Depth Map Generation for Wearable Object Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

High-resolution and high-frame-rate image processing in wearable terminals leads to increased processing amounts, making it difficult to reduce image processing load while maintaining detection accuracy.

Innovation Solution

An image processing device that uses depth acquisition circuitry to generate depth maps from image data and object detection circuitry to identify objects by utilizing distance information, allowing for efficient object detection in a predetermined depth range.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If high-resolution and high-frame-rate image processing is performed, then detection accuracy is improved, but image processing amount increases

Engineering Contradiction:
Improvedetection accuracyVSAvoidimage processing amount
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent divides the image processing task into two segments: depth map generation using parallax information from multiple cameras, and object detection using the generated depth map. This segmentation allows the system to process depth information separately from detailed image analysis, reducing the overall processing burden while maintaining detection accuracy through specialized depth-based algorithms.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a depth map as an intermediary data structure between raw image capture and object detection. The depth map contains distance information for each pixel and serves as a mediator that simplifies subsequent object detection operations by providing pre-computed depth cues, thereby reducing the computational complexity of the detection process.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If resolution or frame rate is lowered, then image processing amount is reduced, but detection accuracy deteriorates

Engineering Contradiction:
Improveimage processing amountVSAvoiddetection accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent transitions from two-dimensional image data to three-dimensional depth information by generating depth maps that include distance values for each pixel. This dimensional transformation enables the system to perform object detection in 3D space using parallax information, allowing accurate detection with reduced processing requirements by exploiting the additional depth dimension rather than relying solely on high-resolution 2D image analysis.

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

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

This approach reduces the image processing load while maintaining detection accuracy, enabling efficient object detection and reducing power consumption in wearable terminals.

Implementation Method 1

depth acquisition circuitry that uses a parallax corresponding to image data to electronically generate a depth map of an image

Methodology Applied
Scientific EffectParallax: Parallax

Data Source

PatentUS11089289B2Image processing device, imaging device, image processing method and program
Publication Date: 2021.08.10 SONY SEMICON SOLUTIONS CORP
  • US11089289B2 patent drawing
  • US11089289B2 patent drawing
  • US11089289B2 patent drawing

AI summary

An image processing device includes depth acquisition circuitry that uses a parallax corresponding to image data to electronically generate a depth map of an image and object detection circuitry that uses distance information and the depth map to electronically detect a specific object in the image by identifying specific pixels in the image data. The depth map includes information that pertains to distances from a reference position for each pixel in the image.