Evaluating the accuracy of line thinning algorithms after processing scanned line data

The development and rapid growth of computer mapping has led to many discussions concerning the accuracy of techniques used to generate these computer representations. The purpose of this study is to analyze the accuracy of thinning methods applied to scanned map data, which is only one in a series of processes used to accomplish digital conversion of conventional maps.

In preliminary tests, nine thinning methods based on the successive layer removal process are evaluated. Seven raster images are thinned using these methods. The raster results are compared based on the number of pixels deleted and on the number of retained pixels that fall either on or off the medial axis of the original matrix. The four algorithms that produce the best results are then used for final testing.

For the final tests, 25 digital lines are plotted and scanned. The raster images are thinned using the four successive layer removal methods and a line following method developed for this study. The raster output is evaluated using the preliminary testing method. The final vector output is compared to the original input based on line length, anchor line length, and fractal dimension.