Depth-Based Object Tracking for Similar-Object Occlusion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing object tracking techniques struggle with accurately identifying and tracking a target object when similar objects are present, leading to erroneous identifications and occlusion relationship estimations.

Innovation Solution

An information processing apparatus that acquires images in series and depth information, detects candidate areas for the tracking target, estimates occlusion states based on time-series distance information, and determines the target using an estimation unit to accurately track the object.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If template matching or brightness information is used for object tracking, then the tracking method is simple and fast, but similar objects with similar features cause erroneous identifications

Engineering Contradiction:
Improvetracking speedVSAvoididentification accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent introduces depth information as a new dimension to distinguish objects that appear similar in 2D images. By comparing depth values of candidate objects across multiple frames, the system can identify which object is actually the tracking target even when they have similar appearance features, thus resolving the contradiction between simple tracking and accurate identification.

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

Solution Approach 2:

The patent uses depth information as an intermediary factor to resolve ambiguities in object identification. Instead of directly comparing appearance features which lead to erroneous identifications, the system uses depth data as a mediator to disambiguate between similar-looking objects and determine the correct tracking target.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of information

If occlusion relationship estimation is performed using appearance features, then the estimation can be done with existing image data, but erroneous estimation occurs when similar objects are present

Engineering Contradiction:
Improveocclusion informationVSAvoidocclusion estimation accuracy
Core Design Contradiction:
Loss of informationVSReliability

Solution Approach 1:

The patent adds depth information as an additional dimension for estimating occlusion relationships. By considering depth values alongside appearance features, the system can accurately determine which object is in front of another, even when they have similar appearances, thereby improving occlusion estimation accuracy without losing information.

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

3Measurement precision

If multiple candidate objects with similar features are detected, then the detection sensitivity is high, but it becomes difficult to determine the correct tracking target

Engineering Contradiction:
Improvedetection sensitivityVSAvoidtarget identification difficulty
Core Design Contradiction:
Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The patent resolves the difficulty of identifying the correct target among multiple similar candidates by introducing depth information as a distinguishing feature. By comparing depth values across frames, the system can reliably identify the correct target even when multiple candidates have similar appearance features and high detection sensitivity.

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

Data Source

PatentUS20260004547A1Information processing apparatus, image capturing apparatus, information processing method, and storage medium
Publication Date: 2026.01.01 CANON KK
  • US20260004547A1 patent drawing
  • US20260004547A1 patent drawing
  • US20260004547A1 patent drawing

AI summary

An information processing apparatus includes an acquisition unit configured to acquire images captured in time series and distance information in a depth direction on a plurality of areas in each of the images, a detection unit configured to detect a candidate area of an object to be a tracking target from each of the images based on an image feature of each of the images, an estimation unit configured to estimate an occlusion state indicating whether the object to be the tracking target is occluded by another object different from the tracking target for the candidate area detected from each of the images based on time-series data of the distance information, and a determination unit configured to determine the candidate area of the object to be the tracking target from the candidate area detected from each of the images based on an estimation result of the occlusion state.