the processing approach based on region characteristics. High entropy regions (with good contrast and texture) are identified and processed with higher priority for three-dimensional coordinate calculation, while low entropy regions are processed differently or with lower priority. This localized differentiation ensures that regions most suitable for accurate depth estimation are handled optimally, resolving the contradiction between processing simplicity and measurement precision.
2Ease of operation
If all regions are processed equally for three-dimensional coordinate calculation, then the process is uniform and simple to implement, but the accuracy deteriorates when some regions have poor visibility conditions
Solution Approach 1:
The patent changes the parameter of region selection by introducing region entropy as a criterion. Instead of processing all regions equally, the system calculates entropy values for different regions and uses these values to prioritize which regions to process first for three-dimensional coordinate calculation. This parameter change allows the system to adapt to varying visibility conditions across different regions, improving reliability while maintaining reasonable operational simplicity through automated entropy-based selection.
3Productivity
If three-dimensional coordinates are calculated for all regions simultaneously, then the model generation is faster in ideal conditions, but the precision of depth estimation decreases in poor visibility conditions
Solution Approach 1:
The patent applies preliminary action by first identifying and processing high entropy regions before tackling low entropy regions. The system performs a preliminary assessment of region entropy and prioritizes coordinate calculation for regions with better visibility characteristics. This staged approach ensures that the most reliable depth information is established first, improving precision while still maintaining overall productivity by efficiently processing regions in order of their suitability for accurate measurement.
Data Source
AI summary
A computer-implemented method is provided for creating a three-dimensional virtual model of an environment (2) of a motor vehicle (1). The method includes capturing images of the environment (2); determining a plurality of regions (3; 4; 5; 6; 7; 8; 9) in each of the images and calculating a region entropy for each of the regions (3; 4; 5; 6; 7; 8; 9). The method then proceeds by selecting at least one region (4; 9) that has a higher region entropy than other regions (3; 5; 6; 7; 8); and generating the three-dimensional virtual model by first ascertaining three-dimensional coordinates for the selected at least one region (4; 9) and only thereafter ascertaining three-dimensional coordinates for the unselected regions (3; 5; 6; 7; 8). The three-dimensional virtual model includes the three-dimensional coordinates of all regions (3; 4; 5; 6; 7; 8; 9).