Depth Image Planar Surface Extraction via Dynamic Region Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for real-time planar surface extraction from depth images face challenges in detecting small planar structures and performing well in hostile environments with poor lighting or texture-less conditions, often resulting in high miss-detection and over-detection rates and inefficiencies due to computational costs.
Innovation Solution
A method involving dynamic region size adjustment using a depth change indication map, recursive planar region extraction, and clustering with an auto-balanced search tree to efficiently extract and segment planar surfaces, ensuring accurate plane fitting and reduced computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional planar surface extraction methods are used, then computational simplicity is maintained, but detection accuracy for small planar structures deteriorates and miss-detection rate increases
Solution Approach 1:
The patent segments the depth image into multiple scales by applying Gaussian pyramids, dividing the image into different resolution levels (e.g., 1/4, 1/8, 1/16 original resolution). This multi-scale segmentation enables detection of planar structures at various sizes, particularly improving small planar structure detection while distributing computational load across scales.
Solution Approach 2:
The patent transforms the 2D depth image into a 3D point cloud representation and processes it across multiple dimensional scales using Gaussian pyramids. This dimensional transformation and multi-scale processing adds a scale dimension to the detection space, enabling simultaneous detection of planar structures at different sizes without proportionally increasing computational complexity.
2Reliability
If traditional planar surface extraction methods are used, then processing speed is maintained, but detection reliability in hostile environments deteriorates
Solution Approach 1:
The patent performs preliminary processing by constructing Gaussian pyramids and pre-segmenting the depth image into multiple scales before actual planar detection. This preliminary multi-scale preparation ensures that when detection is performed, planar structures of various sizes are already represented in the appropriate scale, improving reliability in hostile environments without requiring excessive computational resources during the actual detection phase.
Solution Approach 2:
The patent changes the scale parameter by processing the depth image at multiple resolutions (different scale factors in Gaussian pyramids). This parameter variation allows the detection algorithm to adapt to different planar structure sizes and environmental conditions, improving reliability while maintaining processing efficiency through systematic scale progression.
3Measurement precision
If small planar structures are emphasized for detection, then detection completeness improves, but false detection rate increases
Solution Approach 1:
The patent dynamically adjusts the detection scale by implementing multi-scale Gaussian pyramids, where the detection process adapts to different planar structure sizes automatically. Small planar structures are detected at finer scales while larger structures are detected at coarser scales, creating a dynamic detection system that reduces false detections by matching the appropriate scale to each detected structure.
Solution Approach 2:
The patent performs preliminary multi-scale segmentation using Gaussian pyramids before detection, pre-organizing the image data at different scales. This preliminary action allows the detection algorithm to focus on appropriate scales for each planar structure, improving detection completeness for small structures while reducing false detections through scale-appropriate processing.
Data Source
AI summary
According to some embodiments, an imaging processing method for extracting a plurality of planar surfaces from a depth map includes computing a depth change indication map (DCI) from a depth map in accordance with a smoothness threshold. The imaging processing method further includes recursively extracting a plurality of planar region from the depth map, wherein the size of each planar region is dynamically adjusted according to the DCI. The imaging processing method further includes clustering the extracted planar regions into a plurality of groups in accordance with a distance function; and growing each group to generate pixel-wise segmentation results and inlier points statistics simultaneously.