Color Decontamination for Image Compositing via Border Region Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image compositing techniques using alpha mattes often result in undesirable background color contamination in the border regions of foreground objects when composited onto new backgrounds, leading to computationally infeasible global optimization processes, especially for high-resolution images.
Innovation Solution
A method for color decontamination that estimates foreground colors in border regions based on original image colors, neighboring pixel colors, and alpha matte values, using weighted averages and luminance blending to create a composite image with reduced background influence, allowing for efficient processing and user-controlled blending.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If global optimization is used to approximate foreground and background portions, then color accuracy is improved, but computational complexity becomes infeasible for high-resolution images
Solution Approach 1:
The patent segments the image processing into distinct regions: foreground pixels (α=1), background pixels (α=0), and border region pixels (0<α<1). This segmentation allows different processing strategies to be applied to each region, avoiding the need for global optimization across all pixels while maintaining color accuracy in the border regions where contamination occurs.
Solution Approach 2:
The patent applies local quality by using different color estimation approaches for different regions. For border region pixels, it estimates foreground colors based on neighboring foreground pixels and alpha values. For pure foreground pixels, it uses the original image colors directly. This localized approach reduces computational complexity while maintaining color accuracy where needed.
2Manufacturing precision
If alpha matte compositing is used to extract foreground objects, then object separation is achieved, but background color contamination appears in border regions
Solution Approach 1:
The patent extracts the contaminating background color component from border region pixels by estimating what the pure foreground color should be based on neighboring foreground pixels and alpha values. This extraction removes the harmful background influence while preserving the desired foreground appearance in the composite image.
Solution Approach 2:
The patent converts the harmful background color contamination into useful information by using the alpha values and neighboring pixel colors to estimate the true foreground color. The contamination effect is transformed into a basis for calculating the correct foreground color through weighted averaging and color decontamination formulas.
3Productivity
If weighted average color estimation is used for border pixels, then processing efficiency is improved, but color accuracy may be compromised
Solution Approach 1:
The patent uses parameter changes by adjusting the weighting factors in the color estimation based on alpha values. Pixels with alpha closer to 1 receive higher weight from neighboring foreground pixels, while pixels with alpha closer to 0 receive higher weight from the original image colors. This dynamic parameter adjustment maintains color accuracy while enabling efficient processing through closed-form solutions.
Data Source
AI summary
Various embodiments of methods and apparatus for removing unwanted background color from a border region surrounding a foreground object in an image in order to composite the foreground object of the image with a new background image are described. Embodiments of color decontamination for image compositing may accept an image and an alpha matte corresponding to the image as input. In some embodiments, estimated foreground colors are determined for pixels in a border region between the foreground and the background of the input image. In some embodiments, the input image may be created by down-sampling a higher resolution image and pixels with estimated foreground colors may be up-sampled. In some embodiments, a composite image may be created based on the input image, the alpha matte, the estimated foreground colors of pixels in the border region and a new background image. In some embodiments, a texture preserving luminance blending operation may be performed using the estimated foreground colors for pixels in the border region to create refined foreground colors for pixels in the border region. In some embodiments, a composite image may be created further based on the refined foreground colors for pixels in the border region. In some embodiments, parameters controlling the creation of a composite image may be set in response to user input.