This MATLAB library provides functionality for 1:1 conversion of line drawings into SVG files. The purpose is for lossless conversion of raster line drawings into vectorized line drawings. The converted files will be identical to the original files and infinitely zoomable. The SVG output is structured using a separate polyline between each joint.
Read in a raster line drawing, exLarge.png
im = imread('exLarge.png');
Run the function vectorizeLineSmart.m
on the image to get coordinate lists
svgData = vectorizeLineSmart(im);
Run the function printSVGpoly.m
to print the data to an SVG file
printSVGpoly(svgData, im, 'output.svg');
PNG (Original) | SVG (Converted) |
---|---|
![]() |
-
Why is this better than other alternatives?
-
After trying many solutions online, you will find that none are 1:1 conversions. Online solutions are designed around graphic design where a line may be larger than 1 pixel. They use machine learning to quickly convert raster to vector at the cost of some features.
-
This library is optimized for use on 1 pixel wide line drawings. These files might be the result of a skeletonization. Converting the file using this library will 100% preserve the image.
-
-
How fast are the functions?
-
On the example, the
exLarge.png
file which is 506x564 pixels and has 3403 edge pixels takes approximately 0.2 seconds to run thevectorizeLineSmart.m
function. The runtime will be roughly proportional to the dimensions*numPixels. The library is currently single-threaded. -
Alternatively, the
vectorizeLineDense.m
function can be used which will return the same image, but every pixel will have a connection to adjacent pixels. This will take approximately half the time, 0.1 seconds. Attempting to open the image in a renderer will take more time.
-
-
How is a 1:1 conversion achieved?
-
The
vectorizeLineSmart.m
function looks at any endpoints or joints in the line drawing. It traces a line from these points until it reaches another endpoint or joint. This process is repeated until all connections have been created. -
Using the
svgDataIntermediate
variable withprintSVGpoly.m
will create a image composed of polylines between the joints. Using thesvgDataSimple
variable withprintSVG.m
will create an image composed of straight line segments in 8-connectivity space. Using thesvgDataDense
variable withprintSVG.m
will create an image composed of pixel connections.
-
https://github.com/karsonkevin2/heatmap-skeletonization