Ground Plane Estimation Using Adaptive Spatial-Temporal Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the ground plane in images, particularly in dynamic environments like vehicles, face challenges with aberrant results due to vertical objects and inadequate precision from limited textured areas, and become unreliable when the environment deviates significantly from nominal conditions.
Innovation Solution
A method that adapts topological and temporal constraints based on the quality of image data and previous estimates, selecting valid points, determining spatially and temporally filtered parameters, and using indicators to refine parameter calculations, ensuring robust ground plane estimation across varying environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If v-disparity calculation with Hough transform is used to estimate ground plane parameters, then the ground plane can be determined from depth image data, but the method yields aberrant results when vertical objects (such as walls) are present in the image
Solution Approach 1:
The patent applies local quality by analyzing the spatial distribution characteristics of points in the depth image. It identifies regions with different statistical properties (e.g., points near the ground versus points on vertical objects) and processes them differently. The method examines local density and depth variations to distinguish ground points from points on vertical structures, thereby improving reliability without sacrificing precision.
Solution Approach 2:
The patent changes parameters by dynamically adjusting filtering thresholds and statistical measures based on the observed data distribution. Instead of using fixed Hough transform parameters, the method adapts its parameters according to the local density, depth range, and spatial configuration of points, allowing it to distinguish ground planes from vertical objects more effectively.
2Quantity of substance
If pixel matching between stereoscopic images is used to obtain depth data, then depth information can be calculated for textured areas, but the precision becomes inadequate when textured areas are limited or absent
Solution Approach 1:
The patent merges multiple sources of information to compensate for insufficient textured areas. It combines depth data from pixel matching with spatial distribution analysis, statistical measures of point clouds, and temporal information from sequential images. This fusion approach maintains precision even when texture-based matching yields limited or noisy depth data.
Solution Approach 2:
The patent implements feedback by using the observed spatial distribution patterns and statistical properties of depth points to refine and validate the depth data. The method continuously adjusts its processing based on the quality and consistency of the depth information, improving precision by filtering out unreliable measurements and emphasizing consistent patterns across multiple observations.
3Device complexity
If fixed topological constraints are applied to ground plane estimation, then the estimation can be simplified, but the method becomes unreliable when the environment deviates significantly from nominal conditions
Solution Approach 1:
The patent applies dynamics by making the constraints adaptive rather than fixed. The method dynamically adjusts filtering criteria, threshold values, and processing parameters based on the observed environmental conditions, data quality, and statistical properties of the point cloud. This allows the system to maintain simplicity while adapting to diverse environments including non-nominal conditions such as steep slopes, uneven terrain, or limited visibility.
Data Source
AI summary
A method of determining a triplet of parameters defining a ground plane based on a depth image includes determining a plurality of triplets of parameters, each defining a ground plane, spatial filtering of the parameters, and temporal filtering of the parameters. The temporal filtering is dependent on an indicator of quality of depth image data available for determination of the ground plane. The temporal filtering is dependent on an indicator of quality of the ground plane determined after spatial filtering. Similarly, the spatial filtering can be parameterized as a function of the two indicators. Globally, the more the estimated plane for a depth image can “explain” the points in this image, the more accurately planes of the subsequent images are determined. For the spatial filtering, the ground plane is searched in a more limited space, and, for the temporal filtering, the previously estimated ground planes considered to a lesser degree.