Monocular Depth Map Object Distance Determination

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing techniques for determining distance from a camera to an object are highly sensitive to camera calibration and placement, limiting their accuracy and requiring additional sensors like radar or LIDAR.

Innovation Solution

The method generates a depth map from monocular image data, detects objects, applies bounding boxes, masks the depth map by setting pixels to the closest pixel's depth value, and uses machine-learning programs to determine distances independently of camera calibration and placement, potentially fusing with data from other sensors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional distance determination techniques are used, then distance can be measured, but the measurement is highly sensitive to camera calibration and placement, reducing accuracy

Engineering Contradiction:
Improvedistance measurement accuracyVSAvoidsensitivity to camera calibration
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent introduces an intermediary object with known dimensions (such as a license plate or traffic sign) as a mediator between the camera and the distance measurement process. This intermediary provides reference features that are independent of camera calibration, allowing the system to determine distances accurately without being sensitive to camera placement or calibration parameters. The known dimensions of the intermediary object serve as a reference scale that decouples the measurement from camera-specific parameters.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent creates a digital representation or model of the intermediary object with precisely known dimensions and features. By comparing the captured image of the intermediary object against this digital copy or reference model, the system can extract geometric information and calculate distances without relying on camera calibration. The digital copy serves as a calibration-independent reference that enables accurate measurement.

Inventive Principle:
Principle #26Copying

2Measurement precision

If additional sensors like radar or LIDAR are used to improve distance measurement accuracy, then measurement precision increases, but device complexity and cost increase

Engineering Contradiction:
Improvedistance measurement accuracyVSAvoidnumber of sensors
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent makes the camera system multi-functional by enabling it to perform both image capture and accurate distance measurement through the use of intermediary objects with known dimensions. Instead of requiring separate dedicated sensors for distance measurement, the same camera that captures visual information also determines distances by analyzing the appearance and perspective of the intermediary object. This universal approach eliminates the need for additional specialized sensors.

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

Solution Approach 2:

The camera system serves itself by using the intermediary object in the scene to provide its own measurement reference. Rather than requiring external calibration equipment or additional sensors, the system leverages the known dimensions of objects naturally present in the environment (or deliberately placed) to perform self-calibration and distance determination. The intermediary object becomes a self-provided reference that enables the camera to measure distances independently.

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS12014508B2Distance determination from image data
Publication Date: 2024.06.18 FORD GLOBAL TECH LLC
  • US12014508B2 patent drawing
  • US12014508B2 patent drawing
  • US12014508B2 patent drawing

AI summary

A computer includes a processor and a memory storing instructions executable by the processor to receive image data from a camera, generate a depth map from the image data, detect an object in the image data, apply a bounding box circumscribing the object to the depth map, mask the depth map by setting depth values for pixels in the bounding box in the depth map to a depth value of a closest pixel in the bounding box, and determine a distance to the object based on the masked depth map. The closest pixel is closest to the camera of the pixels in the bounding box.