Depth Estimation Using Two-Picture Matching Curves

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional methods for generating depth maps require additional hardware for single-lens cameras, such as range sensors or stereoscopic systems, which can be cumbersome and inefficient.

Innovation Solution

A method using a two-picture matching curve, generated by capturing images at different focus positions and applying Gaussian convolutions, to determine object distances without the need for extra hardware, employing a mathematical formulation with CCD or CMOS sensor data to create an iteration map that estimates relative object distances.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional depth map generation methods (range sensors or stereoscopic systems) are used, then depth information can be obtained, but additional hardware is required which increases device complexity

Engineering Contradiction:
Improvedepth measurement capabilityVSAvoidhardware requirements
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The single-lens camera system performs depth estimation using only its existing optical components and sensor, without requiring additional range sensors or stereoscopic camera pairs. The system leverages the camera's own focus mechanism and image processing capabilities to generate depth maps, making the system self-sufficient for depth measurement

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The method utilizes changes in focus position as a controllable parameter to encode depth information. By capturing images at different focus positions and analyzing the focus variation, the system translates optical parameter changes into depth measurements, eliminating the need for specialized depth-sensing hardware

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If multiple images at different focus positions are captured and processed, then accurate depth estimation is achieved, but processing time and computational complexity increase

Engineering Contradiction:
Improveobject distance estimation accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

A two-picture matching curve is pre-generated by capturing images at different focus positions and computing the matching relationships between them. This curve serves as a lookup table or reference model that stores the focus-position-to-depth mapping, allowing rapid depth estimation during actual operation without repeating the full multi-image processing sequence

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The depth estimation process is divided into two distinct phases: an offline calibration phase where the two-picture matching curve is generated, and an online estimation phase where the curve is used for rapid depth calculation. This segmentation allows computationally intensive operations to be performed only once, while subsequent depth measurements use the pre-computed curve for efficiency

Inventive Principle:
Principle #1Segmentation

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

Enables precise determination of object distances in a scene using a simple lens model, improving image processing efficiency by eliminating the need for additional hardware and providing accurate depth maps through iterative calculations and clustering algorithms.

Implementation Method 1

capturing images at different focus positions

Methodology Applied
Scientific EffectOptical focusing: Focusing

Implementation Method 2

The change in blur is then computed for the test target or object

Methodology Applied
Scientific EffectDepth of field: Depth of Field

Data Source

PatentEP2494524B1Algorithms for estimating precise and relative object distances in a scene
Publication Date: 2021.06.16 SONY ELECTRONICS INC
  • EP2494524B1 patent drawingFigure 1
  • EP2494524B1 patent drawingFigure 2A~2B
  • EP2494524B1 patent drawingFigure 3A~3B

AI summary

A two picture matching curve information is able to be used to determine precise object distance or relative object distance in a scene. Acquiring two images with different blur information in addition to the curve information enables a device to determine distance information of objects in a scene. The distance information is able to be used in image processing including generating a depth map which is then able to be used in many imaging applications.