Pattern generation for geometrical objects using morphological transformations

摘要

A new approach to generate a pattern for any object of binary image is proposed. One of important compression techniques of any monochrome images without losing the information is chain code encoding principle. It is segmenting the connected component of each region at point of boundary. The boundary is located by selecting the point with coordinates and then transmitted. This continues to reach the starting point, so this way object will be segmented from monochrome images with background. This novel method first applying the morphological transformations for enhancing the image, elimination of background, and binarisation, secondly using modified chain code encoding system for segmenting the object with edges and corner detection. This method generates the pattern for each object by applying modified chain code and morphological transformations procedure which extracts the coordinates of the shapes with considerable improvement in accuracy over classical chain code. We could reconstruct the original objects from the generated pattern which shows more accurate for considerable geometrical objects.