Digital Image Segmentation for Natural Video Panning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current automated image panning and zooming methods in digital image presentations, such as the Ken Burns effect, often produce unappealing and unnatural animations, failing to engage viewers effectively.

Innovation Solution

A method that segments digital images into relevant parts based on color, distance, and size, using techniques like histogram equalization, median filtering, and pixel intensity segmentation to automatically generate video sequences that mimic human eye movement, focusing on eye-catching areas and maximizing contrast.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If automated image panning and zooming methods are used to create video sequences from digital images, then the presentation becomes more dynamic and engaging, but the animation often appears unnatural and unappealing to viewers

Engineering Contradiction:
Improveautomated video sequence generationVSAvoidnatural appearance of animation
Core Design Contradiction:
Extent of automationVSReliability

Solution Approach 1:

The digital image is divided into multiple segments representing different objects or regions of interest. Each segment is analyzed independently to determine its characteristics (color, distance from center, size), and video sequences are generated by transitioning between these segments. This segmentation allows the system to make informed decisions about which regions to pan to or zoom in on, creating more natural and meaningful animations compared to random or crude automated methods.

Inventive Principle:
Principle #1Segmentation

2Reliability

If manual definition of starting and ending points for video sequences is used, then the animation can be precisely controlled to match human eye movement patterns, but the process requires significant user time and effort

Engineering Contradiction:
Improveaccuracy of eye movement simulationVSAvoiduser time for manual configuration
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system automatically analyzes the digital image to identify meaningful segments and determine optimal starting and ending points for video sequences. By computing scores based on color, distance from image center, and segment size, the system autonomously selects which regions to transition between, eliminating the need for manual user configuration while still producing natural-looking animations that simulate human eye movement patterns.

Inventive Principle:
Principle #25Self-service

3Ease of operation

If random panning and zooming is applied to digital images, then sometimes acceptable results are achieved that mimic human eye scanning, but more often the animation detracts from the image quality

Engineering Contradiction:
Improvesimplicity of automated applicationVSAvoidquality of animation result
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The system uses multiple parameters (color values, distance from center, segment size) to evaluate and score different image segments. By changing and comparing these parameters across multiple segments, the system identifies the most meaningful regions for video sequence transitions. This multi-parameter approach ensures that animations focus on visually significant areas rather than applying random panning and zooming, thereby maintaining high animation quality while keeping the process fully automated.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2614489B1Method and system for obtaining a control information related to a digital image
Publication Date: 2014.04.16 DEUTSCHE TELEKOM AG
  • EP2614489B1 patent drawingFigure 1~3
  • EP2614489B1 patent drawingFigure 4a~5

AI summary

The present invention relates to a method for obtaining a control information related to a digital image, the method comprising the steps of: - processing the digital image by - providing a segmentation of the digital image into a plurality of image segments comprising at least a first image segment and a second image segment, and by - determining the first image segment and the second image segment such that within the plurality of image segments, a product comprising, on the one hand, a colour value of the first image segment or of the second image segment and, on the other hand, the distance between the first image segment and the second image segment is maximized, and ~ using the location of the first image segment and the location of the second image segment to define the image control information. The present invention further relates to a system and a program comprising a computer readable program code.