Super Pixel Scene Segmentation for Focal Stack Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional image segmentation from focal stack images is inefficient, particularly in areas with depth discontinuity and weak texture, and lacks a balance between incorporating user input and minimizing user burden, with per-pixel focus measurement being unreliable and sophisticated techniques requiring significant computational resources.

Innovation Solution

The method groups pixels into super pixels, calculates focal measures and color probabilities for these super pixels, and segments images based on both focal measures and color probabilities, allowing for user input through selecting a focus frame to enhance segmentation efficiency and accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If per-pixel focus measurement is used for segmentation, then segmentation can be performed, but reliability is poor especially in areas with depth discontinuity and weak texture

Engineering Contradiction:
Improvesegmentation reliabilityVSAvoidfocus measurement precision
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The patent merges multiple focal stack images to compute a composite focus measure for each super pixel. By combining information from multiple images with different focus planes, the system overcomes the unreliability of per-pixel measurements in challenging areas, achieving more robust segmentation reliability.

Inventive Principle:
Principle #5Merging (Combining)

2Manufacturing precision

If sophisticated segmentation techniques like graph-cut algorithm are used, then segmentation accuracy may improve, but computation time increases significantly due to large graph size

Engineering Contradiction:
Improvesegmentation accuracyVSAvoidcomputation time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent segments the image into super pixels first, then performs segmentation calculations on these larger units rather than individual pixels. This hierarchical segmentation approach maintains segmentation accuracy while dramatically reducing the number of nodes in the graph, thus reducing computation time.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from pixel-level operations to super pixel-level operations, effectively changing the dimension of operation. By grouping pixels into super pixels, the system reduces the data dimension while preserving essential segmentation information, enabling efficient processing.

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

3Reliability

If user input is incorporated into segmentation calculations, then segmentation quality may improve, but user burden increases when significant detail is required

Engineering Contradiction:
Improvesegmentation qualityVSAvoiduser burden
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The patent implements semi-automatic segmentation where users provide minimal input (such as selecting a focus frame or providing rough annotations) rather than detailed complete segmentation. This partial action approach achieves sufficient segmentation quality without imposing excessive user burden.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The system performs automatic super pixel formation, focal measure calculation, and segmentation computation without requiring extensive user input. The algorithm self-services by automatically processing the images and generating segmentation results, with user input serving only as optional guidance.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP2863362B1Method and apparatus for scene segmentation from focal stack images
Publication Date: 2019.04.24 NOKIA TECHNOLOGIES OY
  • EP2863362B1 patent drawingFigure 1
  • EP2863362B1 patent drawingFigure 2
  • EP2863362B1 patent drawingFigure 3

AI summary

A method, apparatus and computer program product are provided to facilitate scene segmentation from focal stack images. The method may include receiving a set of focal stack images, calculating a focal measure for each of a plurality ofpixels of the set of focal stack images, and grouping each of a plurality of pixels for which the focal measure was calculated into a plurality of super pixels. The method may also include calculating a focal measure for each of the plurality of super pixels, segmenting a respective focal stack image based on the focal measure of each of the plurality of super pixels, calculating a color probability for each of the plurality of super pixels, and segmenting each focal stack image based on color probability of each of the plurality of super pixels.