Sketch Vectorization With Aligned Guide Images for Stroke Fidelity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems fail to generate high-fidelity vector representations of hand-drawn sketches due to alignment issues and irregularities, such as stroke overshoots, undershoots, and misalignments, leading to inaccuracies in converting sketches to digital vector graphics.
Innovation Solution
A stroke-guided vectorization system generates an aligned guide image by using an adaptive exposure filter based on vector strokes from the original guide image to define pixel-specific thresholds, distinguishing between sketch strokes and background, and training a deep learning network with these improved training pairs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional approaches are used to convert paper sketches to digital vector graphics, then the conversion process is simple and fast, but the fidelity and accuracy of the vector representation are poor with important content lost
Solution Approach 1:
The system performs preliminary actions by generating an aligned guide image that merges the input sketch with a reference guide image before the actual vectorization process. This pre-alignment step corrects registration errors, stroke overshoots, and undershoots in advance, providing a cleaner input for the subsequent vectorization pipeline and improving overall fidelity without complicating the core vectorization algorithm
Solution Approach 2:
The aligned guide image serves as an intermediary element between the noisy input sketch and the final vector output. By introducing this intermediate representation that combines information from both the sketch and guide image, the system mediates the transformation process, preserving important content while reducing artifacts, thereby improving vector representation fidelity
2Measurement precision
If the system uses an adaptive exposure filter with per-pixel thresholds to distinguish strokes from background, then the accuracy of stroke detection is improved, but the computational complexity increases
Solution Approach 1:
The system applies local quality by using per-pixel thresholds instead of global thresholds for stroke detection. Each pixel's threshold is adaptively determined based on local characteristics of the aligned guide image, allowing the system to accurately distinguish strokes from background in different regions with varying contrast and lighting conditions, thereby improving stroke detection accuracy
Solution Approach 2:
The system dynamically changes parameters by computing adaptive exposure filters and per-pixel thresholds based on the local intensity distribution in the aligned guide image. This parameter adaptation allows the stroke detection to respond to local variations in the image, improving accuracy while the computational overhead is managed through efficient histogram-based threshold computation
Data Source
AI summary
A stroke-guided vectorization system is described that generates, from an input sketch and guide image depicting an approximate vector representation of the sketch, an aligned guide image depicting an improved vector representation of the sketch. To do so, the stroke-guided vectorization system determines black levels representing a vector stroke in the input sketch and white levels representing a background in the input sketch. The stroke-guided vectorization system determines a black threshold value and a white threshold value for discrete portions of the aligned guide image using subsets of the black levels and subsets of the white levels determined using the input sketch. Each discrete portion of the aligned guide image is then mapped to a vector stroke or a background based on the black threshold value and the white threshold value of the portion.