Deformable Contour Model for Parking Lot Boundary Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mapping systems cannot accurately depict the shape and boundary of parking lots using probe data, limiting navigation assistance for drivers.
Innovation Solution
A mapping system that analyzes probe data using a deformable contour model to determine the boundary of parking lots, associating probe data points with grid cells, calculating parking likelihood, and clustering to identify lot clusters, with an energy function-based approach to define and merge boundaries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If trajectory analysis or stay point analysis is used to identify parking lots, then parking locations can be detected, but the shape and boundary of parking lots cannot be defined
Solution Approach 1:
The patent segments the parking lot identification process into two distinct phases: first identifying parking locations using trajectory/stay point analysis, then determining the boundary using a deformable contour model. This segmentation allows each phase to specialize - location detection uses probe data patterns while boundary definition uses spatial expansion from identified clusters.
Solution Approach 2:
The patent introduces a deformable contour model as an intermediary between location detection and boundary definition. This model acts as a mediator that takes identified parking locations as input and generates the final boundary shape, bridging the gap between point-based detection and area-based representation.
2Manufacturing precision
If a deformable contour model is used to determine parking lot boundaries, then accurate shape and size representation is achieved, but system complexity increases
Solution Approach 1:
The patent performs preliminary clustering of probe data points into parking lot clusters before applying the deformable contour model. This preliminary organization of data simplifies the subsequent boundary determination process by providing structured input (clusters) rather than raw scattered points, reducing the complexity burden on the contour model.
Solution Approach 2:
The patent employs a deformable contour model that dynamically adjusts the boundary shape based on the distribution of parking lot clusters. The contour is not fixed but adapts its shape and size to match the actual parking lot boundaries, allowing accurate representation while using a relatively simple expansion algorithm from cluster centers.
3Ease of manufacture
If probe data points are clustered to identify parking lot clusters, then boundary determination is facilitated, but processing time increases
Solution Approach 1:
The patent segments the probe data processing into discrete grid cells before clustering. By dividing the spatial domain into cells and performing clustering within each cell independently, the overall processing time is reduced through parallelization, while still achieving accurate parking lot cluster identification for boundary determination.
Data Source
AI summary
A mapping system, method and computer program product are provided to identify a parking lot from probe data. In the context of a mapping system, processing circuitry is configured to determine a parking likelihood for each grid cell based upon probe data points associated with the respective grid cells and to identify likely parking locations in instances in which the parking likelihood for a respective grid cell satisfies a predefined threshold. The processing circuitry is also configured to cluster likely parking locations to identify parking lot clusters and to determine the boundary of the parking lot pursuant to a deformable contour model which causes a polygon to expand from a respective parking lot cluster so as to represent the boundary of the parking lot. The processing circuitry is further configured to update a map to include the parking lot having the boundary determined pursuant to the deformable contour model.