Search

In this report three thinning algorithms are developed: one each for use with rectangular, hexagonal, and triangular arrays. The approach to the development of each algorithm is the same. Pictorial results produced by each of the algorithms are presented and the relative performances of the algorithms are compared. It is found that the algorithm operating with the triangular array is the most sensitive to image irregularities and noise, yet it will yield a thinned image with an overall reduced number of points. It is concluded that the algorithm operating in conjunction with the hexagonal array has features which strike a balance between those of the other two arrays. [ABSTRACT FROM AUTHOR]

A method for extracting a smooth polygonal contour from a digitized image is illustrated. The ordered sequence of contour points and the connection graph of the image are first obtained by a modified Ledley algorithm in one image scan. A minimal perimeter polygon subjected to specified constraints is then chosen as the approximating contour. The determination of the minimal polygon can be reduced to a nonlinear programming problem, solved by an algorithm which takes into account the weak bonds between variables. Some examples are presented, and the corresponding computing times are listed. [ABSTRACT FROM AUTHOR]