Image Blending via Object-Aware Alpha Masking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Alpha blending techniques fail to effectively utilize content information between images, leading to reduced visual appeal, visibility issues, and ghosting effects due to inadequate handling of overlap regions and alpha mask generation.
Innovation Solution
An image processing device and method that identifies objects in overlap regions, generates a Gaussian curve, and determines an alpha mask for blending images based on this curve to ensure objects of interest are not obscured, using a combination of image blending and alpha mask generation techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If alpha blending is used to combine multiple source images, then the blending process can be performed, but objects may fade into the background and visual appeal is reduced
Solution Approach 1:
The patent applies preliminary action by performing object detection and segmentation before the blending operation. The system identifies objects in the foreground image, determines their boundaries, and creates masks that protect these objects during the subsequent alpha blending process, preventing them from fading into the background.
Solution Approach 2:
The patent applies local quality by creating spatially varying alpha values based on object locations. Different regions of the image receive different blending treatments - areas containing objects are protected with higher alpha values while other regions can be blended more aggressively, allowing localized control over the blending effect.
2Device complexity
If a fixed overlap window size is used in blending, then the device complexity is reduced, but seam and ghosting effects occur
Solution Approach 1:
The patent applies dynamics by making the overlap window size adaptive rather than fixed. The system dynamically adjusts the window size based on detected object characteristics such as size, shape, and position. This allows the blending parameters to change according to the content being processed, eliminating seams and ghosting effects while maintaining manageable device complexity through automated adjustment.
3Productivity
If alpha blending is performed without considering content information, then the processing speed is maintained, but visual appeal and object distinction are reduced
Solution Approach 1:
The patent applies segmentation by dividing the image into distinct regions based on object boundaries detected through image processing. The system segments the foreground image to identify objects, creates separate masks for different objects, and applies blending operations to each segment independently. This content-aware approach improves visual appeal while maintaining processing efficiency through localized operations.
Data Source
AI summary
An apparatus and method for composing images in an image processing device are provided. The method includes identifying an object in an overlap region of a plurality of images; generating a Gaussian curve for the object; determining an alpha mask for the overlap region using the Gaussian curve; and blending the plurality of images based on the alpha mask.