Image Foreground Matting Using Neighborhood and Non-Neighborhood Smoothness Priors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image foreground matting methods are insufficient in handling complicated images, particularly due to challenges in setting proper neighborhood windows for Laplacian matrix calculation, which can lead to insufficient detail capture or destruction of color line models, and are inadequate when only relying on neighborhood or non-neighborhood smoothness priors.
Innovation Solution
A novel image foreground matting method that combines neighborhood and non-neighborhood smoothness priors, involving steps to mark image areas, initialize opacity probabilities, calculate data and smoothness constraint weights, and solve energy equations to distinguish foreground and background pixels, using both Laplacian and local linear embedded dimension reduction approaches to construct graph patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large neighborhood window is used for Laplacian matrix calculation, then more detail information is captured, but the color line model is destroyed leading to bad results
Solution Approach 1:
The patent divides the neighborhood window into two distinct regions: a center region where the color line model is enforced to maintain reliability, and a surrounding ring region where gradient constraints are applied to capture detail information. This spatial segmentation allows both requirements to be satisfied simultaneously without conflict.
2Reliability
If a small neighborhood window is used for Laplacian matrix calculation, then the color line model is preserved, but detail information is insufficient
Solution Approach 1:
The patent extends the constraint application from a single-dimensional small window to a two-dimensional structure by adding a surrounding ring region. This allows the system to maintain the color line model in the center while capturing detail information from the extended ring area, effectively adding a spatial dimension to the solution.
3Stability of the object's composition
If only neighborhood smoothness prior is used, then local consistency is maintained, but complicated image problems cannot be solved
Solution Approach 1:
The patent merges two distinct prior constraints: neighborhood smoothness prior (applied in the center region) and non-neighborhood smoothness prior (applied in the surrounding ring region). This combination allows the system to handle complicated image problems by leveraging both local consistency and broader contextual information.
4Stability of the object's composition
If only non-neighborhood smoothness prior is used, then global consistency is improved, but neighborhood structure information is not captured
Solution Approach 1:
The patent applies different quality constraints to different spatial locations: non-neighborhood smoothness prior is applied in the surrounding ring region to improve global consistency, while neighborhood smoothness prior is applied in the center region to preserve local structure information. This localized differentiation resolves the contradiction between global and local requirements.
Data Source
AI summary
The present invention discloses an image foreground matting method based on neighborhood and non-neighborhood smoothness priors. The method primarily comprises the steps of: interactively marking foreground points and background points; initializing α values of each unmarked pixel of the input image by a color sampling method, calculating confidence degree of the pixel, and admitting α values of pixels of which confidence degree is larger than a given threshold as known pixels; calculating data term weights, neighborhood smoothness constraint term weights and non-neighborhood smoothness constraint term weights of each pixel in the input image to construct graph patterns of all pixels of the input image; and according to α values of the known pixels, under the constraint of the graph patterns, solving probabilities that each pixel belongs to the foreground by minimizing the energy equation so as to obtain alpha mattes.