Moving Image Cut-Out Tracking With Regression-Based Cutting Locus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques for generating cut-out images in moving images struggle with determining appropriate cutting-out regions, leading to unwanted shifts due to object movement or detection errors, resulting in a disjointed viewing experience.
Innovation Solution
An image processing apparatus that detects objects in a moving image, determines a region of interest, derives a cutting locus based on object movement, and generates cut-out images using this locus to maintain consistent cutting-out regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If the cutting-out region position is determined solely based on object position in each image, then the cutting-out region can automatically follow object movement, but the cutting-out region may shift to unintended regions due to object behavior changes or detection errors
Solution Approach 1:
The patent introduces a regression line as an intermediary element between the detected object positions and the cutting-out region position. Instead of directly using object positions, the system calculates a regression line from multiple detected positions and determines the cutting-out region based on this regression line. This intermediary approach filters out temporary detection errors and object behavior variations, providing more stable and reliable cutting-out region positioning while maintaining automatic operation.
2Speed
If the cutting-out region follows object movement directly, then the cut-out images can track the object, but the viewing experience becomes disjointed when objects move erratically or are temporarily undetected
Solution Approach 1:
The patent performs preliminary actions by calculating the regression line from multiple detected object positions before determining the cutting-out region. This preliminary calculation of the regression line creates a predictive model that smooths out position variations. When objects move or are temporarily undetected, the regression line provides a stable reference that maintains cutting-out region continuity, preventing disjointed viewing experience while still tracking object movement trends.
3Stability of the object's composition
If linear regression analysis is used to smooth cutting-out region movement, then the cutting locus becomes smoother, but regions not originally intended may still be selected depending on object behavior
Solution Approach 1:
The patent changes the parameters used for determining cutting-out region position by incorporating not only the regression line from object positions but also reference position information. By adjusting and combining multiple parameters (regression line position, reference positions, and their relationships), the system achieves both smooth cutting locus and accurate cutting-out region selection, preventing unintended region selection while maintaining smooth movement.
Data Source
AI summary
An image processing apparatus acquires a moving image to be processed, detects an object from each of images included in the moving image, determines a position of a region of interest in each of the images included in the moving image, based on a result of the detection, derives a cutting locus for the moving image based on a locus corresponding to movement of the position of the region of interest and a reference position for a cutting-out region, the cutting locus being a locus corresponding to movement of a position of the cutting-out region, and generates a cut-out image from the cutting-out region identified based on the cutting locus in each of the images included in the moving image.