Edge Closing of Synthetic and Real Images using Polynomial Fitting

Usually, in image processing, edge detectors are mostly used as a basis for high level processing. In many cases, the edge detection gives erroneous and imprecise results, as incomplete or open contours. These inaccuracies mislead the subsequent processing. In this article, we propose a method to close contours of synthetic and real images. This method combines polynomial contour fitting with edge closing. A preliminary resampling of contour points is done to avoid the use of high degree polynomial. The edge closing is performed using an adaptive search window. The results in synthetic and real images are presented to validate the proposed technique and to illustrate its limitations.