Automated Region of Interest Identification in Digital Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Manual identification of regions of interest in digital images for industrial applications is time-consuming and prone to precision issues, leading to extended processing times and potential inaccuracies in quality control and defectiveness assessment.

Innovation Solution

An automated method for identifying regions of interest within digital images using a calibration process involving a checkerboard or grid, determining the real coordinates of illuminator end points through mirror surface alignment and triangulation, and calculating the shortest light beam path to define the region of interest precisely.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual identification of region of interest is used, then the operator can select pixels representing the region of interest, but the processing time is excessively extended

Engineering Contradiction:
Improveregion of interest identification precisionVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces the manual mechanical operation of selecting pixels with an automated computational algorithm. The system automatically identifies the region of interest by detecting edges, calculating gradients, and determining the bounding box that encompasses the object, eliminating the need for manual pixel selection while maintaining identification precision.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs self-identification of the region of interest through automated image processing algorithms. The algorithm independently analyzes the digital image, detects object boundaries, and determines the region of interest without requiring external manual intervention, thereby resolving the time loss associated with manual operations.

Inventive Principle:
Principle #25Self-service

2Loss of time

If manual definition of region of interest is performed once at the start of production, then setup time is reduced, but processing precision declines due to position and dimension variations in different images

Engineering Contradiction:
Improvesetup timeVSAvoidregion of interest identification precision
Core Design Contradiction:
Loss of timeVSMeasurement precision

Solution Approach 1:

The patent implements a dynamic region of interest identification system that adapts to each individual image. Rather than using a fixed predetermined region, the algorithm dynamically determines the region of interest for each image by detecting actual object boundaries, thereby maintaining precision across images with varying object positions and dimensions while requiring minimal setup time.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system performs preliminary automated analysis of each image to identify the region of interest before the actual quality control processing. This preliminary action of automatic region detection ensures that the subsequent processing is always focused on the correct region, maintaining precision without requiring lengthy manual setup for each image configuration.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If automated region identification is implemented, then processing time is reduced, but the complexity of the system increases

Engineering Contradiction:
Improveimage processing speedVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the image processing task into distinct computational stages: edge detection, gradient calculation, contour identification, and bounding box determination. By dividing the automated region identification process into these manageable segments, the system achieves high processing speed while keeping each individual computational step relatively simple and well-defined.

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

This method significantly reduces the time required to set up and analyze digital images, enhances precision, and allows for objective and automatic identification of regions of interest, improving the efficiency and accuracy of industrial quality control processes.

Implementation Method 1

acquiring a digital image in which a light source is in superposition with the mirror surface; determining a normal vector of the mirror surface in a predefined coordinate system

Methodology Applied
Scientific EffectLight reflection: Reflection

Implementation Method 2

determining real coordinates of end points of the illuminator... by means of a triangulation

Methodology Applied
Scientific EffectTriangulation:

Data Source

PatentEP3477590B1Method for identifying a region of interest within a digital image
Publication Date: 2020.03.04 CMZ SISTEMI ELETTRONICI SRL
  • EP3477590B1 patent drawingFigure 1
  • EP3477590B1 patent drawingFigure 2a~2b
  • EP3477590B1 patent drawingFigure 3~4

AI summary

A method for identifying a region of interest (1) within a digital image (2) relating to a surface (3) of a real object (4), the acquisition of the digital image (2) being carried out by means of an artificial vision system (100) comprising a television camera (101) and an illuminator (102) provided with a two-dimensional surface comprising at least three vertex points (V1, V2, V3). The method comprises executing a calibration step to determine the geometric relation existing between the points (i) of the surface (3) of the real object (4) in the real space (S) and the pixels of the digital image (2) of the surface (3); defining the position in real space (S) of the television camera (101) and of at least three vertex points (V1, V2, V3) of the two-dimensional surface of the illuminator (102) and defining as the real reflection point (ir) of the light coming from each vertex point (V1, V2, V3) of the illuminator (102) on the surface (3) of the real object (4). The method comprises lastly converting, by means of the geometric relation, the real reflection point (ir) identified for each vertex point (V1, V2, V3) into a virtual reflection point (pr) in the digital image (2) and of defining the region of interest (1) between the virtual reflection points (pr).