Sketch Vectorization via Superpixel Active Boundary Traversal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital graphics generation systems fail to produce high-fidelity vector representations of sketches, often losing important content and requiring complex editing due to the lack of semantic relationship preservation between vector geometry and the original sketch strokes.
Innovation Solution
A vectorization system that generates a vector representation of an input image by converting it into a grayscale version, segmenting it into superpixel regions, classifying borders as active or inactive, and generating vector paths along active boundaries to create a compact and editable vector representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If conventional approaches convert paper sketches to digital vector graphics using Bezier curves, then the conversion process is simple, but the vector representation loses important content and fails to maintain high fidelity to the original sketch
Solution Approach 1:
The patent segments the sketch image into superpixel regions, where each superpixel is a contiguous group of pixels representing a uniform region. This segmentation allows the system to preserve semantic information by maintaining relationships between adjacent superpixels, thereby preventing loss of sketch content while converting to vector format.
Solution Approach 2:
The patent introduces superpixels as an intermediary representation between the original sketch image and the final vector graphic. These superpixels serve as a mediator that captures semantic relationships and spatial structure, enabling high-fidelity vectorization without losing important sketch content.
2Extent of automation
If conventional approaches generate vector representations, then the process is automated, but the resulting vectors require complex editing due to lack of semantic relationship preservation
Solution Approach 1:
By segmenting the image into superpixels and preserving their semantic relationships in the vector output, the patent enables automated vectorization that produces semantically accurate results. This reduces the need for complex manual editing, as the vector representation maintains the structural and semantic properties of the original sketch.
Solution Approach 2:
The patent incorporates feedback mechanisms where the vector generation process references the superpixel segmentation and active boundary detection to ensure semantic accuracy. This feedback loop ensures that the automated process produces vectors that faithfully represent the original sketch, reducing editing requirements.
3Measurement precision
If the vectorization system processes every pixel to maintain high fidelity, then the vector representation is accurate, but the processing time and computational complexity increase
Solution Approach 1:
The patent segments the image into superpixels, which reduces the number of individual processing units from pixels to superpixel regions. This segmentation maintains measurement precision by preserving semantic information while reducing computational complexity and processing time compared to processing every individual pixel.
Solution Approach 2:
The patent applies local quality by treating different regions of the image differently through superpixel segmentation. By identifying and processing only the relevant regions (foreground superpixels and active boundaries) rather than the entire image uniformly, the system maintains accuracy while reducing processing time.
Data Source
AI summary
Generating a vector representation of a hand-drawn sketch is described. To do so, the sketch is segmented into different superpixel regions. Superpixels are grown by distributing superpixel seeds throughout an image of the sketch and assigning unassigned pixels to a neighboring superpixel based on pixel value differences. The border between each pair of adjacent superpixels is then classified as either an active or an inactive boundary, with active boundaries indicating that the border corresponds to a salient sketch stroke. Vector paths are generated by traversing edges between pixel vertices along the active boundaries. To minimize vector paths included in the vector representation, vector paths are greedily generated first for longer curves along active boundaries until each edge is assigned to a vector path. Regions encompassed by vector paths corresponding to a foreground superpixel are filled to produce a high-fidelity vector representation of the sketch.