Feature Point Scale Modification for Accurate Environment Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing distance measurement devices and environment map generation systems struggle with accurate scale modification of feature points, leading to incomplete and inaccurate environment maps, especially for distant objects.
Innovation Solution
A distance measurement device and environment map generation apparatus that includes a feature point detecting unit, a first distance calculating unit, a second distance calculating unit, and a scale modifying unit, which detect feature points, calculate distances, and modify the scale of feature points using a ratio of first and second distances to improve accuracy and generate high-accuracy environment maps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only feature points simultaneously calculated by monocular stereo processing and multiocular stereo processing are modified, then processing complexity is reduced, but the entire environment map is not modified and scale accuracy is insufficient
Solution Approach 1:
The patent segments the environment map into multiple regions based on depth information, processing different regions with different methods. Close objects use multiocular stereo processing while distant objects use monocular stereo processing with scale modification, allowing selective modification of feature points rather than processing the entire map uniformly
Solution Approach 2:
The patent performs preliminary depth estimation and region classification before executing the full processing pipeline. By pre-dividing the environment map into close and distant regions, the system prepares the groundwork for selective feature point modification, ensuring that only relevant regions are processed in detail while maintaining overall map consistency
2Speed
If monocular stereo processing is used for distant objects, then processing speed is improved, but distance measurement accuracy deteriorates
Solution Approach 1:
The patent applies different processing qualities to different spatial regions. For distant objects where monocular processing is used, the system enhances local quality by selectively modifying feature point distances based on depth information and environmental context, thereby improving accuracy without sacrificing the speed advantage of monocular processing
Solution Approach 2:
The patent dynamically adjusts processing parameters based on object distance and region type. By changing the degree of feature point modification and the weight given to depth information according to the spatial location, the system optimizes the balance between processing speed and measurement accuracy for different distances
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Based on an image imaged by an imaging unit (2), feature points of a plurality of objects present around a vehicle are detected by a feature point detecting unit (35). A first distance from the feature point to the imaging unit (2) is calculated based on a temporal change of the feature point. A second distance from a certain object included in the plurality of objects present in the image to the imaging unit is calculated using some pixels of the certain object. The first distance of the plurality of feature points, other than the certain feature point, simultaneously detected by the feature point detecting unit (35) is modified based on a ratio of the first distance and the second distance.