Theo Pavlidis

An asynchronous thinning algorithm

Abstract A problem common to many areas of pictorial information processing is the transformation of a bilevel (two-color) image into a line drawing. The first step in such a process is discussed: transformation of the bilevel image into another bilevel image that is “thin.” An algorithm is proposed that can be implemented in either parallel or sequential fashion, and therefore is suitable for a mixed operation where a group of processors operate in parallel with each one examining sequentially the pixels of a part of the image assigned to it. It is possible to process images in pieces and thin correctly parts that are intersected by the dividing lines. Therefore, the method can be used on large images, such as maps and engineering drawings. The algorithm may also be run with certain options that label the thinned image so that exact reconstruction of the original is possible.

Contour filling in raster graphics

The paper discusses algorithms for filling contours in raster graphics. Its major feature is the use of the line adjacency graph for the contour in order to fill correctly nonconvex and multiply connected regions, while starting from a “seed.” Because the same graph is used for a “parity check” filling algorithm, the two types of algorithms can be combined into one. This combination is useful for either finding a seed through a parity check, or for resolving ambiguities in parity on the basis of connectivity.

Restoration of binary images using stochastic relaxation with annealing

This paper investigates the application of variations of Stochastic Relaxation with Annealing (SRA) as proposed by Geman and Geman [1] to the Bayesian restoration of binary images corrupted by white noise. After a general review we present some specific prior models and show examples of their application. It appears that a proper selection of the prior model is critical for the success of the method. We obtained better results on artificial images which fitted the model closely than on real images for which there was no precise model.

Scan Conversion of Regions Bounded by Parabolic Splines

A scan-conversion Algorithm has Been Implemented as Part of an on-line Program for Interactive Graphics and Document preparation.

Curve Fitting with Splines

In many applications where curve fitting is used, one would like to modify parts of the curve without affecting other parts. We shall say that a scheme has a local property if local modifications do not propagate. Clearly, the polynomials discussed in Section 10.2 do not have this feature and Bezier polynomials exhibit it only approximately. A change in the location or multiplicity of one of the guiding points requires the recalculation of the whole curve, even though the changes will have little effect far from the changed guiding point. Piecewise polynomial functions offer a direct way of achieving local control. We shall discuss such functions first in the y = y (x) form and later in parametric representations. The following is a general expression for a piecewise polynomial function:

Processing of Gray Scale Images

p(x) = {p_i}(x)\quad {x_i} \leqslant x \leqslant {x_{i + 1}}\quad i = 0,1...,k - 1

Creating Three-Dimensional Graphic Displays

(11.1a)