Height-Map Image Segmentation for Overlapping Cell Clusters

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods struggle to accurately divide analysis target areas in all-in-focus images, particularly when cell clusters overlap, making it difficult to control the shape of divided areas and extract meaningful regions.

Innovation Solution

An image processing method that utilizes a height map to divide analysis target areas into multiple regions based on height distribution, identifies inadequate areas, and integrates them with neighboring regions using a determination score, generating an all-in-focus image and height map from multiple picked-up images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If the Watershed method is used to divide overlapping cell clusters, then division of analysis target area is attempted, but it becomes difficult when Seed regions cannot be obtained and the shape of divided areas cannot be controlled

Engineering Contradiction:
Improveease of area divisionVSAvoidcontrol of divided area shape
Core Design Contradiction:
Ease of operationVSManufacturing precision

Solution Approach 1:

The patent changes the parameter basis from traditional Watershed Seed region-based division to height map-based division. By using height information from the height map as the dividing parameter, the method can naturally separate overlapping cell clusters without requiring Seed regions, while the determined shape of divided areas is controlled by the height distribution characteristics.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If multiple picked-up images are combined to generate all-in-focus image, then whole target is in focus, but overlapping cell clusters appear merged and cannot be individually extracted

Engineering Contradiction:
Improvefocus accuracyVSAvoidextraction precision of individual cell clusters
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent segments the analysis target area into multiple divided areas based on height map information. This segmentation allows individual cell clusters that appear merged in the all-in-focus image to be separated into distinct regions, enabling precise extraction and analysis of each cluster while maintaining the overall focus accuracy of the all-in-focus image.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4704028A1Image processing method, image processing apparatus, image pickup apparatus, and computer-readable program
Publication Date: 2026.03.04 SCREEN HOLDINGS CO LTD
  • EP4704028A1 patent drawingFigure 1~2
  • EP4704028A1 patent drawingFigure 3
  • EP4704028A1 patent drawingFigure 4

AI summary

An image processing apparatus includes a memory memorizing a target image representing a target viewed from a predetermined direction and a height map indicating heights, in the predetermined direction, of portions of the target corresponding to respective pixel positions of the target image, a target area specifying part for specifying an analysis target area in the target image, an area division part for dividing the analysis target area into divided areas in accordance with the height map, an inadequate area specifying part for specifying an inadequate area among the divided areas, which does not satisfy a predetermined determination condition, and an area integrating part for acquiring a new divided area by calculating a determination score in a case where the inadequate area is integrated with each of divided areas contacting the inadequate area and then integrating the inadequate area with one divided area based on the determination score.