In variable data printing (VDP), it is desirable to automatically fit an
arbitrary shaped image object into an arbitrarily shaped copy hole in a
template with maximized use of the available space. In this paper, we
describe a practical image processing method that segments out the
object and determines the scale and translation factors for the optimal
fitting. For our application, an image object
... [Show full abstract] is placed on a uniformed
color background in a JPEG compressed image file. The compression
artifacts around the object boundary area complicate object
segmentation. In order to identify object boundary precisely, we
developed an orientation-dependent adaptive region growing method, which
significantly improve the boundary accuracy. In the first step, we
identify the background pixels using zero thresholding. Connectivity
analysis is then performed to remove very small blobs. Mathematical
morphological operations are applied to background pixels in order to
smooth the border. In the second step, object boundaries are refined
using the proposed orientation-dependent adaptive region growing. In
determining the optimal scale and translation, we use an image-based
exhaustive search algorithm that steps through a set of scaling factors
in descending order until a complete fit is found. The exit scaling
factor along with the associated mass center of the feasible
translations are then used for the object placement.
