Vectorization (image tracing)
In the field of computer graphics processing, the term vectorization, sometimes (image) tracing, on the automated process of converting raster graphics into vector graphics related.
- 4.1 Natural object
- 4.2 lettering
- 4.3 Computer Graphics
In the vectorization simple geometric objects will attempt to identify in a scanning image. The common variant is looking over edge detection areas equal or similar brightness or color, also known as posterization known, which are then expressed as a Graphical Primitive or Bezier curves. Exports will be in various vector graphic file formats such as SVG, AI, CDR, EPS, TTF, PDF or various CAD formats.
Traditional edge detection methods that are to be used at the tracing eg Roberts operator, Canny algorithm and the Sobel operator.
Strengths and limitations of the methodology
In contrast to the reverse process, rasterization, an automatic (or semi-automatic ) vectorization is not clearly defined or a unique process, there are many different methods and degrees of freedom. Each of these different ways is with her Vektorisierungsansatz to another abstracting approximation of the raster image. Also, different methods and parameterizations are different for different good picture content, formats and resolutions suitable WorldCat is often a manual by trying different combinations necessary.
Furthermore, images of natural objects are difficult to vectorize representative in contrast to images of artificial objects, such as Computer graphics, as these typically have less color or brightness gradient and less irregular structures.
Vectorisations high resolution images with few colors and gradients ( plan drawings, writing, computer graphics ) can be significantly smaller than the original image with high representativeness well.
For images of natural objects to achieve a just acceptable traced result is a significant increase in the amount of data is often necessary. This is particularly true for the vectorization of images of natural objects in the JPEG format, a graphic format which is particularly suitable for soft transitions, and this is efficiently coded.
Since the fully automatic vectorization of images of objects with gradients only leads to unsatisfactory results, need to be finalized still often manually. This can range with a vector graphics program on the distance of vectorization artifacts, the Neueinbringung of shadows or gradients to the complete manual rebuild, see example below Tux.
Major fields of application
- When CAD drawings ( plan drawing, etc.) scanned, vectorized and exported as CAD files, this process is also paper- to- CAD conversion or drawing conversion called.
- In Geographic Information System (GIS ) satellite or aerial images are vectorized in order to create maps.
- In graphic design and photography sometimes be vectorized images to make them easier to use and scalable.
- When Font design drawn designs are transferred via vectorization in a font vector format.
- Handwritten notes and signatures are well- vectorizable, since they are typically only two colors ( white ink color) and have a great contrast. This is also the first step in a text recognition.
Influence of various Vektorisierungsprogrammen and parameters
Since photographs of natural objects often include many smooth transitions and small details, and the result also depends strongly on the grid resolution of the input image is an acceptable automatic vectorization difficult. Frequently has long manually to the parameters number of colors, noise reduction / quality are experimenting until an acceptable compromise reached from accomplished visual fidelity and file size.
Vectorized with 18 colors exported with Scan2CAD and in the SVG format, 340 KB
Vectorized with 12 colors in Vector Magic with the parameter High Quality, 369 KB
Vectorized with Auto Trace in Delineate GUI, 677 KB
Vectorized with Vector Magic using the High Quality and unlimited colors, 744 KB ( columns are artifacts occurred)
Vectorized with Rave Grid, 1.64 MB
Lettering of Christopher Columbus as a JPEG raster graphic ( 1,308 × 481 pixels), 63 kB
Vectorized with Sketch, edited manually in addition, 219 kB
- Autotracer.org online vectorization based on the Auto Trace Library (English)
- Vectormagic.com commercial online vectorization (English)
- Computer Graphics
- Algorithm (computer graphics )
- Image processing