Map Difference Extraction via Branch Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting differences between two maps often fail to accurately correlate nodes and edges across meshes, leading to erroneous difference extraction due to errors in map division and correlation.
Innovation Solution
An information processing device that divides map regions into branch and non-branch areas based on overlap degrees, calculates graph editing paths to minimize editing cost, and outputs these paths as differences between reference and input maps, using a control unit to acquire and process map data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If maps are divided into meshes for difference extraction, then the extraction process can be simplified and computational load reduced, but errors occur where the same node or edge is included in different meshes leading to erroneous correlation
Solution Approach 1:
The patent applies segmentation by dividing the map into multiple regions and further dividing each region into grid cells. This hierarchical segmentation allows efficient processing while maintaining accuracy through proper correlation of nodes and edges across boundaries. The segmentation is performed with consideration for road geometry to prevent erroneous correlations.
Solution Approach 2:
The patent introduces an intermediary correlation determination process that uses overlap areas between adjacent grid cells as mediation zones. Nodes and edges in these overlap areas are subjected to special correlation determination using multiple criteria (distance, angle, road type) to ensure accurate matching across mesh boundaries, preventing erroneous correlations while maintaining processing efficiency.
2Device complexity
If traditional mesh-based difference extraction is used, then processing is simpler, but accuracy deteriorates due to erroneous correlation of nodes and edges across meshes
Solution Approach 1:
The patent segments the map into regions and grid cells with hierarchical structure, enabling both simplified processing within cells and accurate correlation across boundaries through the overlap area mechanism.
Solution Approach 2:
The patent applies different processing qualities to different locations: simple grid-based processing within non-overlap areas, and enhanced correlation determination with multiple criteria in overlap areas. This local quality differentiation maintains overall simplicity while ensuring accuracy at critical boundary regions.
3Measurement precision
If the entire map is processed as a single graph, then correlation accuracy is maintained, but computational load increases significantly
Solution Approach 1:
The patent divides the large-scale graph processing into smaller sub-graphs corresponding to grid cells and regions. This segmentation reduces the computational complexity from O(n²) for the entire map to O(k²) for each small cell, where k is the cell size, enabling efficient processing while maintaining accuracy through the overlap area correlation mechanism.
Data Source
AI summary
An information processing device for extracting a difference between a reference map and an input map, the information processing device comprising: a control unit that divides each region of a reference map and an input map into a branch region where a branch of a road or a lane exists, and a non-branch region other than the branch region; determine a branch region and a non-branch region of the input map corresponding to the branch region and the non-branch region of the reference map based on an overlap degree of the branch region of the reference map and the branch region of the input map, and determine a graph editing path in which an editing cost determined by a predetermined criterion is minimized for each of the corresponding regions; and output the graph editing path for multiple corresponding regions as a difference between the reference map and the input map.