3D Structured Light Camera Depth Decoding via Adaptive Binarization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing 3D structured light cameras require high-power consumption and increased costs due to the need for an extra embedded ASIC for depth decoding, limiting their adoption in cost-sensitive applications.

Innovation Solution

A 3D structured light camera device with a light source and a method for performing depth value calculations that uses adaptive binarization, morphological operations, and binary encoding compression to optimize speckle matching, eliminating the need for an extra ASIC and reducing power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If an extra embedded ASIC is used for depth decoding, then depth measurement accuracy is improved, but power consumption and device cost increase

Engineering Contradiction:
Improvedepth measurement accuracyVSAvoidpower consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent combines the depth decoding algorithm with the existing image processing pipeline by integrating it into the main processor's task queue. The depth decoding function is merged with the existing image capture and processing operations, eliminating the need for a separate dedicated ASIC while maintaining depth measurement accuracy through coordinated processing of multiple image frames.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The main processor is designed to perform multiple functions including image capture, image processing, and depth decoding through a unified task queue system. This multi-functional approach allows the single processor to handle diverse computational tasks without requiring separate dedicated hardware for each function, thereby reducing overall device complexity and power consumption.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Measurement precision

If an extra embedded ASIC is used for depth decoding, then depth measurement accuracy is improved, but device cost increases

Engineering Contradiction:
Improvedepth measurement accuracyVSAvoiddevice cost
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent combines the depth decoding algorithm with the existing image processing pipeline by integrating it into the main processor's task queue. The depth decoding function is merged with the existing image capture and processing operations, eliminating the need for a separate dedicated ASIC while maintaining depth measurement accuracy through coordinated processing of multiple image frames.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The main processor performs depth decoding as part of its own computational capabilities rather than requiring external dedicated hardware. The processor serves its own depth measurement needs by executing depth decoding algorithms within its existing architecture, eliminating the need for additional ASIC components and reducing device cost.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If traditional depth decoding algorithms are used, then depth information is obtained, but computing time and processing resources increase

Engineering Contradiction:
Improvedepth information accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by capturing and storing multiple image frames at different exposure times before depth decoding is required. These pre-captured frames are stored in a task queue and ready for immediate processing, eliminating the need for real-time computation during depth measurement and significantly reducing processing time when depth information is needed.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The image capture operation continues uninterrupted while depth decoding processes stored frames in the background through the task queue system. The camera continuously captures images at different exposure times, and the depth decoding algorithm processes these pre-captured frames concurrently, maintaining continuous useful action without requiring sequential blocking of either operation.

Inventive Principle:
Principle #20Continuity of useful action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The proposed solution achieves highly efficient and cost-effective depth value calculations, reducing memory usage by up to 8 times and accelerating speckle matching by over 12 times, while maintaining pixel-level accuracy and supporting long-distance depth estimation.

Implementation Method 1

a light source configured to emit light, the light being modulated in accordance with a speckle pattern

Methodology Applied
Scientific EffectLight modulation:

Implementation Method 2

an IR camera configured to capture the modulated light reflected from the target object

Methodology Applied
Scientific EffectPhotoelectric effect: Photoelectric Effect

Data Source

PatentEP4542169A1Three-dimensional structured light camera device having a light source and method therefor
Publication Date: 2025.04.23 NXP BV
  • EP4542169A1 patent drawingFigure 1
  • EP4542169A1 patent drawingFigure 2
  • EP4542169A1 patent drawingFigure 3

AI summary

A three-dimensional structured light camera device (310) comprises a LUT memory (307) configured to store a reference speckle image of a 3D target object; and a processor (306) configured to: perform object detection on a target speckle image; perform sub-pixel-level, SPL, processing that comprises a plurality of interpolations of the reference speckle image and the target speckle image; perform an adaptive binarization process wherein the adaptive binarization process searches the LUT memory (307) for SPL speckle features of the stored reference speckle image that match SPL speckle features of the SPL target speckle image; find SPL speckle features of the target speckle image that match searched SPL speckle features of the stored reference speckle image; and perform a depth value calculation of the 3D target object in response to the plurality of interpolations and the found SPL speckle features of the target SPL speckle image.