3D Point Cloud Filtering Using Depth-Frequency Segmentation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Three-dimensional point cloud data has a large data amount, and existing methods of reducing data amount through fixed grids risk eliminating essential data, leading to increased processing load and reduced control accuracy.

Innovation Solution

A signal processing device generates frequency data indicating point density in the depth direction and applies data reduction processing to efficiently extract relevant point cloud data, reducing unnecessary data based on this frequency data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If three-dimensional point cloud data is used for control, then recognition accuracy and object recognition capability are improved, but data amount and processing load increase

Engineering Contradiction:

Improverecognition accuracyVSAvoiddata amount

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent divides the three-dimensional space into multiple depth regions and segments the point cloud data accordingly. By processing and filtering data in segmented depth regions rather than as a whole, the system maintains recognition accuracy while reducing the overall data amount that needs to be processed for control decisions.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different processing strategies to different depth regions based on local characteristics. Regions with high object density receive different treatment compared to sparse regions, optimizing the balance between maintaining necessary recognition accuracy and reducing processing load in each local area.

Inventive Principle:

Principle #3Local quality

2Productivity

If fixed grid filtering is applied to reduce data amount, then processing load is reduced, but essential point cloud data may be eliminated

Engineering Contradiction:

Improveprocessing efficiencyVSAvoidcontrol accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent replaces fixed grid filtering with dynamic depth region-based filtering. The filtering boundaries and parameters are adapted based on the actual distribution of points in depth space, allowing the system to dynamically preserve essential data while removing redundant information, thus maintaining control accuracy while improving processing efficiency.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes the filtering parameters based on depth region characteristics rather than using fixed parameters. By adjusting filtering thresholds and region boundaries according to the actual point cloud distribution in different depth zones, the system avoids eliminating essential data while achieving effective data reduction for control processing.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20250252663A1Signal processing device, signal processing method, and data production method

Publication Date: 2025.08.07 SONY SEMICON SOLUTIONS CORP

AI summary

A signal processing device according to the present technology includes a frequency data generation section that generates frequency data that is data indicating a frequency of a point in at least a depth direction, on the basis of three-dimensional point cloud data acquired through three-dimensional measurement of a target space, and a data reduction section that executes data reduction processing for the three-dimensional point cloud data on the basis of the frequency data generated by the frequency data generation section.