Convert a color image into vector graphics:

Convert a natural image to vector graphics using 8 colors:

Convert a bitmap of a digit to vector graphics:

Convert a grayscale image to vector graphics:

By default, an automatic color quantization is performed:

Specify the number of colors to use:

Specify the explicit colors to use in the vector graphics:

Create an outline image:

Simplify color regions using mean-shift and smooth outlines via curvature flow filter:

Convert to vector graphics:

Successively simplify the color regions of an image with a mean-shift filter:

Generate vector graphic maps. Import a pixel-based map of the Rocky Mountains near Denver:

Delete county borders:

Convert colored elevation map into graphic contours:

Perform vector graphic operations introducing contour lines and changing colors:

Transform the map without altering line thickness:

Compare the result with the corresponding pixel-based transformation:

Creating a scalable icon of a kingfisher:

Obtain a rough mask:

Refine the mask with the grow-cut algorithm:

Regularize the color segments via curvature flow:

Cover the background with green:

Select the dominant colors of the bird and add green as the background color:

Generate a vector graphic:

Remove the green background segment:

Insert the expression into a larger graphic to obtain a scalable icon:

Typically, the scalable graphics version of the image is much smaller in byte count:

ImageGraphics returns color segments similar to color quantization in the LAB color space:

Segment and select shapes:

Colorize the segmented image:

Convert the image into vector graphics. Double-click to select and move a segment

"MarchingSquares" and "DualMarchingSquares" introduce gaps at T-junctions:

Segment edges in the resulting graphic are not colored:

Colorize the edges by providing explicit FaceForm and EdgeForm specifications:

Lines and curves of one-pixel thickness may not be detected correctly:

Make the lines thicker for a better conversion:

If the minimum color distance is larger than the image color variation, the output may be of a single color:

Lower the color distance:

Segments touching across pixel corners are connected:

Use Method"MarchingSquares" to keep segments separated:

A low-resolution image may cause some pixelation artifacts:

Some of that artifact may be removed after image upsampling:

Alternatively, use a high-resolution image: