Lidar Image Reconstruction for Precise Depth and Object Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing LIDAR systems face challenges in providing robust distance accuracy down to a few cm at an economical cost, especially in varying conditions, and struggle to provide comprehensive environmental data, including identification of specific objects like traffic signals and moving objects, while requiring extensive computational resources for real-time 3D point cloud analysis.

Innovation Solution

The system employs kernel-based image processing techniques on LIDAR data, utilizing dedicated circuitry and AI coprocessors to correlate LIDAR and color pixels, enabling robust distance measurement and object identification, with image reconstruction and classification techniques integrated into a single system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If 2D lidar images are used for autonomous vehicle navigation, then device complexity is reduced, but measurement precision and detection accuracy deteriorate due to lack of depth information

Engineering Contradiction:
Improvelidar system complexityVSAvoiddepth measurement precision
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent applies dimensionality change by transforming 2D lidar images into 3D point cloud representations through multi-view geometry and epipolar constraint calculations. This allows the system to maintain simple 2D sensor hardware while recovering full 3D spatial information, directly resolving the contradiction between device simplicity and measurement precision.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If 3D lidar data is processed to improve navigation accuracy, then measurement precision improves, but processing time and computational complexity increase

Engineering Contradiction:
Improvespatial detection precisionVSAvoiddata processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary action by pre-calculating and storing epipolar constraint matrices and projection relationships during system initialization. This preprocessing enables real-time 3D reconstruction during operation, reducing computational burden and processing time while maintaining high measurement precision.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If multiple lidar views are integrated to improve accuracy, then measurement precision improves, but device complexity and data processing requirements increase

Engineering Contradiction:
Improvespatial measurement accuracyVSAvoidmulti-view integration complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces epipolar constraints as an intermediary mathematical framework that simplifies multi-view integration. By using fundamental matrices and epipolar geometry as mediators, the system can efficiently fuse multiple lidar views without requiring complex direct coordination between all sensor pairs, thus reducing overall system complexity while improving measurement accuracy.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4025934B1Processing of lidar images
Publication Date: 2026.05.06 OUSTER INC
  • EP4025934B1 patent drawingFigure 1A~1B
  • EP4025934B1 patent drawingFigure 2
  • EP4025934B1 patent drawingFigure 3

AI summary

Systems and methods are provided for processing lidar data. The lidar data can be obtained in a particular manner that allows reconstruction of rectilinear images for which image processing can be applied from image to image. For instance, kernel-based image processing techniques can be used. Such processing techniques can use neighboring lidar and/or associated color pixels to adjust various values associated with the lidar signals. Such image processing of lidar and color pixels can be performed by dedicated circuitry, which may be on a same integrated circuit. Further, lidar pixels can be correlated to each other. For instance, classification techniques can identify lidar and/or associated color pixels as corresponding to the same object. The classification can be performed by an artificial intelligence (AI) coprocessor. Image processing techniques and classification techniques can be combined into a single system.