Image Binarization For 3D Printing

This paper presents a method to print two dimensional images with a 3D printer.It allows a user to create a solid version of an arbitrary chosen photograph using a 3D printer.The final objective of our method is to print an image to a flat,solid,two-tonal artistic style image.The introduced algorithm converts an image to data which are read-able by a fused deposition modeling(FDM)3D printer.This set of data is recognized by a specific software that generates the file containing printing instructions.A flat and solid image is then printed with no need for user's intervention.Our goal of printing solid two-tonal image is achieved by firstly turning the origi-nal image to a black and white version.The algorithm creating the model uses a non-photorealistic rendering(NPR)technique to convert the image to a two-tonal version,using black and white colors.It uses a line drawing algorithm to extract features of the image and combine the drawing with clean tonal blocks of black and white colors extracted from the original image.The output of this first step is a black and white image with large regions,with a high structural coherence and no noise or clutters.The second step of our method is to convert the two-tonal image to a model suitable for 3D printing.We design an adapted Delaunay triangulation to this purpose whose role is to represent black parts of the image.Our algorithm can be seen as a method to cleanly"cut out" black areas and convert them to volumes.Together,these volumes form the printing model which only contains black regions and has therefore many "holes" where white regions initially were in the black and white image.In our conception of the model,the black parts already contain all information needed for printing,including features,structure and shape information.These black parts are printed using a single color over a flat uniform surface of different color which serves as background.The background is then seen through the "holes" of the model and the total image,with its black and white regions,has in the end been printed.Our contribution is to introduce a heuristic to convert regions of interest of the black and white image into polygons which are used to create the model for printing.The binarization method was introduced in another paper,and even though its principle is detailed in this paper,its implementation is not part of our contributions.This tonal simplification is required to make the printing step automatic and make our method easy to implement.Experimentally,we use plastic materials to print,and pleasant results are obtained.Many kinds of photographs with various scenes can be printed in a moderate time.Various aspects as size,colors or thickness can be adjusted to obtain different results.Algorithms for binarization and polygonal conversion are fully-automatic and the printing process does not require any intervention.Moreover,no post-processing step is required after printing.