Image Processing Apparatus for Human Object Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing techniques face difficulties in accurately tracking specific objects, such as faces and human bodies, due to challenges in feature detection and specification, leading to unsuitable tracking results and inadequate analysis of object movement.

Innovation Solution

An image processing apparatus and method that includes an image acquisition unit, object detection unit, object tracking unit, human body detection unit, object association unit, and trajectory management unit, which utilize background subtraction, pattern matching, and determination parameters to accurately associate and track objects and human bodies, managing movement information and attributes to enhance tracking accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional object detection techniques are used, then general object tracking can be achieved, but accurate tracking of specific objects with difficult-to-detect features cannot be achieved

Engineering Contradiction:
Improvetracking accuracyVSAvoidfeature detection difficulty
Core Design Contradiction:
Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The patent segments the detection process into multiple specialized units: object detection unit for general objects, human body detection unit for human bodies, and face detection unit for faces. Each unit focuses on detecting specific features, allowing the system to overcome the difficulty of detecting specific objects by dividing the complex detection task into manageable segments with dedicated detection strategies for each object type.

Inventive Principle:
Principle #1Segmentation

2Adaptability or versatility

If general object tracking is performed without specifying the target, then tracking processing can be applied to any object, but appropriate tracking results for specific objects cannot be obtained

Engineering Contradiction:
Improvetracking target specificationVSAvoidtracking result accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent applies local quality by assigning different detection and tracking attributes to different objects based on their specific characteristics. The object association unit associates detected objects with specific attributes (general object, human body, face) and manages them differently. This allows the system to be versatile in tracking any object while maintaining high accuracy for specific objects by applying object-specific detection and tracking strategies.

Inventive Principle:
Principle #3Local quality

3Measurement precision

If multiple detection units are introduced to improve specific object detection, then detection accuracy for specific objects improves, but system complexity increases

Engineering Contradiction:
Improvespecific object detection accuracyVSAvoidsystem structure complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent implements multi-functionality through the object association unit and trajectory management unit, which serve as central coordination components. These units manage multiple detection results, associate them with appropriate objects, and maintain trajectory information across different detection types. This universal management approach allows the system to handle multiple object types with different detection requirements without creating completely separate tracking systems for each object type.

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

Data Source

PatentEP3518146B1Image processing apparatus and image processing method
Publication Date: 2024.09.18 CANON KK
  • EP3518146B1 patent drawingFigure 1
  • EP3518146B1 patent drawingFigure 2A~2B
  • EP3518146B1 patent drawingFigure 3

AI summary

An image processing apparatus includes an object detection unit configured to detect an object in an image by using a first method which is a method using a subtraction process related to the image; a human detection unit configured to detect a human in the image by using a second method different from the first method, the second method being a method using a feature amount or a pattern corresponding to a human; a determination unit configured to determine whether an overlap ratio exceeds a predetermined threshold, the overlap ratio being based on an overlapping region where an object area and a human area overlap with each other, the object area corresponding to the object detected by the object detection unit, and the human area corresponding to the human detected by the human detection unit; a storingunit configured to store information indicating whether the object is a human; and an updating unit configured to update, if the overlap ratio of the object exceeds the predetermined threshold, the information for the object stored by the storing unit to information indicating that the object is a human.